PTRNet

A Novel Multimodal Learning Method for Predicting Treatment Resistance in MPO-AAV with Lung Involvement

[Model] [Paper] [BibTeX]

Method Overview

Install

1. Download the repository and open the PTRNet

git clone https://github.com/yinangit/PTRNet.git
cd PTRNet

2. Required requirements

cd env
conda env create -f environment.yml
conda activate PTRNet
cd tab-transformer-pytorch
pip install -e .

3. Optional requirements (for feature extraction)

conda activate PTRNet
cd env/CLIP
pip install -e .
cd ../lungmask
pip install -e .
cd ../timm-0.9.12
pip install -e .

Run

• Train

conda activate PTRNet
python scripts/train_modality_ablation.py --saveName modality_ablation --model_mode union --d_model 256

• Test

conda activate PTRNet
python scripts/test.py --saveName modality_ablation --model_mode union --d_model 256 --weight_path log/modality_ablation/model_final.pt

• Ablation

Taking ablation of hyperparameter γ as an example:

conda activate PTRNet
cd launch
bash ablation_gamma.sh

Acknowledgements

Citation