MedSAM - GitLite

KO | EN

gitlite — search

#react #go #machine-learning #rust #linux #docker #java #windows #awesome #kubernetes #awesome-list #mcp #ios #openai

MedSAM

(★ 4,375)

Segment Anything in Medical Images

파일 탐색기

최종 버전 다운로드 (.zip)

- architecture.png
- box-task-illustration.png
- img_demo.png
- MedSAM_supp.pdf
- pathology.png
- seg_demo.gif
- seg_pathology.mp4
- task-illustration.png
- - infer_deeplabv3_res50_2D.py
  - infer_deeplabv3_res50_3D.py
  - README.md
  - train_deeplabv3_res50.py
- - infer_nnunet_2D.py
  - infer_nnunet_3D.py
  - README.md
- - infer_SAM_2D_npz.py
  - infer_SAM_3D_npz.py
  - README.md
- - point_seg_demo.gif
  - README.md
  - train_point_prompt.py
  - tutorial_point_prompt_seg.ipynb
- - label_interpolate.py
  - medsam_infer_3Dbox_adrenal.py
  - README.md
- - README.md
  - text_seg_demo.gif
  - train_text_prompt.py
  - tutorial_text_prompt_seg.ipynb
- - __init__.py
  - common.py
  - image_encoder.py
  - mask_decoder.py
  - prompt_encoder.py
  - sam.py
  - transformer.py
- - __init__.py
  - amg.py
  - onnx.py
  - transforms.py
- __init__.py
- automatic_mask_generator.py
- build_sam.py
- predictor.py
- ckpt_convert.py
- demo.py
- format_convert.py
- pre_CT_MR.py
- pre_grey_rgb.py
- README.md
- split.py
- SurfaceDice.py
- - README.md
.gitignore
.pre-commit-config.yaml
gui.py
LICENSE
MedSAM_Inference.py
pre_CT_MR.py
README.md
setup.py
train_multi_gpus.py
train_multi_gpus.sh
train_one_gpu.py
tutorial_quickstart.ipynb

# 설치 가이드

1. 코드 내려받기

git clone https://github.com/bowang-lab/MedSAM

깃허브에서 프로젝트 코드 전체를 내 컴퓨터로 내려받습니다.

cd MedSAM

방금 내려받은 프로젝트 폴더 안으로 이동합니다.

2. Python

쉬움 추천

사전 준비물

Git GitHub에서 프로젝트 코드를 내려받으려면 필요합니다.
Python 3 설치 시 'Add Python to PATH' 옵션을 꼭 체크하세요 (Windows).

pip install .

PyPI에 배포된 패키지를 바로 설치합니다. 소스 클론이 필요 없습니다.

jupyter notebook

브라우저에서 노트북(.ipynb) 파일들을 열람하고 실행할 수 있는 Jupyter 화면을 켭니다.

✅ 에러 메시지 없이 실행되고 터미널에 안내 문구가 출력되면 정상입니다.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# MedSAM
This is the official repository for MedSAM: Segment Anything in Medical Images.

Welcome to join our [mailing list](https://forms.gle/hk4Efp6uWnhjUHFP6) to get updates.

## News

- 2025.04.07: Release [MedSAM2](https://github.com/bowang-lab/MedSAM2) for 3D and video segmentation.
- 2025.02: Welcome to join CVPR 2025 Challenges: [Interactive](https://www.codabench.org/competitions/5263/) and [Text-guided](https://www.codabench.org/competitions/5651/) 3D Biomedical Image Segmentation
- 2024.01.15: Welcome to join [CVPR 2024 Challenge: MedSAM on Laptop](https://www.codabench.org/competitions/1847/)
- 2024.01.15: Release [LiteMedSAM](https://github.com/bowang-lab/MedSAM/blob/LiteMedSAM/README.md) and [3D Slicer Plugin](https://github.com/bowang-lab/MedSAMSlicer), 10x faster than MedSAM!

## Installation
1. Create a virtual environment `conda create -n medsam python=3.10 -y` and activate it `conda activate medsam`
2. Install [Pytorch 2.0](https://pytorch.org/get-started/locally/)
3. `git clone https://github.com/bowang-lab/MedSAM`
4. Enter the MedSAM folder `cd MedSAM` and run `pip install -e .`

## Get Started
Download the [model checkpoint](https://drive.google.com/drive/folders/1ETWmi4AiniJeWOt6HAsYgTjYv_fkgzoN?usp=drive_link) and place it at e.g., `work_dir/MedSAM/medsam_vit_b`

We provide three ways to quickly test the model on your images

1. Command line

```bash
python MedSAM_Inference.py # segment the demo image
```

Segment other images with the following flags
```bash
-i input_img
-o output path
--box bounding box of the segmentation target
```

2. Jupyter-notebook

We provide a step-by-step tutorial on [CoLab](https://colab.research.google.com/drive/19WNtRMbpsxeqimBlmJwtd1dzpaIvK2FZ?usp=sharing)

You can also run it locally with `tutorial_quickstart.ipynb`.

3. GUI

Install `PyQt5` with [pip](https://pypi.org/project/PyQt5/): `pip install PyQt5 ` or [conda](https://anaconda.org/anaconda/pyqt): `conda install -c anaconda pyqt`

```bash
python gui.py
```

Load the image to the GUI and specify segmentation targets by drawing bounding boxes.

https://github.com/bowang-lab/MedSAM/assets/19947331/a8d94b4d-0221-4d09-a43a-1251842487ee

## Model Training

### Data preprocessing

Download [SAM checkpoint](https://dl.fbaipublicfiles.com/segment_anything/sam_vit_b_01ec64.pth) and place it at `work_dir/SAM/sam_vit_b_01ec64.pth` .

Download the demo [dataset](https://zenodo.org/record/7860267) and unzip it to `data/FLARE22Train/`.

This dataset contains 50 abdomen CT scans and each scan contains an annotation mask with 13 organs. The names of the organ label are available at [MICCAI FLARE2022](https://flare22.grand-challenge.org/).

Run pre-processing

Install `cc3d`: `pip install connected-components-3d`

```bash
python pre_CT_MR.py
```

- split dataset: 80% for training and 20% for testing
- adjust CT scans to [soft tissue](https://radiopaedia.org/articles/windowing-ct) window level (40) and width (400)
- max-min normalization
- resample image size to `1024x1024`
- save the pre-processed images and labels as `npy` files

### Training on multiple GPUs (Recommend)

The model was trained on five A100 nodes and each node has four GPUs (80G) (20 A100 GPUs in total). Please use the slurm script to start the training process.

```bash
sbatch train_multi_gpus.sh
```

When the training process is done, please convert the checkpoint to SAM's format for convenient inference.

```bash
python utils/ckpt_convert.py # Please set the corresponding checkpoint path first
```

### Training on one GPU

```bash
python train_one_gpu.py
```

## Acknowledgements
- We highly appreciate all the challenge organizers and dataset owners for providing the public dataset to the community.
- We thank Meta AI for making the source code of [segment anything](https://github.com/facebookresearch/segment-anything) publicly available.
- We also thank Alexandre Bonnet for sharing this great [blog](https://encord.com/blog/learn-how-to-fine-tune-the-segment-anything-model-sam/)

## Reference

```
@article{MedSAM,
  title={Segment Anything in Medical Images},
  author={Ma, Jun and He, Yuting and Li, Feifei and Han, Lin and You, Chenyu and Wang, Bo},
  journal={Nature Communications},
  volume={15},
  pages={654},
  year={2024}
}
```