- 2024.08.30: This repo is released.
# clone this repo
git clone https://github.com/xiaohuawan/MitoStructSeg.git
cd MitoStructSeg
# create environment
conda create -n MitoStructSeg python=3.9.19
conda activate MitoStructSeg
pip install -r requirements.txt| Dataset | Dataset Name | Source Domain | Target Domain | Validation | |||
|---|---|---|---|---|---|---|---|
| Quark Cloud Disk | Google Cloud Disk | Quark Cloud Disk | Google Cloud Disk | Quark Cloud Disk | Google Cloud Disk | ||
| Human Myocardium Dataset | Patient#1 | Download | Download | Download | Download | Download | Download |
| Patient#2 | Download | Download | Download | Download | |||
| Patient#3 | Download | Download | Download | Download | |||
| Mouse Kidney Dataset | Mouse Kidney | Download | Download | Download | Download | Download | Download |
| Model Name | Description | Quark Cloud Disk | Google Cloud Disk |
|---|---|---|---|
| Patient#1.ckpt | MitoStructSeg trained on Patient#1 | download (pwd: xdJe) |
download |
| Patient#2.ckpt | MitoStructSeg trained on Patient#2 | download (pwd: L82x) |
download |
| Patient#3.ckpt | MitoStructSeg trained on Patient#3 | download (pwd: kpdH) |
download |
| Mouse Kidney.ckpt | MitoStructSeg trained on Mouse Kidney | download (pwd: b5Nb) |
download |
| classification.pt | model for evaluating classification | download (pwd: NEMP) |
download |
├── data
│ ├── patient1
│ │ ├── Source_domain
│ │ │ ├── data_block1
│ │ │ ├── data_block2
│ │ │ ├── label_block1
│ │ │ └── label_block2
│ │ ├── Target_domain
│ │ │ └── data
│ │ └── Valid
│ │ ├── data
│ │ └── label
├── models
│ ├── Patient#1.ckpt
│ ├── Patient#2.ckpt
│ ├── Patient#3.ckpt
│ └── Mouse_Kidney.ckpt
├── src
│ ├── config
│ ├── dataset
│ ├── model
│ ├── scripts
│ └── utils
-
Fill in the training configuration file with appropriate values.
-
Start training!
cd /MitoStructSeg/src python main.py -c Patient#1_config
We store our trained models at GoogleDrive or Quark Cloud
-
Fill in the training configuration file with appropriate values.
-
Start inference!
cd /MitoStructSeg/src python inference.py -c Patient#1_config
The system is divided into four main sections: classification assessment, image segmentation, precise calculation.
1.Configuration
-
Download Node.js 18.17.1
-
Create Symbolic Links
ln -s /root/node-v18.17.1-linux-x64/bin/node /usr/local/bin/node
-
Edit the Environment Configuration File
export NODEJS_HOME=/usr/local/lib/node/nodejs export PATH=$NODEJS_HOME/bin:$PATH
-
Verify the Installation
node -v npm -v
2.Usage
-For Windows:
-Run the following command directly in the terminal:
```shell
python start_win.py
```
-For Linux:
-Run the following command directly in the terminal:
```shell
python start_linux.py
```
Figure 1. Mitochondrial Health Assessment Interface.
Figure 2. Segmentation of 2D Images Interface.
Figure 3. Membrane Structure Calculation Interface.
presentation.workflow.mp4
3D visualization of segmentation results in 800×800×400 voxel blocks
🟢 Green: Healthy mitochondria | 🔴 Red: Damaged mitochondria
Patient #1
Patient.1.mp4
Patient #2
Patient.2.mp4
Patient #3
Patient.3.mp4
Mouse_Kidney.mp4
This program is built upon a set of great works: