Segmentation.md

PanDerm - Skin Lesion Segmentation

This implementation uses PanDerm for skin lesion segmentation.

Install Environment

Main dependencies: Pytorch, Pytorch Lightning, MMSegmentation

cd segmentation
conda create -n dermseg python=3.10
conda activate dermseg
pip install torch==2.2.1 torchvision==0.17.1 --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

pip install -U openmim
mim install mmengine==0.10.4
mim install mmcv==2.1.0
mim install mmsegmentation==1.2.2

Data Preparation

1. Download datasets:
   • ISIC2018 task1
   • HAM10000
2. Change --parent_path in run.sh accordingly

Fine-tuning with PanDerm Weight

1. Download Pretrained Weights

   • PanDerm weights

2. Set Pretrained Path

   • Modify --parent_path in run.sh accordingly

3. Start Training

   bash run.sh

   Note: Adjust the path config to your desired storage location

4. Evaluation

   • Add --evaluate to run.sh for evaluation mode

   cd segmentation
   add --evaluate

Starter Code for Your Own Dermoscopic Images

See evaluate.ipynb to learn:

• Using it for skin lesion segmentation