Skip to content

Commit 2143b79

Browse files
shiyutangshiyutang
authored
[Cherry-pick]Cherry-pick medicalseg docs to release/2.7 (#2783)
1 parent 113cbd8 commit 2143b79

11 files changed

Lines changed: 134 additions & 95 deletions

File tree

‎EISeg/med3d/README.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -121,7 +121,7 @@ os.system(f"'{sys.executable}' -m pip install paddlepaddle-gpu==2.3.1.post111 -f
121121

122122
| 数据 | 模型 | 下载链接 |
123123
|:-:|:-:|:-:|
124-
| MRI椎骨数据 | 交互式 Vnet |[模型](链接: https://pan.baidu.com/s/1vu0ZIbGumlFvRlMGbMvWAg)-pw: dt8q \| [椎骨数据](https://aistudio.baidu.com/aistudio/datasetdetail/81211)|
124+
| MRI椎骨数据 | 交互式 Vnet |[模型](链接: https://pan.baidu.com/s/1vu0ZIbGumlFvRlMGbMvWAg )-pw: dt8q \| [椎骨数据](https://aistudio.baidu.com/aistudio/datasetdetail/81211)|
125125

126126
</p>
127127

‎contrib/MedicalSeg/README.md‎

Lines changed: 44 additions & 35 deletions
Original file line numberDiff line numberDiff line change
@@ -1,10 +1,32 @@
11
English | [简体中文](README_CN.md)
22

3-
# MedicalSeg
4-
Medical image segmentation is a pixel-wise/voxel-wise classification of images generated by medical imaging, so that different organs or tissues can be distinguished, is widely used in medical diagnosis and treatment planning. Medical image segmentation can be divided into 2D medical image segmentation and 3D medical image segmentation. 2D medical image segmentation is supported in PaddleSeg. For details, please see [Fundus Data Segmentation Instructions](../../configs/unet/), while 3D image segmentation is handled by MedicalSeg.
3+
# 3D Medical image Segmentaion Solution
54

6-
MedicalSeg is an easy-to-use 3D medical image segmentation toolkit that supports the whole segmentation process including data preprocessing, model training, and model deployment. Specially, We provide data preprocessing acceleration, high precision model on [COVID-19 CT scans](https://www.kaggle.com/andrewmvd/covid19-ct-scans) lung dataset and [MRISpineSeg](https://aistudio.baidu.com/aistudio/datasetdetail/81211) spine dataset, support for multiple datasets including [MSD](http://medicaldecathlon.com/), [Promise12](https://promise12.grand-challenge.org/), [Prostate_mri](https://liuquande.github.io/SAML/) and etc, and a [3D visualization demo](visualize.ipynb) based on [itkwidgets](https://github.com/InsightSoftwareConsortium/itkwidgets). The following image visualize the segmentation results on these two datasets:
5+
## Contents
6+
0. Introduction
7+
1. News
8+
2. EISeg-Med3D
9+
3. Performance
10+
4. Quick Start
11+
5. Full Tutorial
12+
6. Train On Your Dataset
13+
7. Structure
14+
8. TODO
15+
9. License
16+
10. Acknowledgement
17+
18+
## <img src="https://user-images.githubusercontent.com/34859558/190043857-bfbdaf8b-d2dc-4fff-81c7-e0aac50851f9.png" width="25"/> Introduction
19+
20+
Medical image segmentation is a pixel-wise/voxel-wise classification of images generated by medical imaging, so that different organs or tissues can be distinguished. It is widely used in medical diagnosis and treatment planning.
21+
22+
Medical image segmentation can be divided into 2D medical image segmentation and 3D medical image segmentation. 2D medical image segmentation is supported in PaddleSeg. For details, please see [Fundus Data Segmentation Instructions](../../configs/unet/), while 3D image segmentation is handled by MedicalSeg.
23+
24+
**MedicalSeg is an easy-to-use 3D medical image segmentation solution** supporting the whole segmentation process including data preprocessing, model training, and model deployment.
725

26+
The MedicalSeg panorama is as follows, and its main features include:
27+
* Contains the APIs of the whole process medical image segmentation process from data labeling, training, to deployment.
28+
* Including the 3D medical image annotation platform EISeg-Med3D to achieve efficient, accurate and easy-to-use labeling.
29+
* Support six cutting-edge models nnUNet, nnFormer, SwinUNet, TransUNet, UNETR, VNet and corresponding high-precision pre-training models.
830

931
<p align="center">
1032
<img src="https://github.com/shiyutang/files/raw/main/ezgif.com-gif-maker%20(1).gif" width="30.6%" height="20%"><img src="https://github.com/shiyutang/files/raw/main/ezgif.com-gif-maker.gif" width="40.6%" height="20%">
@@ -21,34 +43,20 @@ MedicalSeg is an easy-to-use 3D medical image segmentation toolkit that supports
2143
</p>
2244
</p>
2345

24-
25-
## Contents
26-
0. [News](##News)
27-
1. [Community](##Community)
28-
2. [EISeg-Med3D](##EISeg-Med3D)
29-
3. [Performance](##Performance)
30-
4. [Quick Start](##QuickStart)
31-
5. [Full Tutorial](##FullTutorial)
32-
6. [Train On Your Dataset](##TrainOnYourDataset)
33-
7. [Structure](##Structure)
34-
8. [TODO](##TODO)
35-
9. [License](##License)
36-
10. [Acknowledgement](##Acknowledgement)
37-
38-
39-
4046
## <img src="https://user-images.githubusercontent.com/34859558/190043516-eed25535-10e8-4853-8601-6bcf7ff58197.png" width="25"/> News
4147
- [2022-9] Added 3 cutting-edge models to support whole process deployment applications, including nnformer, TransUnet and nnUnet, allowing you to experience a stronger and more accurate segmentation effect; a new 3D medical image intelligent annotation platform [EISeg-Med3D]( ../../EISeg/med3d/README_en.md) to quickly and easily achieve accurate 3D medical image annotation.
4248
- [2022-4] MedicalSeg releases version 0.1, which provides the whole process from data preprocessing in 3D medical image segmentation to training and deployment, including native support for five datasets, and high-precision preprocessing on vertebrae and lungs Train the model.
4349

44-
## <img src="../../docs/images/chat.png" width="25"/> Community
4550

46-
* If you have any questions, suggestions and feature requests, please create an issues in [GitHub Issues](https://github.com/PaddlePaddle/PaddleSeg/issues).
47-
* Welcome to scan the following QR code and join paddleseg wechat group to communicate with us.
51+
## <img src="../../docs/images/chat.png" width="25"/> Communicate with us
52+
**MedicalSeg just released! If you find any problem using it or want to share any future develop suggestions, please open a github issue or join us by scanning the following QR code.**
53+
4854
<div align="center">
4955
<img src="https://user-images.githubusercontent.com/48433081/174770518-e6b5319b-336f-45d9-9817-da12b1961fb1.jpg" width = "200" />
5056
</div>
5157

58+
59+
5260
## <img src="https://user-images.githubusercontent.com/34859558/188419267-bd117697-7456-4c72-8cbe-1272264d4fe4.png" width="25"/> EISeg-Med3D
5361
In order to solve the problem of low efficiency of 3D medical manual annotation, and to truly empower medical care with AI starting from data annotation, we built [EISeg-Med3D](../../EISeg/med3d/README_en.md), a user-friendly, efficient and intelligent 3D medical image annotation platform, which realizes intelligent and efficient 3D medical data annotation by integrating 3D interactive segmentation models in the annotation process. The main features are as follows:
5462

@@ -58,6 +66,8 @@ In order to solve the problem of low efficiency of 3D medical manual annotation,
5866

5967
* **Convenient**：Install our plugin within three steps; labeling results and progress are automatically saved; the transparency of labeling results can be adjusted to improve labeling accuracy; user-friendly interface interaction makes labeling worry-free and hassle-free。
6068

69+
The detailed doc of using EISeg-Med3D is [here](../../EISeg/med3d/README_en.md)
70+
6171
<div align="center">
6272
<p align="center">
6373
<img src="https://user-images.githubusercontent.com/34859558/188415269-10526530-0415-4632-8223-0e5d755db29c.gif" align="middle" width = 600"/>
@@ -67,25 +77,24 @@ In order to solve the problem of low efficiency of 3D medical manual annotation,
6777
------------------
6878

6979

70-
## <img src="https://user-images.githubusercontent.com/34859558/190044217-8f6befc2-7f20-473d-b356-148e06265205.png" width="25"/> Performance
80+
## <img src="https://user-images.githubusercontent.com/34859558/190044217-8f6befc2-7f20-473d-b356-148e06265205.png" width="25"/> Model Performance
7181

7282
### 1. Accuracy
7383

74-
We successfully validate our framework with [Vnet](https://arxiv.org/abs/1606.04797) on the [COVID-19 CT scans](https://www.kaggle.com/andrewmvd/covid19-ct-scans) and [MRISpineSeg](https://www.spinesegmentation-challenge.com/) dataset. With the lung mask as label, we reached dice coefficient of 97.04% on COVID-19 CT scans. You can download the log to see the result or load the model and validate it by yourself :).
75-
76-
#### **Result on COVID-19 CT scans**
84+
We have added cutting-edge models including nnUNet, nnFormer, SwinUNet, and TransUNet, and all of them surpassed the original paper in terms of segmentation accuracy to varying degrees. Among them, the accuracy of the reproduced TransUNet exceeded the original paper by 3.6%. 81.8% mDice segmentation accuracy.
7785

78-
| Backbone | Resolution | lr | Training Iters | Dice | Links |
79-
|:-:|:-:|:-:|:-:|:-:|:-:|
80-
|-|128x128x128|0.001|15000|97.04%|[model](https://bj.bcebos.com/paddleseg/paddleseg3d/lung_coronavirus/vnet_lung_coronavirus_128_128_128_15k_1e-3/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg3d/lung_coronavirus/vnet_lung_coronavirus_128_128_128_15k_1e-3/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=9db5c1e11ebc82f9a470f01a9114bd3c)|
81-
|-|128x128x128|0.0003|15000|92.70%|[model](https://bj.bcebos.com/paddleseg/paddleseg3d/lung_coronavirus/vnet_lung_coronavirus_128_128_128_15k_3e-4/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg3d/lung_coronavirus/vnet_lung_coronavirus_128_128_128_15k_3e-4/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/scalar?id=0fb90ee5a6ea8821c0d61a6857ba4614)|
86+
Below we show our existing models, pre-trained model parameters and accuracy in the form of a table, welcome to download the log to view the results or load the pre-trained model to improve the training effect on the relevant data set :).
8287

83-
#### **Result on MRISpineSeg**
8488

85-
| Backbone | Resolution | lr | Training Iters | Dice(20 classes) | Dice(16 classes) | Links |
86-
|:-:|:-:|:-:|:-:|:-:|:-:|:-:|
87-
|-|512x512x12|0.1|15000|74.41%| 88.17% |[model](https://bj.bcebos.com/paddleseg/paddleseg3d/mri_spine_seg/vnet_mri_spine_seg_512_512_12_15k_1e-1/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg3d/mri_spine_seg/vnet_mri_spine_seg_512_512_12_15k_1e-1/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/scalar?id=36504064c740e28506f991815bd21cc7)|
88-
|-|512x512x12|0.5|15000|74.69%| 89.14% |[model](https://bj.bcebos.com/paddleseg/paddleseg3d/mri_spine_seg/vnet_mri_spine_seg_512_512_12_15k_5e-1/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg3d/mri_spine_seg/vnet_mri_spine_seg_512_512_12_15k_5e-1/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/index?id=08b0f9f62ebb255cdfc93fd6bd8f2c06)|
89+
|Model| Segment | Dataset | mDice | README | Links |
90+
|:-:|:-:|:-:|:-:|:-:|:-:|
91+
|[nnFormer](https://arxiv.org/abs/2109.03201)| Heart |[ACDC](https://acdc.creatis.insa-lyon.fr/#phase/5846c3ab6a3c7735e84b67f2)|91.8%|[README](configs/acdc/README.md)|[model](https://paddleseg.bj.bcebos.com/paddleseg3d/acdc/nnformer_acdc_160_160_14_250k_4e-4/model.pdparams) \| [log](https://paddleseg.bj.bcebos.com/paddleseg3d/acdc/nnformer_acdc_160_160_14_250k_4e-4/train.log)\| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/scalar?id=b9a90b8aba579997a6f088b840a6e96d)|
92+
|[Vnet](https://arxiv.org/abs/1606.04797)|Lung|[COVID-19 CT scans](https://www.kaggle.com/andrewmvd/covid19-ct-scans)|97.0%|[README](configs/lung_coronavirus/README.md)|[model](https://bj.bcebos.com/paddleseg/paddleseg3d/lung_coronavirus/vnet_lung_coronavirus_128_128_128_15k_1e-3/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg3d/lung_coronavirus/vnet_lung_coronavirus_128_128_128_15k_1e-3/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=9db5c1e11ebc82f9a470f01a9114bd3c)|
93+
|[nnUNet](https://www.nature.com/articles/s41592-020-01008-z)|Lung Tumor|[MSD-Lung](http://medicaldecathlon.com/)|67.9%|[README](configs/nnunet/msd_lung/README.md)|[model](https://aistudio.baidu.com/aistudio/datasetdetail/162872) \| [log](https://aistudio.baidu.com/aistudio/datasetdetail/150774)|
94+
|[Vnet](https://arxiv.org/abs/1606.04797)|Spine|[MRISpineSeg](https://www.spinesegmentation-challenge.com/)|74.7%|[README](configs/mri_spine_seg/README.md)|[model](https://bj.bcebos.com/paddleseg/paddleseg3d/mri_spine_seg/vnet_mri_spine_seg_512_512_12_15k_5e-1/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg3d/mri_spine_seg/vnet_mri_spine_seg_512_512_12_15k_5e-1/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/index?id=08b0f9f62ebb255cdfc93fd6bd8f2c06)|
95+
|[UNETR](https://arxiv.org/abs/2103.10504)|Brain Tumor|[MSD-brain](http://medicaldecathlon.com/)|71.8%|[README](configs/msd_brain_seg/README.md)|[model](https://bj.bcebos.com/paddleseg/paddleseg/medicalseg/msd_brain_seg/unetr_msd_brain_seg_1e-4/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/paddleseg/medicalseg/msd_brain_seg/unetr_msd_brain_seg_1e-4/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/scalar?id=04e012eef21ea8478bdc03f9c5b1032f)|
96+
|[SwinUNet](https://arxiv.org/abs/2105.05537)|Multi-orgrans|[Synapse](https://www.synapse.org/#!Synapse:syn3193805/files/)|82.1%|[README](configs/synapse/README.md)|[model](https://paddleseg.bj.bcebos.com/paddleseg3d/synapse/abdomen/swinunet_abdomen_224_224_1_14k_5e-2/model.pdparams) \| [log](https://paddleseg.bj.bcebos.com/paddleseg3d/synapse/abdomen/swinunet_abdomen_224_224_1_14k_5e-2/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/scalar?id=f62f69b8e9e9210c680dcfc862e3b65b) |****
97+
|[TransUNet](https://arxiv.org/abs/2102.04306)|Multi-orgrans|[Synapse](https://www.synapse.org/#!Synapse:syn3193805/files/)|81.1%|[README](configs/synapse/README.md)|[model](https://paddleseg.bj.bcebos.com/paddleseg3d/synapse/abdomen/transunet_abdomen_224_224_1_14k_1e-2/model.pdparams) \| [log](https://paddleseg.bj.bcebos.com/paddleseg3d/synapse/abdomen/transunet_abdomen_224_224_1_14k_1e-2/train.log) \| [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/scalar?id=d933d970394436aa6969c9c00cf8a6da)|
8998

9099

91100
### 2. Speed

0 commit comments

Comments
 (0)