Awesome
MedSAM-Lite 3D Slicer Plugin
This is the official repository for 3D Slicer Plugin for MedSAM: Segment Anything in Medical Images.
https://github.com/bowang-lab/MedSAMSlicer/assets/19947331/c7ef20f1-7e23-4e6c-8be1-e39931eeb841
Installation
You can watch a video tutorial of installation steps here.
- Install 3D Slicer from its official website. The compatibility of our plugin has been tested with 3D Slicer >= 5.4.0
- Download the specific version of the plugin from releases page and extract to your desired location.
- In the Slicer App, press Ctrl+4 to open Extension Manager.
- In
Install Extension
tab lookupPyTorch​
and install the extension. Our plugin currently relies on this. Restart 3D Slicer afterward. - Select the
Welcome to Slicer
drop-down menu in the toolbar at the top and navigate toDeveloper Tools > Extension Wizard
. - Click on
select Extension
and locate theMedSAM
folder among the extracted files and directories at step 2. Accept if asked to add the new modules. - Now, from the
Welcome to Slicer
drop-down menu, under theSegmentation
sub-menu,MedSAMLite
option is added. By choosing it, you get to the final steps. Choose a folder
to install module dependencies and click onInstall dependencies
. It can take several minutes.- Restart 3D Slicer.
Upgrade
Remove all pre-existing files from step#2 and install the new version as instructed before.
Usage
You can watch a video guide for usage here.
- From the
Welcome to Slicer
drop-down menu, under theSegmentation
sub-menu,MedSAMLite
option is added. By choosing it, you get to the final steps. - Load your image file.
- From
Prepare Data
accordion menu you can choose from preset preprocessing methods (optional, but extremely helpful) - From
Select the Region of Interest
accordion menu, you can click onAttach ROI
to select the region for segmentation. It is possible to useSet Current Frame As Selection's Start
andSet Current Frame As Selection's End
to set ROI boundaries on the L-R slice. - In
Start Segmentation
accordion menu, you have two different ways to infer the segmentation:- If you want to segment multiple regions or a single large one, it is advised to preprocess the whole image first. To do so:
- click on
Send Image
button. It will prepare the whole 3D image for the module to process. It will take several minutes. You will need to do this step only once for each image. As long as the image is not changed (new preprocessing / cropping / etc.) you do not have to redo this step. - Click on
Segmentation
button to get the segmentation results. You can replace your ROI or delete and re-attach it (step 4) as many times as you need and repeat step 5.b without going over step 5.a again.
- click on
- If you need to segment a single smaller region, you can bypass preprocessing and only click on
Single Segmentation
. After confirming, the segmentation would be inferred.
- If you want to segment multiple regions or a single large one, it is advised to preprocess the whole image first. To do so:
- In the same section, from the
Segmentation
andSource volume
drop-down menus you can choose the right segmentation group and manually refine the segmentation results. You can also assign different color to each segmentation mask.