We present a multimodal deep imaging decision attention network (DIDANet) that integrates radiomics with traditional ML modules and introduces them into a neural network. DIDANet effectively predicts ER and postoperative TACE benefits in HCC patients. Additionally, it addresses overfitting in small-sample medical models to some extent.
optuna 3.2.0
pycox 0.2.3
scikit-learn 1.2.2
torch 2.0.1
Please see requirements.txt for all the other requirements.
Clone this repo:
git clone https://github.com/iWiley/DIDANet.git
Create a virtual conda environment named Radiomics-CT-master with the following command:
conda create --name Radiomics-CT-master --file requirements.txt
conda activate Radiomics-CT-master