Home

Awesome

Imaging analysis tools

A table containing imaging analysis tools for biology and neuroscience, with a focus on calcium imaging.

Created by <a href='https://bahanonu.com' target='_blank'>Biafra Ahanonu, PhD</a> (<a href='https://www.hhmi.org/scientists/biafra-ahanonu' target='_blank'>HHMI Hanna Gray Fellow</a>, <a href='https://basbaumlab.ucsf.edu/people/' target='_blank'>Basbaum Lab</a>, UCSF).

DOI

<p align="center"> <a href="https://user-images.githubusercontent.com/5241605/94530890-9c3db280-01f0-11eb-99f0-e977f5edb304.gif"> <img src="https://user-images.githubusercontent.com/5241605/94530890-9c3db280-01f0-11eb-99f0-e977f5edb304.gif" align="center" title="ciapkgMovie" alt="ciapkgMovie" width="60%" style="margin-left:auto;margin-right:auto;display:block;margin-bottom: 1%;"> </a> <p align="center">Calcium imaging analysis with CIAtah (https://github.com/bahanonu/ciatah).</strong> </p> </p>

The table can also be found at:

Notes:

<strong> <figcaption class="caption" ><span class="id">Table 1:</span><span class="content">Ca<sup class="textsuperscript"><span class="rm-lmr-10x-x-109">2+</span></sup> imaging cell extraction and trace reconstruction algorithms</span></figcaption></strong><!--tex4ht:label?: x1-1012r --> <div class="tabular"> <table id="TBL-1" class="tabular"><colgroup id="TBL-1-1g"><col id="TBL-1-1" /></colgroup><colgroup id="TBL-1-2g"><col id="TBL-1-2" /></colgroup><colgroup id="TBL-1-3g"><col id="TBL-1-3" /></colgroup><colgroup id="TBL-1-4g"><col id="TBL-1-4" /></colgroup><colgroup id="TBL-1-5g"><col id="TBL-1-5" /></colgroup><colgroup id="TBL-1-6g"><col id="TBL-1-6" /></colgroup><tr class="hline"><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td></tr><tr style="vertical-align:baseline;" id="TBL-1-1-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-1-1" class="td11"># </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-1-2" class="td11">Method </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-1-3" class="td11">Year</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-1-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-1-5" class="td11">Notes/Code </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-1-6" class="td11">Citation </td> </tr><tr class="hline"><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td></tr><tr style="vertical-align:baseline;" id="TBL-1-2-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-2-1" class="td11">1 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-2-2" class="td11">PhaseCorrelation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-2-3" class="td11">1996</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-2-4" class="td11">Motion correction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-2-5" class="td11">•  Phase correlation for motion correction, to include translation, rotation, and scale-invariance. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-2-6" class="td11"><a href="#Xreddy1996fft">Reddy and Chatterji</a> <a href="#Xreddy1996fft">1996</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-3-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-3-1" class="td11">2 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-3-2" class="td11">Turboreg </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-3-3" class="td11">1998</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-3-4" class="td11">Motion correction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-3-5" class="td11">•  Motion correction.<br class="newline" />•  <a href='http://bigwww.epfl.ch/thevenaz/turboreg/' target='_blank'>http://bigwww.epfl.ch/thevenaz/turboreg/</a><a id="dx1-1014"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-3-6" class="td11"><a href="#Xthevenaz1998pyramid">Thevenaz et al.</a> <a href="#Xthevenaz1998pyramid">1998</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-4-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-4-1" class="td11">3 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-4-2" class="td11">subPixelPhase </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-4-3" class="td11">2002</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-4-4" class="td11">Motion correction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-4-5" class="td11">•  Closed-form solution to subpixel translation estimation using phase correlation. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-4-6" class="td11"><a href="#Xforoosh2002extension">Foroosh et al.</a> <a href="#Xforoosh2002extension">2002</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-5-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-5-1" class="td11">4 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-5-2" class="td11">ROI </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-5-3" class="td11">2005</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-5-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-5-5" class="td11">•  Matrix multiplication; in some methods neuropil/background subtraction implemented. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-5-6" class="td11"><a href="#XKerr2005">Kerr et al.</a> <a href="#XKerr2005">2005</a>; <a href="#XKuchibhotla2014">Kuchibhotla et al.</a> <a href="#XKuchibhotla2014">2014</a>; <a href="#XPeron2015">Peron et al.</a> <a href="#XPeron2015">2015</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-6-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-6-1" class="td11">5 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-6-2" class="td11">CellProfiler </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-6-3" class="td11">2006</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-6-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-6-5" class="td11">•  Multi-algorithm pipeline for cell segmentation.<br class="newline" />•  <a href='https://cellprofiler.org' target='_blank'>https://cellprofiler.org</a><a id="dx1-1015"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-6-6" class="td11"><a href="#Xcarpenter2006cellprofiler">Carpenter et al.</a> <a href="#Xcarpenter2006cellprofiler">2006</a>; <a href="#Xmcquin2018cellprofiler">McQuin et al.</a> <a href="#Xmcquin2018cellprofiler">2018</a>; <a href="#Xlamprecht2007cellprofiler">Lamprecht et al.</a> <a href="#Xlamprecht2007cellprofiler">2007</a></td> </tr><tr style="vertical-align:baseline;" id="TBL-1-7-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-7-1" class="td11">6 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-7-2" class="td11">PCA-ICA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-7-3" class="td11">2009</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-7-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-7-5" class="td11">•  Cell extraction using principal component analysis (PCA) followed by independent component analysis (ICA). </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-7-6" class="td11"><a href="#Xmukamel2009automated">Mukamel et al.</a> <a href="#Xmukamel2009automated">2009</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-8-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-8-1" class="td11">7 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-8-2" class="td11">ANTs </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-8-3" class="td11">2009</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-8-4" class="td11">Image analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-8-5" class="td11">•  Suite of tools for registering and analyzing imaging data.<br class="newline" />•  <a href='http://stnava.github.io/ANTs/' target='_blank'>http://stnava.github.io/ANTs/</a><a id="dx1-1016"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-8-6" class="td11"><a href="#Xavants2009advanced">Avants et al.</a> <a href="#Xavants2009advanced">2009</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-9-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-9-1" class="td11">8 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-9-2" class="td11">elastix </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-9-3" class="td11">2009</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-9-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-9-5" class="td11">•  A general toolbox for rigid and non-rigid image registration.<br class="newline" />•  <a href='https://elastix.lumc.nl' target='_blank'>https://elastix.lumc.nl</a><a id="dx1-1017"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-9-6" class="td11"><a href="#Xklein2009elastix">Klein et al.</a> <a href="#Xklein2009elastix">2009</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-10-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-10-1" class="td11">9 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-10-2" class="td11">Lucas–Kanade framework</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-10-3" class="td11">2009</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-10-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-10-5" class="td11">•  Lucas-Kanade framework for non-uniform motion image registration. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-10-6" class="td11"><a href="#Xgreenberg2009automated">Greenberg and Kerr</a> <a href="#Xgreenberg2009automated">2009</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-11-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-11-1" class="td11">10 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-11-2" class="td11">CIRF (calcium-behavior) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-11-3" class="td11">2011</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-11-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-11-5" class="td11">•  Regressive model to obtain Ca<sup class="textsuperscript"><span class="rm-lmr-10x-x-109">2+</span></sup> signal based on behavior. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-11-6" class="td11"><a href="#XMiri2011">Miri et al.</a> <a href="#XMiri2011">2011</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-12-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-12-1" class="td11">11 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-12-2" class="td11">openBIS </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-12-3" class="td11">2011</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-12-4" class="td11">Data handling </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-12-5" class="td11">•  FAIR data management.<br class="newline" />•  <a href='https://openbis.ch' target='_blank'>https://openbis.ch</a><a id="dx1-1018"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-12-6" class="td11"><a href="#Xbauch2011openbis">Bauch et al.</a> <a href="#Xbauch2011openbis">2011</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-13-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-13-1" class="td11">12 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-13-2" class="td11">Automated ROI analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-13-3" class="td11">2012</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-13-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-13-5" class="td11">•  Automatic ellipses based ROI detection. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-13-6" class="td11"><a href="#XFrancis2012">Francis et al.</a> <a href="#XFrancis2012">2012</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-14-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-14-1" class="td11">13 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-14-2" class="td11">OMERO </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-14-3" class="td11">2012</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-14-4" class="td11">Data handling </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-14-5" class="td11">•  Microscopy data handling.<br class="newline" />•  <a href='https://www.openmicroscopy.org' target='_blank'>https://www.openmicroscopy.org</a><a id="dx1-1019"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-14-6" class="td11"><a href="#Xallan2012omero">Allan et al.</a> <a href="#Xallan2012omero">2012</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-15-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-15-1" class="td11">14 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-15-2" class="td11">ADINA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-15-3" class="td11">2013</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-15-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-15-5" class="td11">•  Sparse dictionary learning. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-15-6" class="td11"><a href="#Xdiego2013automated">Diego et al.</a> <a href="#Xdiego2013automated">2013</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-16-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-16-1" class="td11">15 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-16-2" class="td11">TPP </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-16-3" class="td11">2013</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-16-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-16-5" class="td11">•  Tool for processing two-photon calcium imaging data, e.g. finding cells with SeNeCA.<br class="newline" />•  <a href='http://uemweb.biomed.cas.cz/tpp/' target='_blank'>http://uemweb.biomed.cas.cz/tpp/</a><a id="dx1-1020"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-16-6" class="td11"><a href="#Xtomek2013two">Tomek et al.</a> <a href="#Xtomek2013two">2013</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-17-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-17-1" class="td11">16 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-17-2" class="td11">NMF </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-17-3" class="td11">2014</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-17-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-17-5" class="td11">•  Cell extraction using nonnegative matrix factorization (NMF). Followed by CNMF. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-17-6" class="td11"><a href="#Xpnevmatikakis2014structured">Pnevmatikakis et al.</a> <a href="#Xpnevmatikakis2014structured">2014</a>; <a href="#XMaruyama2014">Maruyama et al.</a> <a href="#XMaruyama2014">2014</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-18-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-18-1" class="td11">17 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-18-2" class="td11">SIMA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-18-3" class="td11">2014</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-18-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-18-5" class="td11">•  Normalized cut segmentation, motion correction, etc.<br class="newline" />•  <a href='https://github.com/losonczylab/sima' target='_blank'>https://github.com/losonczylab/sima</a><a id="dx1-1021"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-18-6" class="td11"><a href="#XKaifosh2014">Kaifosh et al.</a> <a href="#XKaifosh2014">2014</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-19-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-19-1" class="td11">18 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-19-2" class="td11">DataJoint </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-19-3" class="td11">2015</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-19-4" class="td11">Data handling </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-19-5" class="td11">•  Schema for data handling.<br class="newline" />•  <a href='https://github.com/datajoint/datajoint-matlab' target='_blank'>https://github.com/datajoint/datajoint-matlab</a><a id="dx1-1022"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-19-6" class="td11"><a href="#Xyatsenko2015datajoint">Yatsenko et al.</a> <a href="#Xyatsenko2015datajoint">2015</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-20-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-20-1" class="td11">19 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-20-2" class="td11">NWB </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-20-3" class="td11">2015</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-20-4" class="td11">Data handling </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-20-5" class="td11">•  Neurodata Without Borders (NWB) initiative to produce a common data format for electrophysiology and imaging studies.<br class="newline" />•  <a href='https://github.com/NeurodataWithoutBorders' target='_blank'>https://github.com/NeurodataWithoutBorders</a><a id="dx1-1023"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-20-6" class="td11"><a href="#XTeeters2015">Teeters et al.</a> <a href="#XTeeters2015">2015</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-21-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-21-1" class="td11">20 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-21-2" class="td11">Suite2p </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-21-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-21-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-21-5" class="td11">•  Generative model along with GUIs. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-21-6" class="td11"><a href="#Xpachitariu2016suite2p">Pachitariu et al.</a> <a href="#Xpachitariu2016suite2p">2016</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-22-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-22-1" class="td11">21 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-22-2" class="td11">CNMF (CaImAn) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-22-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-22-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-22-5" class="td11">•  Constrained NMF (CNMF).<br class="newline" />•  <a href='https://github.com/flatironinstitute/CaImAn-MATLAB' target='_blank'>https://github.com/flatironinstitute/CaImAn-MATLAB</a><a id="dx1-1024"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-22-6" class="td11"><a href="#Xpnevmatikakis2016simultaneous">Pnevmatikakis et al.</a> <a href="#Xpnevmatikakis2016simultaneous">2016</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-23-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-23-1" class="td11">22 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-23-2" class="td11">CNMF-E </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-23-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-23-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-23-5" class="td11">•  CNMF + background model to handle one-photon data.<br class="newline" />•  <a href='https://github.com/zhoupc/CNMF_E' target='_blank'>https://github.com/zhoupc/CNMF_E</a><a id="dx1-1025"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-23-6" class="td11"><a href="#Xzhou2016efficient">Zhou et al.</a> <a href="#Xzhou2016efficient">2016</a>, <a href="#Xzhou2018efficient">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-24-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-24-1" class="td11">23 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-24-2" class="td11">Apthorpe CNN </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-24-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-24-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-24-5" class="td11">•  Convolutional neural network (CNN). </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-24-6" class="td11"><a href="#Xapthorpe2016automatic">Apthorpe et al.</a> <a href="#Xapthorpe2016automatic">2016</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-25-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-25-1" class="td11">24 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-25-2" class="td11">moco </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-25-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-25-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-25-5" class="td11">•  Fourier-transform based motion correction.<br class="newline" />•  <a href='https://github.com/NTCColumbia/moco' target='_blank'>https://github.com/NTCColumbia/moco</a><a id="dx1-1026"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-25-6" class="td11"><a href="#Xdubbs2016moco">Dubbs et al.</a> <a href="#Xdubbs2016moco">2016</a> </td></tr><tr style="vertical-align:baseline;" id="TBL-1-26-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-26-1" class="td11">25 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-26-2" class="td11">Cytomine </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-26-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-26-4" class="td11">Analysis GUI </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-26-5" class="td11">•  Analysis of large-scale imaging data.<br class="newline" />•  <a href='https://cytomine.be' target='_blank'>https://cytomine.be</a><a id="dx1-1027"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-26-6" class="td11"><a href="#Xmaree2016collaborative">Marée et al.</a> <a href="#Xmaree2016collaborative">2016</a></td> </tr><tr style="vertical-align:baseline;" id="TBL-1-27-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-27-1" class="td11">26 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-27-2" class="td11">ROI clustering </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-27-3" class="td11">2016</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-27-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-27-5" class="td11">•  Select high-intensity pixels then perform clustering to segment.<br class="newline" />•  <a href='https://www.bu.edu/hanlab/files/2016/02/pfgc.zip' target='_blank'>https://www.bu.edu/hanlab/files/2016/02/pfgc.zip</a><a id="dx1-1028"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-27-6" class="td11"><a href="#Xmohammed2016integrative">Mohammed et al.</a> <a href="#Xmohammed2016integrative">2016</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-28-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-28-1" class="td11">27 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-28-2" class="td11"><span class="rm-lmbx-12">CELLMax </span>(conference) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-28-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-28-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-28-5" class="td11">•  Cell segmentation and activity trace extraction using a maximum likelihood approach. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-28-6" class="td11"><a href="#XAhanonu2018sfnposter">Ahanonu et al.</a> <a href="#XAhanonu2018sfnposter">2018</a>, <a href="#XAhanonu2017sfnposter">2017</a>; <a href="#Xahanonu2018neural">Ahanonu</a> <a href="#Xahanonu2018neural">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-29-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-29-1" class="td11">28 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-29-2" class="td11">sc-CNMF </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-29-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-29-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-29-5" class="td11">•  CNMF + GMM/RNN seed cleansing. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-29-6" class="td11"><a href="#Xlu2017seeds">Lu et al.</a> <a href="#Xlu2017seeds">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-30-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-30-1" class="td11">29 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-30-2" class="td11">OASIS </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-30-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-30-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-30-5" class="td11">•  Generalized pool adjacent violators algorithm.<br class="newline" />•  <a href='https://github.com/zhoupc/OASIS_matlab' target='_blank'>https://github.com/zhoupc/OASIS_matlab</a><a id="dx1-1029"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-30-6" class="td11"><a href="#Xfriedrich2017fast">Friedrich et al.</a> <a href="#Xfriedrich2017fast">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-31-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-31-1" class="td11">30 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-31-2" class="td11">ABLE </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-31-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-31-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-31-5" class="td11">•  Multiple active contours and a cost function to identify cells in 2P data.<br class="newline" />•  <a href='https://github.com/StephanieRey/ABLE' target='_blank'>https://github.com/StephanieRey/ABLE</a><a id="dx1-1030"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-31-6" class="td11"><a href="#Xreynolds2017able">Reynolds et al.</a> <a href="#Xreynolds2017able">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-32-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-32-1" class="td11">31 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-32-2" class="td11">SCALPEL </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-32-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-32-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-32-5" class="td11">•  Dictionary learning, dissimilarity, and clustering.<br class="newline" />•  <a href='https://cran.r-project.org/web/packages/scalpel/index.html' target='_blank'>https://cran.r-project.org/web/packages/scalpel/index.html</a><a id="dx1-1031"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-32-6" class="td11"><a href="#Xpetersen2017scalpel">Petersen et al.</a> <a href="#Xpetersen2017scalpel">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-33-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-33-1" class="td11">32 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-33-2" class="td11">HNCcorr </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-33-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-33-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-33-5" class="td11">•  Combinatorial optimization (correlation space analysis).<br class="newline" />•  <a href='https://github.com/hochbaumGroup/HNCcorr' target='_blank'>https://github.com/hochbaumGroup/HNCcorr</a><a id="dx1-1032"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-33-6" class="td11"><a href="#Xspaen2017hnccorr">Spaen et al.</a> <a href="#Xspaen2017hnccorr">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-34-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-34-1" class="td11">33 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-34-2" class="td11">OnACID </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-34-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-34-4" class="td11">Cell extraction (online) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-34-5" class="td11">•  NMF variant for online Ca<sup class="textsuperscript"><span class="rm-lmr-10x-x-109">2+</span></sup> imaging processing. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-34-6" class="td11"><a href="#Xgiovannucci2017onacid">Giovannucci et al.</a> <a href="#Xgiovannucci2017onacid">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-35-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-35-1" class="td11">34 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-35-2" class="td11">EXTRACT </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-35-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-35-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-35-5" class="td11">•  Robust statistical estimation. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-35-6" class="td11"><a href="#Xinan2017robust">Inan et al.</a> <a href="#Xinan2017robust">2017</a> </td></tr><tr style="vertical-align:baseline;" id="TBL-1-36-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-36-1" class="td11">35 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-36-2" class="td11">NETCAL </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-36-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-36-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-36-5" class="td11">•  Calcium imaging analysis GUI.<br class="newline" />•  <a href='https://github.com/orlandi/netcal' target='_blank'>https://github.com/orlandi/netcal</a><a id="dx1-1033"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-36-6" class="td11"><a href="#Xorlandinetcal">Orlandi et al.</a></td> </tr><tr style="vertical-align:baseline;" id="TBL-1-37-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-37-1" class="td11">36 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-37-2" class="td11">NoRMCorre </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-37-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-37-4" class="td11">Motion correction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-37-5" class="td11">•  Piecewise rigid motion correction.<br class="newline" />•  <a href='https://github.com/simonsfoundation/NoRMCorre' target='_blank'>https://github.com/simonsfoundation/NoRMCorre</a><a id="dx1-1034"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-37-6" class="td11"><a href="#Xpnevmatikakis2017normcorre">Pnevmatikakis and Giovannucci</a> <a href="#Xpnevmatikakis2017normcorre">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-38-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-38-1" class="td11">37 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-38-2" class="td11">CellReg </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-38-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-38-4" class="td11">Cross-session alignment</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-38-5" class="td11">•  Alignment of cells across days using a probabilistic approach.<br class="newline" />•  <a href='https://github.com/zivlab/CellReg' target='_blank'>https://github.com/zivlab/CellReg</a><a id="dx1-1035"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-38-6" class="td11"><a href="#Xsheintuch2017tracking">Sheintuch et al.</a> <a href="#Xsheintuch2017tracking">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-39-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-39-1" class="td11">38 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-39-2" class="td11">NeuroSeg </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-39-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-39-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-39-5" class="td11">•  Filtering and seed/clustering based cell segmentation.<br class="newline" />•  <a href='https://github.com/baidatong/NeuroSeg' target='_blank'>https://github.com/baidatong/NeuroSeg</a><a id="dx1-1036"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-39-6" class="td11"><a href="#Xguan2018neuroseg">Guan et al.</a> <a href="#Xguan2018neuroseg">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-40-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-40-1" class="td11">39 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-40-2" class="td11">CNMF-E+ </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-40-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-40-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-40-5" class="td11">•  Shrinkage estimation to improve CNMF-E initialization. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-40-6" class="td11"><a href="#Xtakekawa2017automatic">Takekawa et al.</a> <a href="#Xtakekawa2017automatic">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-41-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-41-1" class="td11">40 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-41-2" class="td11">Toolbox-Romano </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-41-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-41-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-41-5" class="td11">•  Full analysis pipeline with ROI-based segmentation<br class="newline" />•  <a href='https://github.com/zebrain-lab/Toolbox-Romano-et-al' target='_blank'>https://github.com/zebrain-lab/Toolbox-Romano-et-al</a><a id="dx1-1037"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-41-6" class="td11"><a href="#Xromano2017integrated">Romano et al.</a> <a href="#Xromano2017integrated">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-42-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-42-1" class="td11">41 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-42-2" class="td11">SamuROI </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-42-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-42-4" class="td11">Analysis GUI </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-42-5" class="td11">•  GUI for data visualization<br class="newline" />•  <a href='https://github.com/samuroi/SamuROI' target='_blank'>https://github.com/samuroi/SamuROI</a><a id="dx1-1038"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-42-6" class="td11"><a href="#Xrueckl2017samuroi">Rueckl et al.</a> <a href="#Xrueckl2017samuroi">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-43-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-43-1" class="td11">42 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-43-2" class="td11">KNIME </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-43-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-43-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-43-5" class="td11">•  Workflow manager for data analysis.<br class="newline" />•  <a href='https://www.knime.com' target='_blank'>https://www.knime.com</a><a id="dx1-1039"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-43-6" class="td11"><a href="#Xfillbrunn2017knime">Fillbrunn et al.</a> <a href="#Xfillbrunn2017knime">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-44-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-44-1" class="td11">43 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-44-2" class="td11">U-Net2DS </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-44-3" class="td11">2017</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-44-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-44-5" class="td11">•  Evaluated several deep learning models on Neurofinder, U-Net2DS best.<br class="newline" />•  <a href='https://github.com/alexklibisz/deep-calcium' target='_blank'>https://github.com/alexklibisz/deep-calcium</a><a id="dx1-1040"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-44-6" class="td11"><a href="#Xklibisz2017fast">Klibisz et al.</a> <a href="#Xklibisz2017fast">2017</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-45-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-45-1" class="td11">44 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-45-2" class="td11"><span class="rm-lmbx-12">CLEAN </span>(conference) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-45-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-45-4" class="td11">Cell sorting </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-45-5" class="td11">•  Machine learning based cell sorting of cell extraction outputs based on image and activity trace features. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-45-6" class="td11"><a href="#XAhanonu2018sfnposter">Ahanonu et al.</a> <a href="#XAhanonu2018sfnposter">2018</a>; <a href="#Xahanonu2018neural">Ahanonu</a> <a href="#Xahanonu2018neural">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-46-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-46-1" class="td11">45 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-46-2" class="td11">FISSA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-46-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-46-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-46-5" class="td11">•  Neuropil decontamination using local region around cell.<br class="newline" />•  <a href='https://github.com/rochefort-lab/fissa' target='_blank'>https://github.com/rochefort-lab/fissa</a><a id="dx1-1041"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-46-6" class="td11"><a href="#Xkeemink2018fissa">Keemink et al.</a> <a href="#Xkeemink2018fissa">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-47-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-47-1" class="td11">46 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-47-2" class="td11">LSSC </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-47-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-47-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-47-5" class="td11">•  Spectral clustering; variant to find local subset of eigenvectors. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-47-6" class="td11"><a href="#Xmishne2018automated">Mishne et al.</a> <a href="#Xmishne2018automated">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-48-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-48-1" class="td11">47 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-48-2" class="td11">PMD - PCA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-48-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-48-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-48-5" class="td11">•  Spatially-localized penalized matrix decomposition for denoising; compression; and improved demixing.<br class="newline" />•  <a href='https://github.com/paninski-lab/funimag' target='_blank'>https://github.com/paninski-lab/funimag</a><a id="dx1-1042"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-48-6" class="td11"><a href="#Xbuchanan2018penalized">Buchanan et al.</a> <a href="#Xbuchanan2018penalized">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-49-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-49-1" class="td11">48 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-49-2" class="td11">MIN1PIPE </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-49-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-49-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-49-5" class="td11">•  Pre-processing to enhance neural signals then sc-CNMF for cell extraction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-49-6" class="td11"><a href="#XLu2018min1pipe">Lu et al.</a> <a href="#XLu2018min1pipe">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-50-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-50-1" class="td11">49 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-50-2" class="td11">CaImAn (preprint) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-50-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-50-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-50-5" class="td11">•  CNMF + several other processing tools. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-50-6" class="td11"><a href="#XGiovannucci2018">Giovannucci et al.</a> <a href="#XGiovannucci2018">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-51-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-51-1" class="td11">50 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-51-2" class="td11">SEUDO (preprint) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-51-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-51-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-51-5" class="td11">•  Mixture of Gaussians + maximum likelihood; post-hoc activity trace correction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-51-6" class="td11"><a href="#Xgauthier2018detecting">Gauthier et al.</a> <a href="#Xgauthier2018detecting">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-52-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-52-1" class="td11">51 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-52-2" class="td11">ACSAT </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-52-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-52-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-52-5" class="td11">•  Global and local adaptive thresholding to identify neurons.<br class="newline" />•  <a href='https://github.com/sshen8/acsat' target='_blank'>https://github.com/sshen8/acsat</a><a id="dx1-1043"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-52-6" class="td11"><a href="#Xshen2018automatic">Shen et al.</a> <a href="#Xshen2018automatic">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-53-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-53-1" class="td11">52 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-53-2" class="td11">onlineMotionCorrection </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-53-3" class="td11">2018</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-53-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-53-5" class="td11">•  Tested multiple algorithms and developed an online motion correction pipeline.<br class="newline" />•  <a href='https://github.com/amitani/onlineMotionCorrection' target='_blank'>https://github.com/amitani/onlineMotionCorrection</a><a id="dx1-1044"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-53-6" class="td11"><a href="#Xmitani2018real">Mitani and Komiyama</a> <a href="#Xmitani2018real">2018</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-54-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-54-1" class="td11">53 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-54-2" class="td11"><span class="rm-lmbx-12">CIAtah </span></td><td style="white-space:nowrap; text-align:left;" id="TBL-1-54-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-54-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-54-5" class="td11">•  1P and 2P Imaging analysis pipeline supporting PCA-ICA, CNMF, CELLMax, EXTRACT, etc.<br class="newline" />•  <a href='https://github.com/bahanonu/ciatah' target='_blank'>https://github.com/bahanonu/ciatah</a><a id="dx1-1045"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-54-6" class="td11"><a href="#Xcorder2019amygdalar">Corder et al.</a> <a href="#Xcorder2019amygdalar">2019</a>; <a href="#Xahanonu2018neural">Ahanonu</a> <a href="#Xahanonu2018neural">2018</a>; <a href="#Xahanonu2022recording">Ahanonu and Corder</a> <a href="#Xahanonu2022recording">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-55-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-55-1" class="td11">54 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-55-2" class="td11">NAOMi (bioRxiv) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-55-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-55-4" class="td11">Simulator </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-55-5" class="td11">•  Generative model for creating simulated calcium imaging movies. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-55-6" class="td11"><a href="#Xcharles2019neural">Charles et al.</a> <a href="#Xcharles2019neural">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-56-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-56-1" class="td11">55 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-56-2" class="td11">CALIMA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-56-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-56-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-56-5" class="td11">•  Calcium imaging analysis GUI. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-56-6" class="td11"><a href="#Xradstake2019calima">Radstake et al.</a> <a href="#Xradstake2019calima">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-57-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-57-1" class="td11">56 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-57-2" class="td11">STNeuroNet </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-57-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-57-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-57-5" class="td11">•  Convolutional neural network to detect and segment cells. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-57-6" class="td11"><a href="#Xsoltanian2019fast">Soltanian-Zadeh et al.</a> <a href="#Xsoltanian2019fast">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-58-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-58-1" class="td11">57 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-58-2" class="td11">AQuA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-58-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-58-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-58-5" class="td11">•  Astrocyte imaging focused. Non-ROI cluster and propagation based detection of events. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-58-6" class="td11"><a href="#Xwang2019accurate">Wang et al.</a> <a href="#Xwang2019accurate">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-59-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-59-1" class="td11">58 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-59-2" class="td11">CaImAn </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-59-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-59-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-59-5" class="td11">•  Popular calcium imaging pipeline that includes CNMF + several other processing tools.<br class="newline" />•  <a href='https://github.com/flatironinstitute/CaImAn' target='_blank'>https://github.com/flatironinstitute/CaImAn</a><a id="dx1-1046"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-59-6" class="td11"><a href="#Xgiovannucci2019caiman">Giovannucci et al.</a> <a href="#Xgiovannucci2019caiman">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-60-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-60-1" class="td11">59 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-60-2" class="td11">DL+RWL1-SF </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-60-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-60-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-60-5" class="td11">•  Dictionary learning and spatial correlation based cell extraction. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-60-6" class="td11"><a href="#Xmishne2019learning">Mishne and Charles</a> <a href="#Xmishne2019learning">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-61-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-61-1" class="td11">60 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-61-2" class="td11">Segment2P </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-61-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-61-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-61-5" class="td11">•  Pre-process images and run through DeepLabV3.<br class="newline" />•  <a href='https://github.com/NoahDolev/Segment2P' target='_blank'>https://github.com/NoahDolev/Segment2P</a><a id="dx1-1047"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-61-6" class="td11"><a href="#Xdolev2019segment2p">Dolev et al.</a> <a href="#Xdolev2019segment2p">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-62-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-62-1" class="td11">61 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-62-2" class="td11">LANMC </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-62-3" class="td11">2019</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-62-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-62-5" class="td11">•  Long short-term memory non-rigid motion correction, reduce computational cost by predicting non-rigid motion. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-62-6" class="td11"><a href="#Xchen2019lanmc">Chen et al.</a> <a href="#Xchen2019lanmc">2019</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-63-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-63-1" class="td11">62 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-63-2" class="td11">marked point processes </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-63-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-63-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-63-5" class="td11">•  Probabilistic generative model, specifically a marked point process, to extract activity traces. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-63-6" class="td11"><a href="#Xshibue2020deconvolution">Shibue and Komaki</a> <a href="#Xshibue2020deconvolution">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-64-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-64-1" class="td11">63 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-64-2" class="td11">LocaNMF </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-64-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-64-4" class="td11">Region extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-64-5" class="td11">•  Localized semi-nonnegative matrix factorization for extracting active regions.<br class="newline" />•  <a href='https://github.com/ikinsella/locaNMF' target='_blank'>https://github.com/ikinsella/locaNMF</a><a id="dx1-1048"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-64-6" class="td11"><a href="#Xsaxena2020localized">Saxena et al.</a> <a href="#Xsaxena2020localized">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-65-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-65-1" class="td11">64 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-65-2" class="td11">EZcalcium </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-65-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-65-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-65-5" class="td11">•  Calcium imaging analysis toolbox.<br class="newline" />•  <a href='https://github.com/porteralab/EZcalcium' target='_blank'>https://github.com/porteralab/EZcalcium</a><a id="dx1-1049"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-65-6" class="td11"><a href="#Xcantu2020ezcalcium">Cantu et al.</a> <a href="#Xcantu2020ezcalcium">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-66-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-66-1" class="td11">65 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-66-2" class="td11">OnACID-E + ring CNN </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-66-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-66-4" class="td11">Cell extraction (online) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-66-5" class="td11">•  OnACID for miniscope and new ring CNN background model to improve accuracy.<br class="newline" />•  <a href='https://github.com/flatironinstitute/CaImAn' target='_blank'>https://github.com/flatironinstitute/CaImAn</a><a id="dx1-1050"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-66-6" class="td11"><a href="#Xfriedrich2020online">Friedrich et al.</a> <a href="#Xfriedrich2020online">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-67-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-67-1" class="td11">66 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-67-2" class="td11">Auto CNMF-E sorting </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-67-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-67-4" class="td11">Cell sorting </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-67-5" class="td11">•  Machine learning (AutoML) based curation of CNMF-E outputs.<br class="newline" />•  <a href='https://github.com/jf-lab/cnmfe-reviewer' target='_blank'>https://github.com/jf-lab/cnmfe-reviewer</a><a id="dx1-1051"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-67-6" class="td11"><a href="#Xtran2020automated">Tran et al.</a> <a href="#Xtran2020automated">2020a</a>,<a href="#Xtran2020automated2">b</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-68-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-68-1" class="td11">67 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-68-2" class="td11">DeepInterpolation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-68-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-68-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-68-5" class="td11">•  Encoder-decoder architecture with 2D conv. to denoise imaging data.<br class="newline" />•  <a href='https://github.com/AllenInstitute/deepinterpolation' target='_blank'>https://github.com/AllenInstitute/deepinterpolation</a><a id="dx1-1052"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-68-6" class="td11"><a href="#Xlecoq2020removing">Lecoq et al.</a> <a href="#Xlecoq2020removing">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-69-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-69-1" class="td11">68 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-69-2" class="td11">BIAFLOWS </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-69-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-69-4" class="td11">Benchmarking </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-69-5" class="td11">•  Framework for benchmarking imaging analysis workflows.<br class="newline" />•  <a href='https://biaflows.neubias.org' target='_blank'>https://biaflows.neubias.org</a><a id="dx1-1053"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-69-6" class="td11"><a href="#Xrubens2020biaflows">Rubens et al.</a> <a href="#Xrubens2020biaflows">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-70-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-70-1" class="td11">69 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-70-2" class="td11">FIBSI </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-70-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-70-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-70-5" class="td11">•  Extension of Ramer-Douglas-Peucker algorithm to identify baseline that is used for signal detection.<br class="newline" />•  <a href='https://github.com/rmcassidy/FIBSI_program' target='_blank'>https://github.com/rmcassidy/FIBSI_program</a><a id="dx1-1054"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-70-6" class="td11"><a href="#Xcassidy2020frequency">Cassidy et al.</a> <a href="#Xcassidy2020frequency">2020</a>; <a href="#Xalles2021chronic">Alles et al.</a> <a href="#Xalles2021chronic">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-71-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-71-1" class="td11">70 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-71-2" class="td11">DISCo </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-71-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-71-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-71-5" class="td11">•  Pixel correlation and deep learning (CNN) + graph based segmentation.<br class="newline" />•  <a href='https://github.com/EKirschbaum/DISCo' target='_blank'>https://github.com/EKirschbaum/DISCo</a><a id="dx1-1055"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-71-6" class="td11"><a href="#Xkirschbaum2020disco">Kirschbaum et al.</a> <a href="#Xkirschbaum2020disco">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-72-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-72-1" class="td11">71 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-72-2" class="td11">DeepCINAC </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-72-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-72-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-72-5" class="td11">•  Trace analysis after human labeling followed by CNNs + bidirectional long-short term memory (LSTM) network.<br class="newline" />•  <a href='https://gitlab.com/cossartlab/deepcinac' target='_blank'>https://gitlab.com/cossartlab/deepcinac</a><a id="dx1-1056"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-72-6" class="td11"><a href="#Xdenis2020deepcinac">Denis et al.</a> <a href="#Xdenis2020deepcinac">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-73-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-73-1" class="td11">72 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-73-2" class="td11">NDSEP </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-73-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-73-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-73-5" class="td11">•  Dataflow framework for real-time calcium imaging processing.<br class="newline" />•  <a href='http://dspcad-www.iacs.umd.edu/bcnm/index.html' target='_blank'>http://dspcad-www.iacs.umd.edu/bcnm/index.html</a><a id="dx1-1057"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-73-6" class="td11"><a href="#Xlee2020real">Lee et al.</a> <a href="#Xlee2020real">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-74-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-74-1" class="td11">73 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-74-2" class="td11">DeepBrainSeg </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-74-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-74-4" class="td11">Segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-74-5" class="td11">•  Dual-pathway CNN to learn local and contextual features. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-74-6" class="td11"><a href="#Xtan2020deepbrainseg">Tan et al.</a> <a href="#Xtan2020deepbrainseg">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-75-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-75-1" class="td11">74 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-75-2" class="td11">RT-3DMC </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-75-3" class="td11">2020</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-75-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-75-5" class="td11">•  Bead or soma tracking for real-time motion correction during 2P imaging.<br class="newline" />•  <a href='https://github.com/SilverLabUCL/SilverLab-Microscope' target='_blank'>https://github.com/SilverLabUCL/SilverLab-Microscope</a><a id="dx1-1058"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-75-6" class="td11"><a href="#Xgriffiths2020real">Griffiths et al.</a> <a href="#Xgriffiths2020real">2020</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-76-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-76-1" class="td11">75 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-76-2" class="td11">Cellpose </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-76-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-76-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-76-5" class="td11">•  Neural network and gradient-based cell segmentation.<br class="newline" />•  <a href='https://github.com/mouseland/cellpose' target='_blank'>https://github.com/mouseland/cellpose</a><a id="dx1-1059"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-76-6" class="td11"><a href="#Xstringer2021cellpose">Stringer et al.</a> <a href="#Xstringer2021cellpose">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-77-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-77-1" class="td11">76 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-77-2" class="td11">NAOMi </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-77-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-77-4" class="td11">Simulator </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-77-5" class="td11">•  Detailed model simulation for benchmarking calcium imaging algorithms.<br class="newline" />•  <a href='https://bitbucket.org/adamshch/naomi_sim/src/master/' target='_blank'>https://bitbucket.org/adamshch/naomi_sim/src/master/</a><a id="dx1-1060"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-77-6" class="td11"><a href="#Xsong2021neural">Song et al.</a> <a href="#Xsong2021neural">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-78-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-78-1" class="td11">77 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-78-2" class="td11">OnACID-E + ring CNN </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-78-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-78-4" class="td11">Cell extraction (online) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-78-5" class="td11">•  OnACID for 1P data and ring CNN background model.<br class="newline" />•  <a href='https://github.com/flatironinstitute/CaImAn' target='_blank'>https://github.com/flatironinstitute/CaImAn</a><a id="dx1-1061"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-78-6" class="td11"><a href="#Xfriedrich2021online">Friedrich et al.</a> <a href="#Xfriedrich2021online">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-79-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-79-1" class="td11">78 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-79-2" class="td11">EXTRACT </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-79-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-79-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-79-5" class="td11">•  Robust statistics based cell extraction.<br class="newline" />•  <a href='https://github.com/schnitzer-lab/EXTRACT-public' target='_blank'>https://github.com/schnitzer-lab/EXTRACT-public</a><a id="dx1-1062"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-79-6" class="td11"><a href="#Xinan2021fast">Inan et al.</a> <a href="#Xinan2021fast">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-80-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-80-1" class="td11">79 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-80-2" class="td11">Minian </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-80-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-80-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-80-5" class="td11">•  Imaging analysis pipeline with CNMF for cell extraction, in part using Jupyter notebooks with GUI elements.<br class="newline" />•  <a href='https://github.com/DeniseCaiLab/minian' target='_blank'>https://github.com/DeniseCaiLab/minian</a><a id="dx1-1063"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-80-6" class="td11"><a href="#Xdong2021minian">Dong et al.</a> <a href="#Xdong2021minian">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-81-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-81-1" class="td11">80 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-81-2" class="td11">Mesmerize </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-81-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-81-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-81-5" class="td11">•  Imaging analysis platform with CaImAn for cell extraction, import support for other cell extraction algorithms.<br class="newline" />•  <a href='https://github.com/kushalkolar/MESmerize' target='_blank'>https://github.com/kushalkolar/MESmerize</a><a id="dx1-1064"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-81-6" class="td11"><a href="#Xkolar2021mesmerize">Kolar et al.</a> <a href="#Xkolar2021mesmerize">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-82-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-82-1" class="td11">81 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-82-2" class="td11">DeepInterpolation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-82-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-82-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-82-5" class="td11">•  Encoder-decoder architecture with 2D conv. to denoise imaging data.<br class="newline" />•  <a href='https://github.com/AllenInstitute/deepinterpolation' target='_blank'>https://github.com/AllenInstitute/deepinterpolation</a><a id="dx1-1065"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-82-6" class="td11"><a href="#Xlecoq2021removing">Lecoq et al.</a> <a href="#Xlecoq2021removing">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-83-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-83-1" class="td11">82 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-83-2" class="td11">BEAR </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-83-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-83-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-83-5" class="td11">•  Neural network approximation of PCA for cell extraction.<br class="newline" />•  <a href='https://github.com/NICALab/BEAR' target='_blank'>https://github.com/NICALab/BEAR</a><a id="dx1-1066"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-83-6" class="td11"><a href="#Xhan2021efficient">Han et al.</a> <a href="#Xhan2021efficient">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-84-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-84-1" class="td11">83 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-84-2" class="td11">CaPTure </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-84-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-84-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-84-5" class="td11">•  ROI segmentation and activity extraction.<br class="newline" />•  <a href='https://github.com/LieberInstitute/CaPTure' target='_blank'>https://github.com/LieberInstitute/CaPTure</a><a id="dx1-1067"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-84-6" class="td11"><a href="#Xtippani2021capture">Tippani et al.</a> <a href="#Xtippani2021capture">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-85-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-85-1" class="td11">84 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-85-2" class="td11">CASCADE </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-85-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-85-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-85-5" class="td11">•  Spike inference based on dual ephys/calcium imaging recordings.<br class="newline" />•  <a href='https://github.com/HelmchenLabSoftware/Cascade' target='_blank'>https://github.com/HelmchenLabSoftware/Cascade</a><a id="dx1-1068"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-85-6" class="td11"><a href="#Xrupprecht2021database">Rupprecht et al.</a> <a href="#Xrupprecht2021database">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-86-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-86-1" class="td11">85 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-86-2" class="td11">VolPy </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-86-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-86-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-86-5" class="td11">•  Voltage imaging analysis pipeline integrated into CaImAn.<br class="newline" />•  <a href='https://github.com/flatironinstitute/CaImAn' target='_blank'>https://github.com/flatironinstitute/CaImAn</a><a id="dx1-1069"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-86-6" class="td11"><a href="#Xcai2021volpy">Cai et al.</a> <a href="#Xcai2021volpy">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-87-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-87-1" class="td11">86 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-87-2" class="td11">DeepCAD </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-87-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-87-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-87-5" class="td11">•  Deep neural network based denoising.<br class="newline" />•  <a href='https://github.com/cabooster/DeepCAD-RT' target='_blank'>https://github.com/cabooster/DeepCAD-RT</a><a id="dx1-1070"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-87-6" class="td11"><a href="#Xli2021reinforcing">Li et al.</a> <a href="#Xli2021reinforcing">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-88-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-88-1" class="td11">87 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-88-2" class="td11">SpecSeg </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-88-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-88-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-88-5" class="td11">•  Spectral density of pixels to identify ROIs. Also incorporates motion correction and cross-session matching.<br class="newline" />•  <a href='https://github.com/Leveltlab/SpectralSegmentation' target='_blank'>https://github.com/Leveltlab/SpectralSegmentation</a><a id="dx1-1071"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-88-6" class="td11"><a href="#Xde2021specseg">de Kraker et al.</a> <a href="#Xde2021specseg">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-89-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-89-1" class="td11">88 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-89-2" class="td11">FIOLA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-89-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-89-4" class="td11">Cell extraction (online) </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-89-5" class="td11">•  GPU- and computational graph-based speed-ups along with non-negative least squares for post-initialization signal extraction.<br class="newline" />•  <a href='https://github.com/nel-lab/FIOLA' target='_blank'>https://github.com/nel-lab/FIOLA</a><a id="dx1-1072"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-89-6" class="td11"><a href="#Xgiovannucci2021fiola">Giovannucci et al.</a> <a href="#Xgiovannucci2021fiola">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-90-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-90-1" class="td11">89 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-90-2" class="td11">PatchWarp </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-90-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-90-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-90-5" class="td11">•  Affine transformation of subfields followed by stitching subfields together.<br class="newline" />•  <a href='https://github.com/ryhattori/PatchWarp' target='_blank'>https://github.com/ryhattori/PatchWarp</a><a id="dx1-1073"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-90-6" class="td11"><a href="#XHattori2021.11.10.468164">Hattori and Komiyama</a> <a href="#XHattori2021.11.10.468164">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-91-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-91-1" class="td11">90 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-91-2" class="td11">MVG-CNN </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-91-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-91-4" class="td11">Region extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-91-5" class="td11">•  Automated sleep states classification using multiplex visibility graphs and deep learning. <a href='https://physionet.org/content/calcium-imaging-sleep-state/1.0.0/' target='_blank'>Data URL</a><a id="dx1-1074"></a>.<br class="newline" />•  <a href='https://github.com/comp-imaging-sci/MVG-CNN' target='_blank'>https://github.com/comp-imaging-sci/MVG-CNN</a><a id="dx1-1075"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-91-6" class="td11"><a href="#Xzhang2021automated">Zhang et al.</a> <a href="#Xzhang2021automated">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-92-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-92-1" class="td11">91 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-92-2" class="td11">Flow-Registration </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-92-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-92-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-92-5" class="td11">•  Variational optical flow for non-uniform motion correction<br class="newline" />•  <a href='https://github.com/phflot/flow_registration' target='_blank'>https://github.com/phflot/flow_registration</a><a id="dx1-1076"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-92-6" class="td11"><a href="#Xflotho2022software">Flotho et al.</a> <a href="#Xflotho2022software">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-93-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-93-1" class="td11">92 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-93-2" class="td11">SUNS </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-93-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-93-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-93-5" class="td11">•  Cell segmentation using shallow U-Nets.<br class="newline" />•  <a href='https://github.com/YijunBao/Shallow-UNet-Neuron-Segmentation_SUNS' target='_blank'>https://github.com/YijunBao/Shallow-UNet-Neuron-Segmentation_SUNS</a><a id="dx1-1077"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-93-6" class="td11"><a href="#Xbao2021segmentation">Bao et al.</a> <a href="#Xbao2021segmentation">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-94-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-94-1" class="td11">93 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-94-2" class="td11">Carignan </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-94-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-94-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-94-5" class="td11">•  Online cell extraction and triggering based on OnACID and CaImAn.<br class="newline" />•  <a href='https://github.com/tzklab/carignan' target='_blank'>https://github.com/tzklab/carignan</a><a id="dx1-1078"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-94-6" class="td11"><a href="#Xtaniguchi2021open">Taniguchi et al.</a> <a href="#Xtaniguchi2021open">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-95-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-95-1" class="td11">94 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-95-2" class="td11">MullenClassifier </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-95-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-95-4" class="td11">Cell sorting </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-95-5" class="td11">•  Feature extraction from cell images and tracs followed by supervised learning classifier. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-95-6" class="td11"><a href="#Xmullen2021automated">Mullen et al.</a> <a href="#Xmullen2021automated">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-96-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-96-1" class="td11">95 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-96-2" class="td11">timeUnet </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-96-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-96-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-96-5" class="td11">•  Deep learning for denoising with temporal information added in.<br class="newline" />•  <a href='https://github.com/BoHuangLab/Transfer-Learning-Denoising/' target='_blank'>https://github.com/BoHuangLab/Transfer-Learning-Denoising/</a><a id="dx1-1079"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-96-6" class="td11"><a href="#Xwang2021image">Wang et al.</a> <a href="#Xwang2021image">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-97-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-97-1" class="td11">96 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-97-2" class="td11">EMC<sup class="textsuperscript"><span class="rm-lmr-10x-x-109">2</span></sup> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-97-3" class="td11">2021</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-97-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-97-5" class="td11">•  Wavelet decomposition to detect bright spots followed by motion correction with multiple hypothesis tracking and computing elastic deformation.<br class="newline" />•  <a href='https://icy.bioimageanalysis.org/plugin/elastic-motion-correction-concatenation-emc2-of-tracks/' target='_blank'>https://icy.bioimageanalysis.org/plugin/elastic-motion-correction-concatenation-emc2-of-tracks/</a><a id="dx1-1080"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-97-6" class="td11"><a href="#Xlagache2021tracking">Lagache et al.</a> <a href="#Xlagache2021tracking">2021</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-98-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-98-1" class="td11">97 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-98-2" class="td11">GraFT </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-98-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-98-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-98-5" class="td11">•  Dictionary-based learning of activity traces followed by graph-based segmentation.<br class="newline" />•  <a href='https://github.com/adamshch/GraFT-analysis' target='_blank'>https://github.com/adamshch/GraFT-analysis</a><a id="dx1-1081"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-98-6" class="td11"><a href="#Xcharles2022graft">Charles et al.</a> <a href="#Xcharles2022graft">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-99-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-99-1" class="td11">98 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-99-2" class="td11">CaPTure </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-99-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-99-4" class="td11">Analysis pipeline </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-99-5" class="td11">•  Binary/watershed segmentation followed by ROI-based mean traces.<br class="newline" />•  <a href='https://github.com/LieberInstitute/CaPTure' target='_blank'>https://github.com/LieberInstitute/CaPTure</a><a id="dx1-1082"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-99-6" class="td11"><a href="#Xtippani2022capture">Tippani et al.</a> <a href="#Xtippani2022capture">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-100-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-100-1" class="td11">99 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-100-2" class="td11">SpecSeg </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-100-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-100-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-100-5" class="td11">•  Cross spectral power-based segmentation of neurons and neurites.<br class="newline" />•  <a href='https://github.com/Leveltlab/SpectralSegmentation' target='_blank'>https://github.com/Leveltlab/SpectralSegmentation</a><a id="dx1-1083"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-100-6" class="td11"><a href="#Xde2022specseg">de Kraker et al.</a> <a href="#Xde2022specseg">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-101-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-101-1" class="td11">100</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-101-2" class="td11">CITE-On </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-101-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-101-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-101-5" class="td11">•  Online cell detection and trace extraction using CNNs.<br class="newline" />•  <a href='https://gitlab.iit.it/fellin-public/cite-on' target='_blank'>https://gitlab.iit.it/fellin-public/cite-on</a><a id="dx1-1084"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-101-6" class="td11"><a href="#Xsita2022deep">Sità et al.</a> <a href="#Xsita2022deep">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-102-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-102-1" class="td11">101</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-102-2" class="td11">DL-assisted 2P fiberscope</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-102-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-102-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-102-5" class="td11">•  Denoising 2P fiberscope data using deep neural network (conditional generative adversarial network).<br class="newline" />•  <a href='https://figshare.com/articles/dataset/Data/19193792' target='_blank'>https://figshare.com/articles/dataset/Data/19193792</a><a id="dx1-1085"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-102-6" class="td11"><a href="#Xguan2022deep">Guan et al.</a> <a href="#Xguan2022deep">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-103-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-103-1" class="td11">102</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-103-2" class="td11">4SM </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-103-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-103-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-103-5" class="td11">•  Generative adversarial network for image segmentation.<br class="newline" />•  <a href='https://github.com/SharifAmit/4SM' target='_blank'>https://github.com/SharifAmit/4SM</a><a id="dx1-1086"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-103-6" class="td11"><a href="#Xkamran2022new">Kamran et al.</a> <a href="#Xkamran2022new">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-104-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-104-1" class="td11">103</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-104-2" class="td11">DeepCAD-RT </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-104-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-104-4" class="td11">Denoising </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-104-5" class="td11">•  Improved version of DeepCAD for real time performance.<br class="newline" />•  <a href='https://github.com/cabooster/DeepCAD-RT/' target='_blank'>https://github.com/cabooster/DeepCAD-RT/</a><a id="dx1-1087"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-104-6" class="td11"><a href="#Xli2023real">Li et al.</a> <a href="#Xli2023real">2023a</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-105-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-105-1" class="td11">104</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-105-2" class="td11">SEUDO </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-105-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-105-4" class="td11">Trace analysis </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-105-5" class="td11">•  Mixture of Gaussians + maximum likelihood; post-hoc activity trace correction.<br class="newline" />•  <a href='https://github.com/adamshch/SEUDO' target='_blank'>https://github.com/adamshch/SEUDO</a><a id="dx1-1088"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-105-6" class="td11"><a href="#Xgauthier2022detecting">Gauthier et al.</a> <a href="#Xgauthier2022detecting">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-106-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-106-1" class="td11">105</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-106-2" class="td11">AxialMotionCorrect </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-106-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-106-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-106-5" class="td11">•  Axial motion correction via multi-plane scanning plus maximum likelihood optimization.<br class="newline" />•  <a href='https://gitlab.com/anflores/axial_motion_correction' target='_blank'>https://gitlab.com/anflores/axial_motion_correction</a><a id="dx1-1089"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-106-6" class="td11"><a href="#Xflores2022axial">Flores-Valle and Seelig</a> <a href="#Xflores2022axial">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-107-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-107-1" class="td11">106</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-107-2" class="td11">FIFER </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-107-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-107-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-107-5" class="td11">•  Feature-based motion correction, finding features using a density-based estimating and clustering algorithm and matching features with a similarity metric for registration.<br class="newline" />•  <a href='https://github.com/Weiyi-Liu-Unique/FIFER' target='_blank'>https://github.com/Weiyi-Liu-Unique/FIFER</a><a id="dx1-1090"></a></td><td style="white-space:nowrap; text-align:left;" id="TBL-1-107-6" class="td11"><a href="#Xliu2022fast">Liu et al.</a> <a href="#Xliu2022fast">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-108-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-108-1" class="td11">107</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-108-2" class="td11">NWB </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-108-3" class="td11">2022</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-108-4" class="td11">Data handling </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-108-5" class="td11">•  Neurodata Without Borders (NWB) to standardize ephys and imaging data across tools.<br class="newline" />•  <a href='https://github.com/NeurodataWithoutBorders' target='_blank'>https://github.com/NeurodataWithoutBorders</a><a id="dx1-1091"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-108-6" class="td11"><a href="#Xrubel2022neurodata">Rübel et al.</a> <a href="#Xrubel2022neurodata">2022</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-109-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-109-1" class="td11">108</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-109-2" class="td11">DeCalciOn </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-109-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-109-4" class="td11">Online analysis pipeline</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-109-5" class="td11">•  Integrate hardware and software to online decode calcium signals.<br class="newline" />•  <a href='https://github.com/zhe-ch/ACTEV' target='_blank'>https://github.com/zhe-ch/ACTEV</a><a id="dx1-1092"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-109-6" class="td11"><a href="#Xchen2023hardware">Chen et al.</a> <a href="#Xchen2023hardware">2023</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-110-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-110-1" class="td11">109</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-110-2" class="td11">jGCaMP8 </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-110-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-110-4" class="td11">Calcium indicator </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-110-5" class="td11">•  Improved calcium indicators with increased sensitivity and reduced background. </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-110-6" class="td11"><a href="#Xzhang2023fast">Zhang et al.</a> <a href="#Xzhang2023fast">2023a</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-111-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-111-1" class="td11">110</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-111-2" class="td11">NeuroSeg-II </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-111-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-111-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-111-5" class="td11">•  2P cell segmentation using region-based convolutional neural network with modifications.<br class="newline" />•  <a href='https://github.com/XZH-James/NeuroSeg2' target='_blank'>https://github.com/XZH-James/NeuroSeg2</a><a id="dx1-1093"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-111-6" class="td11"><a href="#Xxu2023neuroseg">Xu et al.</a> <a href="#Xxu2023neuroseg">2023</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-112-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-112-1" class="td11">111</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-112-2" class="td11">CaliAli </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-112-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-112-4" class="td11">Cross-session alignment</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-112-5" class="td11">•  Cross-session alignment using vasculature information.<br class="newline" />•  <a href='https://github.com/CaliAli-PV/CaliAli' target='_blank'>https://github.com/CaliAli-PV/CaliAli</a><a id="dx1-1094"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-112-6" class="td11"><a href="#Xvergara2023caliali">Vergara et al.</a> <a href="#Xvergara2023caliali">2023</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-113-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-113-1" class="td11">112</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-113-2" class="td11">DeepWonder </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-113-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-113-4" class="td11">Cell extraction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-113-5" class="td11">•  Deep-learning-based cell finding for widefield datasets.<br class="newline" />•  <a href='https://github.com/yuanlong-o/Deep_widefield_cal_inferece' target='_blank'>https://github.com/yuanlong-o/Deep_widefield_cal_inferece</a><a id="dx1-1095"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-113-6" class="td11"><a href="#Xzhang2023rapid">Zhang et al.</a> <a href="#Xzhang2023rapid">2023b</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-114-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-114-1" class="td11">113</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-114-2" class="td11">ASTRA </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-114-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-114-4" class="td11">Cell segmentation </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-114-5" class="td11">•  Deep neural network for astrocyte segmentation.<br class="newline" />•  <a href='https://gitlab.iit.it/fellin-public/astra' target='_blank'>https://gitlab.iit.it/fellin-public/astra</a><a id="dx1-1096"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-114-6" class="td11"><a href="#Xbonato2023astra">Bonato et al.</a> <a href="#Xbonato2023astra">2023</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-115-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-115-1" class="td11">114</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-115-2" class="td11">SRDTrans </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-115-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-115-4" class="td11">Denoise </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-115-5" class="td11">•  Spatial redundancy for training followed by spatiotemporal transformer architecture to reduce CNN bias/issues.<br class="newline" />•  <a href='https://github.com/cabooster/SRDTrans' target='_blank'>https://github.com/cabooster/SRDTrans</a><a id="dx1-1097"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-115-6" class="td11"><a href="#Xli2023spatial">Li et al.</a> <a href="#Xli2023spatial">2023b</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-116-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-116-1" class="td11">115</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-116-2" class="td11">REALS </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-116-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-116-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-116-5" class="td11">•  Motion correction via simultaneous transformation and low rank and sparse decomposition with gradient-based updates.<br class="newline" />•  <a href='https://openaccess.thecvf.com/content/WACV2023/supplemental/Cho_Robust_and_Efficient_WACV_2023_supplemental.zip' target='_blank'>https://openaccess.thecvf.com/content/WACV2023/supplemental/Cho_Robust_and_Efficient_WACV_2023_supplemental.zip</a><a id="dx1-1098"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-116-6" class="td11"><a href="#Xcho2023robust">Cho et al.</a> <a href="#Xcho2023robust">2023</a> </td> </tr><tr style="vertical-align:baseline;" id="TBL-1-117-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-117-1" class="td11">116</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-117-2" class="td11"><span class="rm-lmbx-12">LD-MCM </span></td><td style="white-space:nowrap; text-align:left;" id="TBL-1-117-3" class="td11">2023</td><td style="white-space:nowrap; text-align:left;" id="TBL-1-117-4" class="td11">Motion correction </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-117-5" class="td11">•  Motion correction using deep learning feature identification and control point registration.<br class="newline" />•  <a href='https://github.com/bahanonu/ciatah' target='_blank'>https://github.com/bahanonu/ciatah</a><a id="dx1-1099"></a> </td><td style="white-space:nowrap; text-align:left;" id="TBL-1-117-6" class="td11"><a href="#Xahanonu2023long">Ahanonu et al.</a> <a href="#Xahanonu2023long">2023</a> </td> </tr><tr class="hline"><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td><td><hr /></td></tr><tr style="vertical-align:baseline;" id="TBL-1-118-"><td style="white-space:nowrap; text-align:left;" id="TBL-1-118-1" class="td11"> </td></tr></table> </div><span class="copyright">&copy;</span> Biafra Ahanonu 2018-2023. </figure> </div> <!--l. 68--><p class="indent" > <hr class='headingsep'> </p><!--l. 73--><p class="indent" > <a id="likesection.1"></a><a id="Q1-1-3"></a> </p> <h2 class="likechapterHead"><a id="x1-2000"></a>References</h2> <div class="thebibliography"> <p class="bibitem" ><span class="biblabel"> <a id="Xreddy1996fft"></a><span class="bibsp">   </span></span>B Srinivasa Reddy and Biswanath N Chatterji. An fft-based technique for translation, rotation, and scale-invariant image registration. <span class="rm-lmri-12">IEEE</span> <span class="rm-lmri-12">transactions on image processing</span>, 5(8):1266–1271, 1996. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xthevenaz1998pyramid"></a><span class="bibsp">   </span></span>Philippe Thevenaz, Urs E Ruttimann, and Michael Unser. A pyramid approach to subpixel registration based on intensity. <span class="rm-lmri-12">Image Processing,</span> <span class="rm-lmri-12">IEEE Transactions on</span>, 7(1):27–41, 1998. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xforoosh2002extension"></a><span class="bibsp">   </span></span>Hassan Foroosh, Josiane B Zerubia, and Marc Berthod. Extension of phase correlation to subpixel registration. <span class="rm-lmri-12">IEEE transactions on image</span> <span class="rm-lmri-12">processing</span>, 11(3):188–200, 2002. </p> <p class="bibitem" ><span class="biblabel"> <a id="XKerr2005"></a><span class="bibsp">   </span></span>J N Kerr, D Greenberg, and F Helmchen. Imaging input and output of neocortical networks in vivo. <span class="rm-lmri-12">Proc Natl Acad Sci U S A</span>, 102(39): 14063–14068, 2005. ISSN 0027-8424 (Print) 0027-8424. doi: 10.1073/pnas. 0506029102. </p> <p class="bibitem" ><span class="biblabel"> <a id="XKuchibhotla2014"></a><span class="bibsp">   </span></span>K V Kuchibhotla, S Wegmann, K J Kopeikina, J Hawkes, N Rudinskiy, M L Andermann, T L Spires-Jones, B J Bacskai, and B T Hyman. Neurofibrillary tangle-bearing neurons are functionally integrated in cortical circuits in vivo. <span class="rm-lmri-12">Proc Natl Acad Sci U S</span> <span class="rm-lmri-12">A</span>, 111(1):510–514, 2014. ISSN 0027-8424. doi: 10.1073/pnas.1318807111. </p> <p class="bibitem" ><span class="biblabel"> <a id="XPeron2015"></a><span class="bibsp">   </span></span>Simon P. Peron, Jeremy Freeman, Vijay Iyer, Caiying Guo, and Karel Svoboda. A Cellular Resolution Map of Barrel Cortex Activity during Tactile Behavior. <span class="rm-lmri-12">Neuron</span>, 86(3): 783–799, 2015. ISSN 10974199. doi: 10.1016/j.neuron.2015.03.027. URL <a href="http://dx.doi.org/10.1016/j.neuron.2015.03.027" class="url" ><span class="rm-lmtt-12">http://dx.doi.org/10.1016/j.neuron.2015.03.027</span></a>. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcarpenter2006cellprofiler"></a><span class="bibsp">   </span></span>Anne E Carpenter, Thouis R Jones, Michael R Lamprecht, Colin Clarke, In Han Kang, Ola Friman, David A Guertin, Joo Han Chang, Robert A Lindquist, Jason Moffat, et al. Cellprofiler: image analysis software for identifying and quantifying cell phenotypes. <span class="rm-lmri-12">Genome biology</span>, 7(10):1–11, 2006. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmcquin2018cellprofiler"></a><span class="bibsp">   </span></span>Claire McQuin, Allen Goodman, Vasiliy Chernyshev, Lee Kamentsky, Beth A Cimini, Kyle W Karhohs, Minh Doan, Liya Ding, Susanne M Rafelski, Derek Thirstrup, et al. Cellprofiler 3.0: Next-generation image processing for biology. <span class="rm-lmri-12">PLoS biology</span>, 16(7):e2005970, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xlamprecht2007cellprofiler"></a><span class="bibsp">   </span></span>Michael R Lamprecht, David M Sabatini, and Anne E Carpenter. Cellprofiler™: free, versatile software for automated biological image analysis. <span class="rm-lmri-12">Biotechniques</span>, 42(1):71–75, 2007. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmukamel2009automated"></a><span class="bibsp">   </span></span>Eran A Mukamel, Axel Nimmerjahn, and Mark J Schnitzer. Automated analysis of cellular signals from large-scale calcium imaging data. <span class="rm-lmri-12">Neuron</span>, 63(6):747–760, 2009. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xavants2009advanced"></a><span class="bibsp">   </span></span>Brian B Avants, Nick Tustison, Gang Song, et al. Advanced normalization tools (ants). <span class="rm-lmri-12">Insight j</span>, 2(365):1–35, 2009. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xklein2009elastix"></a><span class="bibsp">   </span></span>Stefan Klein, Marius Staring, Keelin Murphy, Max A Viergever, and

Josien PW Pluim. Elastix: a toolbox for intensity-based medical image registration. <span class="rm-lmri-12">IEEE transactions on medical imaging</span>, 29(1):196–205, 2009.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xgreenberg2009automated"></a><span class="bibsp">   </span></span>David S Greenberg and Jason ND Kerr. Automated correction of fast motion artifacts for two-photon imaging of awake animals. <span class="rm-lmri-12">Journal of</span> <span class="rm-lmri-12">neuroscience methods</span>, 176(1):1–15, 2009. </p> <p class="bibitem" ><span class="biblabel"> <a id="XMiri2011"></a><span class="bibsp">   </span></span>A Miri, K Daie, R D Burdine, E Aksay, and D W Tank. Regression-based identification of behavior-encoding neurons during large-scale optical imaging of neural activity at cellular resolution. <span class="rm-lmri-12">J Neurophysiol</span>, 105(2):964–980, 2011. ISSN 1522-1598 (Electronic) 0022-3077 (Linking). doi: 10.1152/jn.00702.2010. URL <a href="http://www.ncbi.nlm.nih.gov/pubmed/21084686" class="url" ><span class="rm-lmtt-12">http://www.ncbi.nlm.nih.gov/pubmed/21084686</span></a>. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xbauch2011openbis"></a><span class="bibsp">   </span></span>Angela Bauch, Izabela Adamczyk, Piotr Buczek, Franz-Josef Elmer, Kaloyan Enimanev, Pawel Glyzewski, Manuel Kohler, Tomasz Pylak, Andreas Quandt, Chandrasekhar Ramakrishnan, et al. openbis: a flexible framework for managing and analyzing complex data in biology research. <span class="rm-lmri-12">BMC bioinformatics</span>, 12(1):1–19, 2011. </p> <p class="bibitem" ><span class="biblabel"> <a id="XFrancis2012"></a><span class="bibsp">   </span></span>M Francis, X Qian, C Charbel, J Ledoux, J C Parker, and M S Taylor. Automated region of interest analysis of dynamic Ca(2)+ signals in image sequences. <span class="rm-lmri-12">Am J Physiol Cell Physiol</span>, 303(3):C236–43, 2012. ISSN 0363-6143. doi: 10.1152/ajpcell.00016.2012. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xallan2012omero"></a><span class="bibsp">   </span></span>Chris Allan, Jean-Marie Burel, Josh Moore, Colin Blackburn, Melissa Linkert, Scott Loynton, Donald MacDonald, William J Moore, Carlos Neves, Andrew Patterson, et al. Omero: flexible, model-driven data management for experimental biology. <span class="rm-lmri-12">Nature methods</span>, 9(3):245–253, 2012. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xdiego2013automated"></a><span class="bibsp">   </span></span>Ferran Diego, Susanne Reichinnek, Martin Both, and Fred A Hamprecht.

Automated identification of neuronal activity from calcium imaging by sparse dictionary learning. In <span class="rm-lmri-12">Biomedical Imaging (ISBI), 2013 IEEE 10th</span> <span class="rm-lmri-12">International Symposium on</span>, pages 1058–1061. IEEE, 2013.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xtomek2013two"></a><span class="bibsp">   </span></span>Jakub Tomek, Ondrej Novak, and Josef Syka. Two-photon processor and seneca: a freely available software package to process data from two-photon calcium imaging at speeds down to several milliseconds per frame. <span class="rm-lmri-12">Journal</span> <span class="rm-lmri-12">of neurophysiology</span>, 110(1):243–256, 2013. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xpnevmatikakis2014structured"></a><span class="bibsp">   </span></span>Eftychios A Pnevmatikakis, Yuanjun Gao, Daniel Soudry, David Pfau, Clay Lacefield, Kira Poskanzer, Randy Bruno, Rafael Yuste, and Liam Paninski. A structured matrix factorization framework for large scale calcium imaging data analysis. <span class="rm-lmri-12">arXiv preprint arXiv:1409.2903</span>, 2014. </p> <p class="bibitem" ><span class="biblabel"> <a id="XMaruyama2014"></a><span class="bibsp">   </span></span>Ryuichi Maruyama, Kazuma Maeda, Hajime Moroda, Ichiro Kato, Masashi Inoue, Hiroyoshi Miyakawa, and Toru Aonishi. Detecting cells using non-negative matrix factorization on calcium imaging data. <span class="rm-lmri-12">Neural</span> <span class="rm-lmri-12">Netw</span>, 55:11–19, mar 2014. ISSN 0893-6080. doi: 10.1016/j.neunet.2014. 03.007. URL <a href="http://www.ncbi.nlm.nih.gov/pubmed/24705544" class="url" ><span class="rm-lmtt-12">http://www.ncbi.nlm.nih.gov/pubmed/24705544</span></a>. </p> <p class="bibitem" ><span class="biblabel"> <a id="XKaifosh2014"></a><span class="bibsp">   </span></span>Patrick Kaifosh, Jeffrey D Zaremba, Nathan B Danielson, and Attila Losonczy. SIMA: Python software for analysis of dynamic fluorescence imaging data. <span class="rm-lmri-12">Frontiers in neuroinformatics</span>, 8:80, 2014. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xyatsenko2015datajoint"></a><span class="bibsp">   </span></span>Dimitri Yatsenko, Jacob Reimer, Alexander S Ecker, Edgar Y Walker, Fabian Sinz, Philipp Berens, Andreas Hoenselaar, R James Cotton, Athanassios S Siapas, and Andreas S Tolias. Datajoint: managing big scientific data using matlab or python. <span class="rm-lmri-12">BioRxiv</span>, page 031658, 2015. </p> <p class="bibitem" ><span class="biblabel"> <a id="XTeeters2015"></a><span class="bibsp">   </span></span>Jeffery L Teeters, Keith Godfrey, Rob Young, Chinh Dang, Claudia Friedsam, Barry Wark, Hiroki Asari, Simon Peron, Nuo Li, and Adrien

Peyrache. Neurodata without borders: creating a common data format for neurophysiology. <span class="rm-lmri-12">Neuron</span>, 88(4):629–634, 2015.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xpachitariu2016suite2p"></a><span class="bibsp">   </span></span>Marius Pachitariu, Carsen Stringer, Sylvia Schröder, Mario Dipoppa, L Federico Rossi, Matteo Carandini, and Kenneth D Harris. Suite2p: beyond 10,000 neurons with standard two-photon microscopy. <span class="rm-lmri-12">Biorxiv</span>, page 061507, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xpnevmatikakis2016simultaneous"></a><span class="bibsp">   </span></span>Eftychios A Pnevmatikakis, Daniel Soudry, Yuanjun Gao, Timothy A Machado, Josh Merel, David Pfau, Thomas Reardon, Yu Mu, Clay Lacefield, Weijian Yang, et al. Simultaneous denoising, deconvolution, and demixing of calcium imaging data. <span class="rm-lmri-12">Neuron</span>, 89(2):285–299, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xzhou2016efficient"></a><span class="bibsp">   </span></span>P Zhou, SL Resendez, GD Stuber, RE Kass, and L Paninski. Efficient and accurate extraction of in vivo calcium signals from microendoscope video data. <span class="rm-lmri-12">arXiv preprint arXiv:1605.07266</span>, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xzhou2018efficient"></a><span class="bibsp">   </span></span>Pengcheng Zhou, Shanna L Resendez, Jose Rodriguez-Romaguera, Jessica C Jimenez, Shay Q Neufeld, Andrea Giovannucci, Johannes Friedrich, Eftychios A Pnevmatikakis, Garret D Stuber, Rene Hen, et al. Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data. <span class="rm-lmri-12">ELife</span>, 7:e28728, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xapthorpe2016automatic"></a><span class="bibsp">   </span></span>Noah Apthorpe, Alexander Riordan, Robert Aguilar, Jan Homann, Yi Gu, David Tank, and H Sebastian Seung. Automatic neuron detection in calcium imaging data using convolutional networks. In <span class="rm-lmri-12">Advances in</span> <span class="rm-lmri-12">Neural Information Processing Systems</span>, pages 3270–3278, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xdubbs2016moco"></a><span class="bibsp">   </span></span>Alexander Dubbs, James Guevara, and Rafael Yuste. moco: Fast motion correction for calcium imaging. <span class="rm-lmri-12">Frontiers in neuroinformatics</span>, 10:6, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmaree2016collaborative"></a><span class="bibsp">   </span></span>Raphaël Marée, Loïc Rollus, Benjamin Stévens, Renaud Hoyoux, Gilles Louppe, Rémy Vandaele, Jean-Michel Begon, Philipp Kainz, Pierre Geurts, and Louis Wehenkel. Collaborative analysis of multi-gigapixel imaging data using cytomine. <span class="rm-lmri-12">Bioinformatics</span>, 32(9):1395–1401, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmohammed2016integrative"></a><span class="bibsp">   </span></span>Ali I Mohammed, Howard J Gritton, Hua-an Tseng, Mark E Bucklin, Zhaojie Yao, and Xue Han. An integrative approach for analyzing hundreds of neurons in task performing mice using wide-field calcium imaging. <span class="rm-lmri-12">Scientific reports</span>, 6(1):20986, 2016. </p> <p class="bibitem" ><span class="biblabel"> <a id="XAhanonu2018sfnposter"></a><span class="bibsp">   </span></span>B. Ahanonu, L. J. Kitch, T. H. Kim, M. C. Larkin, E. O. Hamel, J. Lecoq, D. E. Aldarondo, and M. J. Schnitzer. Maximum likelihood and machine learning based methods for automated cell sorting of large-scale neural calcium imaging data. Society for Neuroscience, 2018. URL <a href="https://abstractsonline.com/pp8/#!/4649/presentation/41917" class="url" ><span class="rm-lmtt-12">https://abstractsonline.com/pp8/#!/4649/presentation/41917</span></a>. </p> <p class="bibitem" ><span class="biblabel"> <a id="XAhanonu2017sfnposter"></a><span class="bibsp">   </span></span>B. Ahanonu, L. J. Kitch, T. H. Kim, M. C. Larkin, E. O. Hamel, J. Lecoq, and M. J. Schnitzer. Maximum likelihood based cell sorting of large-scale neural calcium imaging data. Society for Neuroscience, 2017. URL <a href="http://www.abstractsonline.com/pp8/index.html#!/4376/presentation/18520" class="url" ><span class="rm-lmtt-12">http://www.abstractsonline.com/pp8/index.html#!/4376/presentation/18520</span></a>. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xahanonu2018neural"></a><span class="bibsp">   </span></span>Biafra Owowonta Ahanonu. <span class="rm-lmri-12">Neural Ensemble Dynamics in Behaving</span> <span class="rm-lmri-12">Animals: Computational Approaches and Applications in Amygdala and</span> <span class="rm-lmri-12">Striatum</span>. Stanford University, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xlu2017seeds"></a><span class="bibsp">   </span></span>Jinghao Lu, Chunyuan Li, and Fan Wang. Seeds cleansing cnmf for spatiotemporal neural signals extraction of miniscope imaging data. <span class="rm-lmri-12">arXiv</span> <span class="rm-lmri-12">preprint arXiv:1704.00793</span>, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xfriedrich2017fast"></a><span class="bibsp">   </span></span>Johannes Friedrich, Pengcheng Zhou, and Liam Paninski. Fast online deconvolution of calcium imaging data. <span class="rm-lmri-12">PLoS computational biology</span>, 13 (3):e1005423, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xreynolds2017able"></a><span class="bibsp">   </span></span>Stephanie Reynolds, Therese Abrahamsson, Renaud Schuck, P Jesper Sjöström, Simon R Schultz, and Pier Luigi Dragotti. Able: An activity-based level set segmentation algorithm for two-photon calcium imaging data. <span class="rm-lmri-12">eNeuro</span>, pages ENEURO–0012, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xpetersen2017scalpel"></a><span class="bibsp">   </span></span>Ashley Petersen, Noah Simon, and Daniela Witten. SCALPEL: Extracting Neurons from Calcium Imaging Data. <span class="rm-lmri-12">ArXiv e-prints</span>, art. arXiv:1703.06946, March 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xspaen2017hnccorr"></a><span class="bibsp">   </span></span>Quico Spaen, Dorit S Hochbaum, and Roberto Asín-Achá. Hnccorr: A novel combinatorial approach for cell identification in calcium-imaging movies. <span class="rm-lmri-12">arXiv preprint arXiv:1703.01999</span>, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xgiovannucci2017onacid"></a><span class="bibsp">   </span></span>Andrea Giovannucci, Johannes Friedrich, Matt Kaufman, Anne Churchland, Dmitri Chklovskii, Liam Paninski, and Eftychios A Pnevmatikakis. Onacid: Online analysis of calcium imaging data in real time. In <span class="rm-lmri-12">Advances in Neural Information Processing Systems</span>, pages 2381–2391, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xinan2017robust"></a><span class="bibsp">   </span></span>Hakan Inan, Murat A Erdogdu, and Mark Schnitzer. Robust estimation of neural signals in calcium imaging. In <span class="rm-lmri-12">Advances in Neural Information</span> <span class="rm-lmri-12">Processing Systems</span>, pages 2901–2910, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xorlandinetcal"></a><span class="bibsp">   </span></span>JG Orlandi, S Fernández-García, A Comella-Bolla, M Masana, G García-Díaz Barriga, M Yaghoubi, A Kipp, JM Canals, MA Colicos, J Davidsen,

et al. Netcal: An interactive platform for large-scale, network and population dynamics analysis of calcium imaging recordings, zenodo (2017).

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xpnevmatikakis2017normcorre"></a><span class="bibsp">   </span></span>Eftychios A Pnevmatikakis and Andrea Giovannucci. Normcorre: An online algorithm for piecewise rigid motion correction of calcium imaging data. <span class="rm-lmri-12">Journal of neuroscience methods</span>, 291:83–94, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xsheintuch2017tracking"></a><span class="bibsp">   </span></span>Liron Sheintuch, Alon Rubin, Noa Brande-Eilat, Nitzan Geva, Noa Sadeh, Or Pinchasof, and Yaniv Ziv. Tracking the same neurons across multiple days in ca2+ imaging data. <span class="rm-lmri-12">Cell reports</span>, 21(4):1102–1115, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xguan2018neuroseg"></a><span class="bibsp">   </span></span>Jiangheng Guan, Jingcheng Li, Shanshan Liang, Ruijie Li, Xingyi Li, Xiaozhe Shi, Ciyu Huang, Jianxiong Zhang, Junxia Pan, Hongbo Jia, et al. Neuroseg: automated cell detection and segmentation for in vivo two-photon ca 2+ imaging data. <span class="rm-lmri-12">Brain Structure and Function</span>, 223(1): 519–533, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xtakekawa2017automatic"></a><span class="bibsp">   </span></span>Takashi Takekawa, Hirotaka Asai, Noriaki Ohkawa, Masanori Nomoto, Reiko Okubo-Suzuki, Khaled Ghandour, Masaaki Sato, Yasunori Hayashi, Kaoru Inokuchi, and Tomoki Fukai. Automatic sorting system for large calcium imaging data. <span class="rm-lmri-12">bioRxiv</span>, page 215145, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xromano2017integrated"></a><span class="bibsp">   </span></span>Sebastián A Romano, Verónica Pérez-Schuster, Adrien Jouary, Jonathan Boulanger-Weill, Alessia Candeo, Thomas Pietri, and Germán Sumbre. An integrated calcium imaging processing toolbox for the analysis of neuronal population dynamics. <span class="rm-lmri-12">PLoS computational biology</span>, 13(6): e1005526, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xrueckl2017samuroi"></a><span class="bibsp">   </span></span>Martin Rueckl, Stephen C Lenzi, Laura Moreno-Velasquez, Daniel Parthier, Dietmar Schmitz, Sten Ruediger, and Friedrich W Johenning.

Samuroi, a python-based software tool for visualization and analysis of dynamic time series imaging at multiple spatial scales. <span class="rm-lmri-12">Frontiers in</span> <span class="rm-lmri-12">neuroinformatics</span>, 11:44, 2017.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xfillbrunn2017knime"></a><span class="bibsp">   </span></span>Alexander Fillbrunn, Christian Dietz, Julianus Pfeuffer, René Rahn, Gregory A Landrum, and Michael R Berthold. Knime for reproducible cross-domain analysis of life science data. <span class="rm-lmri-12">Journal of biotechnology</span>, 261: 149–156, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xklibisz2017fast"></a><span class="bibsp">   </span></span>Aleksander Klibisz, Derek Rose, Matthew Eicholtz, Jay Blundon, and Stanislav Zakharenko. Fast, simple calcium imaging segmentation with fully convolutional networks. In <span class="rm-lmri-12">International Workshop on Deep Learning</span> <span class="rm-lmri-12">in Medical Image Analysis</span>, pages 285–293. Springer, 2017. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xkeemink2018fissa"></a><span class="bibsp">   </span></span>Sander W Keemink, Scott C Lowe, Janelle MP Pakan, Evelyn Dylda, Mark CW Van Rossum, and Nathalie L Rochefort. Fissa: A neuropil decontamination toolbox for calcium imaging signals. <span class="rm-lmri-12">Scientific reports</span>, 8 (1):1–12, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmishne2018automated"></a><span class="bibsp">   </span></span>Gal Mishne, Ronald R Coifman, Maria Lavzin, and Jackie Schiller. Automated cellular structure extraction in biological images with applications to calcium imaging data. <span class="rm-lmri-12">bioRxiv</span>, page 313981, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xbuchanan2018penalized"></a><span class="bibsp">   </span></span>E Kelly Buchanan, Ian Kinsella, Ding Zhou, Rong Zhu, Pengcheng Zhou, Felipe Gerhard, John Ferrante, Ying Ma, Sharon Kim, Mohammed Shaik, et al. Penalized matrix decomposition for denoising, compression, and improved demixing of functional imaging data. <span class="rm-lmri-12">bioRxiv</span>, page 334706, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="XLu2018min1pipe"></a><span class="bibsp">   </span></span>Jinghao Lu, Chunyuan Li, Jonnathan Singh-Alvarado, Zhe Charles Zhou, Flavio Fröhlich, Richard Mooney, and Fan Wang. MIN1PIPE: A

Miniscope 1-Photon-Based Calcium Imaging Signal Extraction Pipeline. <span class="rm-lmri-12">Cell Reports</span>, 23(12):3673–3684, 2018. ISSN 2211-1247.

</p> <p class="bibitem" ><span class="biblabel"> <a id="XGiovannucci2018"></a><span class="bibsp">   </span></span>Andrea Giovannucci, Johannes Friedrich, Pat Gunn, Jeremie Kalfon, Sue Ann Koay, Jiannis Taxidis, Farzaneh Najafi, Jeffrey L Gauthier, Pengcheng Zhou, and David W Tank. CaImAn: An open source tool for scalable Calcium Imaging data Analysis. <span class="rm-lmri-12">bioRxiv</span>, page 339564, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xgauthier2018detecting"></a><span class="bibsp">   </span></span>Jeffrey L Gauthier, Sue Ann Koay, Edward H Nieh, David W Tank, Jonathan W Pillow, and Adam S Charles. Detecting and correcting false transients in calcium imaging. <span class="rm-lmri-12">bioRxiv</span>, page 473470, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xshen2018automatic"></a><span class="bibsp">   </span></span>Simon P Shen, Hua-an Tseng, Kyle R Hansen, Ruofan Wu, Howard J Gritton, Jennie Si, and Xue Han. Automatic cell segmentation by adaptive thresholding (acsat) for large-scale calcium imaging datasets. <span class="rm-lmri-12">eneuro</span>, 5(5), 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmitani2018real"></a><span class="bibsp">   </span></span>Akinori Mitani and Takaki Komiyama. Real-time processing of two-photon calcium imaging data including lateral motion artifact correction. <span class="rm-lmri-12">Frontiers in neuroinformatics</span>, 12:98, 2018. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcorder2019amygdalar"></a><span class="bibsp">   </span></span>Gregory Corder, Biafra Ahanonu, Benjamin F Grewe, Dong Wang, Mark J Schnitzer, and Grégory Scherrer. An amygdalar neural ensemble that encodes the unpleasantness of pain. <span class="rm-lmri-12">Science</span>, 363(6424):276–281, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xahanonu2022recording"></a><span class="bibsp">   </span></span>Biafra Ahanonu and Gregory Corder. Recording pain-related brain activity in behaving animals using calcium imaging calcium imaging and miniature microscopes. In <span class="rm-lmri-12">Contemporary Approaches to the Study of Pain:</span> <span class="rm-lmri-12">From Molecules to Neural Networks</span>, pages 217–276. Springer, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcharles2019neural"></a><span class="bibsp">   </span></span>Adam S Charles, Alex Song, Jeffrey L Gauthier, Jonathan W Pillow, and David W Tank. Neural anatomy and optical microscopy (naomi) simulation for evaluating calcium imaging methods. <span class="rm-lmri-12">bioRxiv</span>, page 726174, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xradstake2019calima"></a><span class="bibsp">   </span></span>FDW Radstake, EAL Raaijmakers, R Luttge, Svitlana Zinger, and Jean-Philippe Frimat. Calima: The semi-automated open-source calcium imaging analyzer. <span class="rm-lmri-12">Computer methods and programs in biomedicine</span>, 179: 104991, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xsoltanian2019fast"></a><span class="bibsp">   </span></span>Somayyeh Soltanian-Zadeh, Kaan Sahingur, Sarah Blau, Yiyang Gong, and Sina Farsiu. Fast and robust active neuron segmentation in two-photon calcium imaging using spatiotemporal deep learning. <span class="rm-lmri-12">Proceedings of the</span> <span class="rm-lmri-12">National Academy of Sciences</span>, 116(17):8554–8563, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xwang2019accurate"></a><span class="bibsp">   </span></span>Yizhi Wang, Nicole V DelRosso, Trisha V Vaidyanathan, Michelle K Cahill, Michael E Reitman, Silvia Pittolo, Xuelong Mi, Guoqiang Yu, and Kira E Poskanzer. Accurate quantification of astrocyte and neurotransmitter fluorescence dynamics for single-cell and population-level physiology. <span class="rm-lmri-12">Nature Neuroscience</span>, 22(11):1936–1944, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xgiovannucci2019caiman"></a><span class="bibsp">   </span></span>Andrea Giovannucci, Johannes Friedrich, Pat Gunn, Jeremie Kalfon, Brandon L Brown, Sue Ann Koay, Jiannis Taxidis, Farzaneh Najafi, Jeffrey L Gauthier, Pengcheng Zhou, et al. Caiman an open source tool for scalable calcium imaging data analysis. <span class="rm-lmri-12">Elife</span>, 8:e38173, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xmishne2019learning"></a><span class="bibsp">   </span></span>Gal Mishne and Adam S Charles. Learning spatially-correlated temporal dictionaries for calcium imaging. In <span class="rm-lmri-12">ICASSP 2019-2019 IEEE</span> <span class="rm-lmri-12">International Conference on Acoustics, Speech and Signal Processing</span> <span class="rm-lmri-12">(ICASSP)</span>, pages 1065–1069. IEEE, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xdolev2019segment2p"></a><span class="bibsp">   </span></span>Noah Dolev, Lior Pinkus, and Michal Rivlin-Etzion. Segment2p: Parameter-free automated segmentation of cellular fluorescent signals. <span class="rm-lmri-12">BioRxiv</span>, page 832188, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xchen2019lanmc"></a><span class="bibsp">   </span></span>Zhe Chen, Hugh T Blair, and Jason Cong. Lanmc: Lstm-assisted non-rigid motion correction on fpga for calcium image stabilization. In <span class="rm-lmri-12">Proceedings of the 2019 ACM/SIGDA International Symposium on</span> <span class="rm-lmri-12">Field-Programmable Gate Arrays</span>, pages 104–109, 2019. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xshibue2020deconvolution"></a><span class="bibsp">   </span></span>Ryohei Shibue and Fumiyasu Komaki. Deconvolution of calcium imaging data using marked point processes. <span class="rm-lmri-12">PLoS computational biology</span>, 16(3): e1007650, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xsaxena2020localized"></a><span class="bibsp">   </span></span>Shreya Saxena, Ian Kinsella, Simon Musall, Sharon H Kim, Jozsef Meszaros, David N Thibodeaux, Carla Kim, John Cunningham, Elizabeth MC Hillman, Anne Churchland, et al. Localized semi-nonnegative matrix factorization (locanmf) of widefield calcium imaging data. <span class="rm-lmri-12">PLOS Computational Biology</span>, 16(4):e1007791, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcantu2020ezcalcium"></a><span class="bibsp">   </span></span>Daniel A Cantu, Bo Wang, Michael W Gongwer, Cynthia X He, Anubhuti Goel, Anand Suresh, Nazim Kourdougli, Erica D Arroyo, William Zeiger, and Carlos Portera-Cailliau. Ezcalcium: Open source toolbox for analysis of calcium imaging data. <span class="rm-lmri-12">bioRxiv</span>, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xfriedrich2020online"></a><span class="bibsp">   </span></span>Johannes Friedrich, Andrea Giovannucci, and Eftychios A Pnevmatikakis. Online analysis of microendoscopic 1-photon calcium imaging data streams. <span class="rm-lmri-12">bioRxiv</span>, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xtran2020automated"></a><span class="bibsp">   </span></span>Lina M Tran, Andrew J Mocle, Adam I Ramsaran, Alex D Jacob,

Paul W Frankland, and Sheena A Josselyn. Automated curation of cnmf-e-extracted roi spatial footprints and calcium traces using open-source automl tools. <span class="rm-lmri-12">bioRxiv</span>, 2020a.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xtran2020automated2"></a><span class="bibsp">   </span></span>Lina M Tran, Andrew J Mocle, Adam I Ramsaran, Alexander D Jacob, Paul W Frankland, and Sheena A Josselyn. Automated curation of cnmf-e-extracted roi spatial footprints and calcium traces using open-source automl tools. <span class="rm-lmri-12">Frontiers in Neural Circuits</span>, 14:42, 2020b. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xlecoq2020removing"></a><span class="bibsp">   </span></span>Jerome Lecoq, Michael Oliver, Joshua H Siegle, Natalia Orlova, and Christof Koch. Removing independent noise in systems neuroscience data using deepinterpolation. <span class="rm-lmri-12">bioRxiv</span>, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xrubens2020biaflows"></a><span class="bibsp">   </span></span>Ulysse Rubens, Romain Mormont, Lassi Paavolainen, Volker Bäcker, Benjamin Pavie, Leandro A Scholz, Gino Michiels, Martin Maška, Devrim Ünay, Graeme Ball, et al. Biaflows: A collaborative framework to reproducibly deploy and benchmark bioimage analysis workflows. <span class="rm-lmri-12">Patterns</span>, 1(3):100040, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcassidy2020frequency"></a><span class="bibsp">   </span></span>Ryan M Cassidy, Alexis G Bavencoffe, Elia R Lopez, Sai S Cheruvu, Edgar T Walters, Rosa A Uribe, Anne Marie Krachler, and Max A Odem. Frequency-independent biological signal identification (fibsi): A free program that simplifies intensive analysis of non-stationary time series data. <span class="rm-lmri-12">bioRxiv</span>, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xalles2021chronic"></a><span class="bibsp">   </span></span>Sascha RA Alles, Max A Odem, Van B Lu, Ryan M Cassidy, and Peter A Smith. Chronic bdnf simultaneously inhibits and unmasks superficial dorsal horn neuronal activity. <span class="rm-lmri-12">Scientific reports</span>, 11(1):1–14, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xkirschbaum2020disco"></a><span class="bibsp">   </span></span>Elke Kirschbaum, Alberto Bailoni, and Fred A Hamprecht. Disco: deep

learning, instance segmentation, and correlations for cell segmentation in calcium imaging. In <span class="rm-lmri-12">Medical Image Computing and Computer Assisted</span> <span class="rm-lmri-12">Intervention–MICCAI 2020: 23rd International Conference, Lima, Peru,</span> <span class="rm-lmri-12">October 4–8, 2020, Proceedings, Part V 23</span>, pages 151–162. Springer, 2020.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xdenis2020deepcinac"></a><span class="bibsp">   </span></span>Julien Denis, Robin F Dard, Eleonora Quiroli, Rosa Cossart, and Michel A Picardo. Deepcinac: a deep-learning-based python toolbox for inferring calcium imaging neuronal activity based on movie visualization. <span class="rm-lmri-12">eneuro</span>, 7(4), 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xlee2020real"></a><span class="bibsp">   </span></span>Yaesop Lee, Jing Xie, Eungjoo Lee, Srijesh Sudarsanan, Da-Ting Lin, Rong Chen, and Shuvra S Bhattacharyya. Real-time neuron detection and neural signal extraction platform for miniature calcium imaging. <span class="rm-lmri-12">Frontiers</span> <span class="rm-lmri-12">in Computational Neuroscience</span>, 14:43, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xtan2020deepbrainseg"></a><span class="bibsp">   </span></span>Chaozhen Tan, Yue Guan, Zhao Feng, Hong Ni, Zoutao Zhang, Zhiguang Wang, Xiangning Li, Jing Yuan, Hui Gong, Qingming Luo, et al. Deepbrainseg: Automated brain region segmentation for micro-optical images with a convolutional neural network. <span class="rm-lmri-12">Frontiers in neuroscience</span>, 14: 179, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xgriffiths2020real"></a><span class="bibsp">   </span></span>Victoria A Griffiths, Antoine M Valera, Joanna YN Lau, Hana Roš, Thomas J Younts, Bóris Marin, Chiara Baragli, Diccon Coyle, Geoffrey J Evans, George Konstantinou, et al. Real-time 3d movement correction for two-photon imaging in behaving animals. <span class="rm-lmri-12">Nature methods</span>, 17(7):741–748, 2020. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xstringer2021cellpose"></a><span class="bibsp">   </span></span>Carsen Stringer, Tim Wang, Michalis Michaelos, and Marius Pachitariu. Cellpose: a generalist algorithm for cellular segmentation. <span class="rm-lmri-12">Nature Methods</span>, 18(1):100–106, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xsong2021neural"></a><span class="bibsp">   </span></span>Alexander Song, Jeff L Gauthier, Jonathan W Pillow, David W Tank, and Adam S Charles. Neural anatomy and optical microscopy (naomi) simulation for evaluating calcium imaging methods. <span class="rm-lmri-12">Journal of</span> <span class="rm-lmri-12">Neuroscience Methods</span>, 358:109173, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xfriedrich2021online"></a><span class="bibsp">   </span></span>Johannes Friedrich, Andrea Giovannucci, and Eftychios A Pnevmatikakis. Online analysis of microendoscopic 1-photon calcium imaging data streams. <span class="rm-lmri-12">PLoS</span> <span class="rm-lmri-12">computational biology</span>, 17(1):e1008565, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xinan2021fast"></a><span class="bibsp">   </span></span>Hakan Inan, Claudia Schmuckermair, Tugce Tasci, Biafra Ahanonu, Oscar Hernandez, Jérôme Lecoq, Fatih Dinç, Mark J Wagner, Murat Erdogdu, and Mark J Schnitzer. Fast and statistically robust cell extraction from large-scale neural calcium imaging datasets. <span class="rm-lmri-12">bioRxiv</span>, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xdong2021minian"></a><span class="bibsp">   </span></span>Zhe Dong, William Mau, Yu Susie Feng, Zachary T Pennington, Lingxuan Chen, Yosif Zaki, Kanaka Rajan, Tristan Shuman, Daniel Aharoni, and Denise J Cai. Minian: An open-source miniscope analysis pipeline. <span class="rm-lmri-12">bioRxiv</span>, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xkolar2021mesmerize"></a><span class="bibsp">   </span></span>Kushal Kolar, Daniel Dondorp, Jordi Cornelis Zwiggelaar, Jørgen Høyer, and Marios Chatzigeorgiou. Mesmerize: a dynamically adaptable user-friendly analysis platform for 2d &#x0026; 3d calcium imaging data. <span class="rm-lmri-12">bioRxiv</span>, page 840488, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xlecoq2021removing"></a><span class="bibsp">   </span></span>Jérôme Lecoq, Michael Oliver, Joshua H Siegle, Natalia Orlova, Peter Ledochowitsch, and Christof Koch. Removing independent noise in systems neuroscience data using deepinterpolation. <span class="rm-lmri-12">Nature Methods</span>, pages 1–8, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xhan2021efficient"></a><span class="bibsp">   </span></span>Seungjae Han, Eun-Seo Cho, Inkyu Park, Kijung Shin, and Young-Gyu Yoon. Efficient neural network approximation of robust pca for automated analysis of calcium imaging data. In <span class="rm-lmri-12">International Conference on Medical</span> <span class="rm-lmri-12">Image Computing and Computer-Assisted Intervention</span>, pages 595–604. Springer, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xtippani2021capture"></a><span class="bibsp">   </span></span>Madhavi Tippani, Elizabeth A Pattie, Brittany A Davis, Claudia V Nguyen, Yanhong Wang, Srinidhi Rao Sripathy, Brady J Maher, Keri Martinowich, Andrew E Jaffe, and Stephanie Cerceo Page. Capture: Calcium peak toolbox for analysis of in vitro calcium imaging data. <span class="rm-lmri-12">bioRxiv</span>, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xrupprecht2021database"></a><span class="bibsp">   </span></span>Peter Rupprecht, Stefano Carta, Adrian Hoffmann, Mayumi Echizen, Antonin Blot, Alex C Kwan, Yang Dan, Sonja B Hofer, Kazuo Kitamura, Fritjof Helmchen, et al. A database and deep learning toolbox for noise-optimized, generalized spike inference from calcium imaging. <span class="rm-lmri-12">Nature</span> <span class="rm-lmri-12">Neuroscience</span>, 24(9):1324–1337, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcai2021volpy"></a><span class="bibsp">   </span></span>Changjia Cai, Johannes Friedrich, Amrita Singh, M Hossein Eybposh, Eftychios A Pnevmatikakis, Kaspar Podgorski, and Andrea Giovannucci. Volpy: automated and scalable analysis pipelines for voltage imaging datasets. <span class="rm-lmri-12">PLoS computational biology</span>, 17(4):e1008806, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xli2021reinforcing"></a><span class="bibsp">   </span></span>Xinyang Li, Guoxun Zhang, Jiamin Wu, Yuanlong Zhang, Zhifeng Zhao, Xing Lin, Hui Qiao, Hao Xie, Haoqian Wang, Lu Fang, et al. Reinforcing neuron extraction and spike inference in calcium imaging using deep self-supervised denoising. <span class="rm-lmri-12">Nature Methods</span>, pages 1–6, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xde2021specseg"></a><span class="bibsp">   </span></span>Leander de Kraker, Koen Seignette, Premnath Thamizharasu, Bastijn JG van den Boom, Ildefonso Ferreira Pica, Ingo Willuhn, Christiaan N Levelt, and

Chris van der Togt. Specseg: cross spectral power-based segmentation of neurons and neurites in chronic calcium imaging datasets. <span class="rm-lmri-12">bioRxiv</span>, pages 2020–10, 2021.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xgiovannucci2021fiola"></a><span class="bibsp">   </span></span>Andrea Giovannucci, Changjia Cai, Cynthia Dong, Marton Rozsa, and Eftychios Pnevmatikakis. Fiola: An accelerated pipeline for fluorescence imaging online analysis. 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="XHattori2021.11.10.468164"></a><span class="bibsp">   </span></span>Ryoma Hattori and Takaki Komiyama. Patchwarp: Corrections of non-uniform image distortions in two-photon calcium imaging data by patchwork affine transformations. <span class="rm-lmri-12">bioRxiv</span>, 2021. doi: 10.1101/2021.11.10.468164. URL <a href="https://www.biorxiv.org/content/early/2021/11/13/2021.11.10.468164" class="url" ><span class="rm-lmtt-12">https://www.biorxiv.org/content/early/2021/11/13/2021.11.10.468164</span></a>. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xzhang2021automated"></a><span class="bibsp">   </span></span>Xiaohui Zhang, Eric C Landsness, Wei Chen, Hanyang Miao, Michelle Tang, Lindsey M Brier, Joseph P Culver, Jin-Moo Lee, and Mark A Anastasio. Automated sleep state classification of wide-field calcium imaging data via multiplex visibility graphs and deep learning. <span class="rm-lmri-12">Journal of</span> <span class="rm-lmri-12">Neuroscience Methods</span>, page 109421, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xflotho2022software"></a><span class="bibsp">   </span></span>Philipp Flotho, Shinobu Nomura, Bernd Kuhn, and Daniel J Strauss. Software for non-parametric image registration of 2-photon imaging data. <span class="rm-lmri-12">Journal of Biophotonics</span>, 15(8):e202100330, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xbao2021segmentation"></a><span class="bibsp">   </span></span>Yijun Bao, Somayyeh Soltanian-Zadeh, Sina Farsiu, and Yiyang Gong. Segmentation of neurons from fluorescence calcium recordings beyond real time. <span class="rm-lmri-12">Nature machine intelligence</span>, 3(7):590–600, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xtaniguchi2021open"></a><span class="bibsp">   </span></span>Masaki Taniguchi, Taro Tezuka, Pablo Vergara, Sakthivel Srinivasan, Takuma Hosokawa, Yoan Chérasse, Toshie Naoi, Takeshi Sakurai, and Masanori Sakaguchi. Open-source software for real-time calcium

imaging and synchronized neuron firing detection. In <span class="rm-lmri-12">2021 43rd Annual</span> <span class="rm-lmri-12">International Conference of the IEEE Engineering in Medicine & Biology</span> <span class="rm-lmri-12">Society (EMBC)</span>, pages 2997–3003. IEEE, 2021.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xmullen2021automated"></a><span class="bibsp">   </span></span>Brian R Mullen, Sydney C Weiser, Desiderio Ascencio, and James B Ackman. Automated classification of signal sources in mesoscale calcium imaging. <span class="rm-lmri-12">bioRxiv</span>, pages 2021–02, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xwang2021image"></a><span class="bibsp">   </span></span>Yina Wang, Henry Pinkard, Emaad Khwaja, Shuqin Zhou, Laura Waller, and Bo Huang. Image denoising for fluorescence microscopy by self-supervised transfer learning. <span class="rm-lmri-12">bioRxiv</span>, pages 2021–02, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xlagache2021tracking"></a><span class="bibsp">   </span></span>Thibault Lagache, Alison Hanson, Jesús E Pérez-Ortega, Adrienne Fairhall, and Rafael Yuste. Tracking calcium dynamics from individual neurons in behaving animals. <span class="rm-lmri-12">PLOS Computational Biology</span>, 17(10): e1009432, 2021. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcharles2022graft"></a><span class="bibsp">   </span></span>Adam S Charles, Nathan Cermak, Rifqi O Affan, Benjamin B Scott, Jackie Schiller, and Gal Mishne. Graft: graph filtered temporal dictionary learning for functional neural imaging. <span class="rm-lmri-12">IEEE Transactions on Image</span> <span class="rm-lmri-12">Processing</span>, 31:3509–3524, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xtippani2022capture"></a><span class="bibsp">   </span></span>Madhavi Tippani, Elizabeth A Pattie, Brittany A Davis, Claudia V Nguyen, Yanhong Wang, Srinidhi Rao Sripathy, Brady J Maher, Keri Martinowich, Andrew E Jaffe, and Stephanie Cerceo Page. Capture: Calcium peaktoolbox for analysis of in vitro calcium imaging data. <span class="rm-lmri-12">BMC</span> <span class="rm-lmri-12">neuroscience</span>, 23(1):71, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xde2022specseg"></a><span class="bibsp">   </span></span>Leander de Kraker, Koen Seignette, Premnath Thamizharasu, Bastijn JG van den Boom, Ildefonso Ferreira Pica, Ingo Willuhn, Christiaan N Levelt, and Chris van der Togt. Specseg

is a versatile toolbox that segments neurons and neurites in chronic calcium imaging datasets based on low-frequency cross-spectral power. <span class="rm-lmri-12">Cell reports</span> <span class="rm-lmri-12">methods</span>, 2(10), 2022.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xsita2022deep"></a><span class="bibsp">   </span></span>Luca Sità, Marco Brondi, Pedro Lagomarsino de Leon Roig, Sebastiano Curreli, Mariangela Panniello, Dania Vecchia, and Tommaso Fellin. A deep-learning approach for online cell identification and trace extraction in functional two-photon calcium imaging. <span class="rm-lmri-12">Nature Communications</span>, 13(1): 1529, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xguan2022deep"></a><span class="bibsp">   </span></span>Honghua Guan, Dawei Li, Hyeon-cheol Park, Ang Li, Yuanlei Yue, Yung-Tian A Gau, Ming-Jun Li, Dwight E Bergles, Hui Lu, and Xingde Li. Deep-learning two-photon fiberscopy for video-rate brain imaging in freely-behaving mice. <span class="rm-lmri-12">Nature communications</span>, 13(1):1534, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xkamran2022new"></a><span class="bibsp">   </span></span>Sharif Amit Kamran, Khondker Fariha Hossain, Hussein Moghnieh, Sarah Riar, Allison Bartlett, Alireza Tavakkoli, Kenton M Sanders, and Salah A Baker. New open-source software for subcellular segmentation and analysis of spatiotemporal fluorescence signals using deep learning. <span class="rm-lmri-12">Iscience</span>, 25(5), 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xli2023real"></a><span class="bibsp">   </span></span>Xinyang Li, Yixin Li, Yiliang Zhou, Jiamin Wu, Zhifeng Zhao, Jiaqi Fan, Fei Deng, Zhaofa Wu, Guihua Xiao, Jing He, et al. Real-time denoising enables high-sensitivity fluorescence time-lapse imaging beyond the shot-noise limit. <span class="rm-lmri-12">Nature Biotechnology</span>, 41(2):282–292, 2023a. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xgauthier2022detecting"></a><span class="bibsp">   </span></span>Jeffrey L Gauthier, Sue Ann Koay, Edward H Nieh, David W Tank, Jonathan W Pillow, and Adam S Charles. Detecting and correcting false transients in calcium imaging. <span class="rm-lmri-12">Nature Methods</span>, 19(4):470–478, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xflores2022axial"></a><span class="bibsp">   </span></span>Andres Flores-Valle and Johannes D Seelig. Axial motion estimation

and correction for simultaneous multi-plane two-photon calcium imaging. <span class="rm-lmri-12">Biomedical Optics Express</span>, 13(4):2035–2049, 2022.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xliu2022fast"></a><span class="bibsp">   </span></span>Weiyi Liu, Junxia Pan, Yuanxu Xu, Meng Wang, Hongbo Jia, Kuan Zhang, Xiaowei Chen, Xingyi Li, and Xiang Liao. Fast and accurate motion correction for two-photon ca2+ imaging in behaving mice. <span class="rm-lmri-12">Frontiers in</span> <span class="rm-lmri-12">Neuroinformatics</span>, 16:851188, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xrubel2022neurodata"></a><span class="bibsp">   </span></span>Oliver Rübel, Andrew Tritt, Ryan Ly, Benjamin K Dichter, Satrajit Ghosh, Lawrence Niu, Pamela Baker, Ivan Soltesz, Lydia Ng, Karel Svoboda, et al. The neurodata without borders ecosystem for neurophysiological data science. <span class="rm-lmri-12">Elife</span>, 11:e78362, 2022. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xchen2023hardware"></a><span class="bibsp">   </span></span>Zhe Chen, Garrett J Blair, Changliang Guo, Jim Zhou, Juan-Luis Romero-Sosa, Alicia Izquierdo, Peyman Golshani, Jason Cong, Daniel Aharoni, and Hugh T Blair. A hardware system for real-time decoding of in vivo calcium imaging data. <span class="rm-lmri-12">Elife</span>, 12:e78344, 2023. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xzhang2023fast"></a><span class="bibsp">   </span></span>Yan Zhang, Márton Rózsa, Yajie Liang, Daniel Bushey, Ziqiang Wei, Jihong Zheng, Daniel Reep, Gerard Joey Broussard, Arthur Tsang, Getahun Tsegaye, et al. Fast and sensitive gcamp calcium indicators for imaging neural populations. <span class="rm-lmri-12">Nature</span>, 615(7954):884–891, 2023a. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xxu2023neuroseg"></a><span class="bibsp">   </span></span>Zhehao Xu, Yukun Wu, Jiangheng Guan, Shanshan Liang, Junxia Pan, Meng Wang, Qianshuo Hu, Hongbo Jia, Xiaowei Chen, and Xiang Liao. Neuroseg-ii: A deep learning approach for generalized neuron segmentation in two-photon ca2+ imaging. <span class="rm-lmri-12">Frontiers in Cellular Neuroscience</span>, 17: 1127847, 2023. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xvergara2023caliali"></a><span class="bibsp">   </span></span>Pablo Vergara, Yuteng Wang, Sakthivel Srinivasan, Yoan Cherasse, Toshie Naoi, Yuki Sugaya, Takeshi Sakurai, Masanobu Kano, and Masanori

Sakaguchi. The caliali tool for long-term tracking of neuronal population dynamics in calcium imaging. <span class="rm-lmri-12">bioRxiv</span>, pages 2023–05, 2023.

</p> <p class="bibitem" ><span class="biblabel"> <a id="Xzhang2023rapid"></a><span class="bibsp">   </span></span>Yuanlong Zhang, Guoxun Zhang, Xiaofei Han, Jiamin Wu, Ziwei Li, Xinyang Li, Guihua Xiao, Hao Xie, Lu Fang, and Qionghai Dai. Rapid detection of neurons in widefield calcium imaging datasets after training with synthetic data. <span class="rm-lmri-12">Nature Methods</span>, 20(5):747–754, 2023b. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xbonato2023astra"></a><span class="bibsp">   </span></span>Jacopo Bonato, Sebastiano Curreli, Sara Romanzi, Stefano Panzeri, and Tommaso Fellin. Astra: a deep learning algorithm for fast semantic segmentation of large-scale astrocytic networks. <span class="rm-lmri-12">bioRxiv</span>, pages 2023–05, 2023. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xli2023spatial"></a><span class="bibsp">   </span></span>Xinyang Li, Xiaowan Hu, Xingye Chen, Jiaqi Fan, Zhifeng Zhao, Jiamin Wu, Haoqian Wang, and Qionghai Dai. Spatial redundancy transformer for self-supervised fluorescence image denoising. <span class="rm-lmri-12">bioRxiv</span>, pages 2023–06, 2023b. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xcho2023robust"></a><span class="bibsp">   </span></span>Junmo Cho, Seungjae Han, Eun-Seo Cho, Kijung Shin, and Young-Gyu Yoon. Robust and efficient alignment of calcium imaging data through simultaneous low rank and sparse decomposition. In <span class="rm-lmri-12">Proceedings of the</span> <span class="rm-lmri-12">IEEE/CVF Winter Conference on Applications of Computer Vision</span>, pages 1939–1948, 2023. </p> <p class="bibitem" ><span class="biblabel"> <a id="Xahanonu2023long"></a><span class="bibsp">   </span></span>Biafra Ahanonu, Andrew Crowther, Artur Kania, Mariela Rosa Casillas, and Allan Basbaum. Long-term optical imaging of the spinal cord in awake, behaving animals. <span class="rm-lmri-12">bioRxiv</span>, pages 2023–05, 2023. </p> </div> <!--l. 91--><p class="indent" > <hr class='headingsep'> <hr class='headingsep'> <h3>Footnotes</h3> <a id="Q1-1-5"></a> </div> </p>