EEG power spectra parameterization and adaptive channel selection towards efficient semi-supervised seizure prediction
This is code for the paper [EEG power spectra parameterization and adaptive channel selection towards efficient semi-supervised seizure prediction] which in the minor revision
@article{lihanyi1761,
title={EEG power spectra parameterization and adaptive channel selection towards efficient semi-supervised seizure prediction},
author={Hanyi Lia, Jiahui Liao, Hongxiao Wang, Chang’an A. Zhan and Feng Yang},
journal={computers in biology and medicine},
year={2024}
}
To obtain the channel selection results based on PAPC, please run the code below the file 'PAPC'
Set the paths in *.json files. Copy files in folder "copy-to-CHBMIT" to your CHBMIT dataset folder.
Prepare preprocessed data for WGAN-GP training. A large storage is required.
python3 main.py --mode save_STFT --dataset DATASET
Train model.
python3 main.py --mode wgan-gp --dataset DATASET
* For comparison, MODEL can be dcgan,wgan,wgan_gp .
5. Leave-one-seizure-out cross-validation.
```console
python3 main.py --mode cvgan --dataset DATASET