Home

Awesome

Zero-shot Nuclei Detection via Visual-Language Pre-trained Models

<img src="VLPMNuD.png" width="800">

Official implementation of Zero-shot Nuclei Detection via Visual-Language Pre-trained Models. The original paper link is here: arXiv link, [MICCAI link](to be update). The proposed method has two process steps:

  1. Generating raw GLIP prediction results.
  2. Self-training via YOLOX. This repository provides source code for the first step. Code for the second step is here.

Installation

  1. Create a Python environment based on the requirements.txt file.

  2. Build up GLIP. Our project is developed based on GLIP.

cd sick/GLIP
python setup.py develop

This instruction will install a maskrcnn_benchmark library, which is required by GLIP.

  1. [Optional] We recommend installing transformers library from source code, which enables us to change some function inside the transformer backbone.If you have installed Transformer via pip and only want to reproduce the experiment result in our paper, you can also skip this step and simply pip install transformers==4.27.4.The following option is intended for further development. You first need to uninstall transformers via pip or conda. Then:
cd github_src/transformers
python setup.py develop

Dataset Introduction

For the convenience of downloading and using, we directly placed the dataset inside the project. It is located in DATASET/coco. For a dataset of tens of MB, it is far more convenient to use git clone directly than to download it from a network disk. It was originally a MoNuSAC dataset and has been customized into COCO dataset format by us. This path will be directly called in the code, so it is not recommended to modify its path. Making it easy for others to reproduce initial results is the original intention of our team, sincerely.

Test: Generate the raw GLIP prediction result

Before testing, please download pre-trained weights of GLIP large model. Then:

python tools/test_grounding_net_wuap.py --config-file configs/pretrain/glip_Swin_L.yaml --weight glip_large_model.pth 
--prompt 'rectangular black nuclei. circle black nuclei. spherical black nuclei. rectangular dark purple nuclei. circle dark purple nuclei. spherical dark purple nuclei.' TEST.IMS_PER_BATCH 1 MODEL.DYHEAD.SCORE_AGG "MEAN" TEST.EVAL_TASK detection MODEL.DYHEAD.FUSE_CONFIG.MLM_LOSS False OUTPUT_DIR OUTPUT

Caution: --prompt param here gives an example prompt obtained from [blip](to be update), you can also input other prompts. A JSON file recording all bbox will be generated in ./jsonfiles.Normally, the mAP score of raw glip should be in the range [0.15, 0.2]. Then you can use the predicted result (i.e. the json file) to run a self-training algorithm, and you will get a near SOTA score (range [0.4, 0.45]). We provide our self-training source code here , which is YOLOX-based and used in our paper. You can choose to use YOLOX-based self-training method or the following "optional" self-training method.

[Optional] Self-Training GLIP using generated raw GLIP prediction

We provide an alternative way to get near SOTA score, in case that you feel YOLOX-based self-training process too complicate and labor-consuming.But you need to make sure each of your gpus be bigger than 16G. You can feed raw GLIP predict back to GLIP itself, train a new GLIP model with unchanged pre-trained weights(i.e. the glip_large_model.pth), and will get a better score, probably mAP in the range [0.35, 0.45]. You can repeat this process again and again. This optional method is actually GLIP-based self-training, rather than YOLOX-based self-training. It is equivalent to YOLOX-based self-training, and you may be happy to avoid building up a YOLOX project. We placed our example raw GLIP prediction file to the path:DATASET/coco/annotations/instances_train2017_0166.json, this result get mAP = 0.166. Example instructions are following:

python tools/train_net.py --config-file configs/pretrain/glip_Swin_L.yaml --train_label "DATASET/coco/annotations/instances_train2017_0166.json" --restart True --use-tensorboard 
--override_output_dir OUTPUT_TRAIN_fanew MODEL.BACKBONE.FREEZE_CONV_BODY_AT 1 SOLVER.IMS_PER_BATCH 1 SOLVER.USE_AMP True SOLVER.MAX_EPOCH 4 TEST.DURING_TRAINING True TEST.IMS_PER_BATCH 1 SOLVER.FIND_UNUSED_PARAMETERS False SOLVER.BASE_LR 0.00001 SOLVER.LANG_LR 0.00001 DATASETS.DISABLE_SHUFFLE True MODEL.DYHEAD.SCORE_AGG "MEAN" TEST.EVAL_TASK detection AUGMENT.MULT_MIN_SIZE_TRAIN (800,) SOLVER.CHECKPOINT_PERIOD 100

Other example instructions : (you must install the Transformers library from github_src as mentioned in Installation because we change some layers and functions in the transformer for better mAP score )

python tools/train_net.py --config-file configs/pretrain/glip_Swin_L.yaml --train_label "/data2/wyj/GLIP/DATASET/coco/annotations/instances_train2017_0166.json" --restart True --use-tensorboard --override_output_dir OUTPUT_TRAIN_fanew MODEL.BACKBONE.FREEZE_CONV_BODY_AT 1 SOLVER.IMS_PER_BATCH 1 SOLVER.USE_AMP True SOLVER.MAX_ITER 500 TEST.DURING_TRAINING True TEST.IMS_PER_BATCH 1 SOLVER.FIND_UNUSED_PARAMETERS False SOLVER.BASE_LR 0.00001 SOLVER.LANG_LR 0.00001 DATASETS.DISABLE_SHUFFLE True MODEL.DYHEAD.SCORE_AGG "MEAN" TEST.EVAL_TASK detection AUGMENT.MULT_MIN_SIZE_TRAIN (800,) SOLVER.CHECKPOINT_PERIOD 100 SWINBLO 3 lang_adap_mlp 2

Citation

If you use VLPM in your work or wish to refer to the results published in this repo, please cite our paper:

@inproceedings{wu2023zero,
  title={Zero-Shot Nuclei Detection via Visual-Language Pre-trained Models},
  author={Wu, Yongjian and Zhou, Yang and Saiyin, Jiya and Wei, Bingzheng and Lai, Maode and Shou, Jianzhong and Fan, Yubo and Xu, Yan},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={693--703},
  year={2023},
  organization={Springer}
}