Transformer-based Multimodal Fusion for Early Diagnosis of Alzheimer’s Disease Using Structural MRI and PET
This repo contains PyTorch implementation of the paper: Transformer-based Multimodal Fusion for Early Diagnosis of Alzheimer’s Disease Using Structural MRI and PET.
Directory structure
├── MRI # MRI images
│ ├── sub-ADNI001S0001.nii.gz
│ ├── sub-ADNI009S0001.nii.gz
│ ├── ...
├── PET # PET images
│ ├── sub-ADNI001S0001.nii.gz
│ ├── sub-ADNI009S0001.nii.gz
│ ├── ...
├── ADNI.csv # label file
Label file
| Subject | Group | Age | ...
————————————————————————————————————————
| sub-ADNI001S0001 | AD | xx | ...
| sub-ADNI009S0001 | CN | xx | ...
| sub-ADNI002S0001 | pMCI | xx | ...
| sub-ADNI003S0001 | sMCI | xx | ...
| ... | ... | .. | ...
pip install -r ./requirements.txt
python kfold_train_adversarial.py --randint False --aug True --batch_size 8 --name <expr_name> --task <ADCN/pMCIsMCI> --model <CNN/Transformer> --dataroot <data_dir>
If you think our research work helpful, please consider citing our original paper.
@inproceedings{zhang2023transformer,
title={Transformer-Based Multimodal Fusion for Early Diagnosis of Alzheimer's Disease Using Structural MRI And PET},
author={Zhang, Yuanwang and Sun, Kaicong and Liu, Yuxiao and Shen, Dinggang},
booktitle={2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI)},
pages={1--5},
year={2023},
organization={IEEE}
}