This repository contains the source code associated with our paper titled "Bidirectional Mapping with Contrastive Learning on Multimodal Neuroimaging Data" which has been accepted at MICCAI 2023.
Paper link: Bidirectional Mapping with Contrastive Learning on Multimodal Neuroimaging Data
The HCP dataset can be accessed at here. The OASIS dataset can be accessed at here. The datasets should be placed under the data/ folder.
Ensure all the necessary packages listed in requirements.txt are installed.
To conduct BMCL experiments, please execute the command below. Note that you should adjust the paths and hyperparameters according to your specific requirements:
python main.py --max_epoch 1000 --batch_size 64 --sample_size 1200 --target gender --dataset HCP --graph_data_dir './data/graphs' --label_path './data/labels.csv' --seq_data_dir './data/sequences' --hidden_dim 512 --hidden_layers 2 --merge_layers 2 --num_heads 4 --lr 1e-4 --factor 0.5 --patience 20 --recon FalseTo perform BMCL experiments with five-fold cross-validation, please execute the command below. Note that you should adjust the paths and hyperparameters according to your specific requirements:
python main_five_fold.py --max_epoch 1000 --batch_size 64 --sample_size 1200 --target gender --dataset HCP --graph_data_dir './data/graphs' --label_path './data/labels.csv' --seq_data_dir './data/sequences' --hidden_dim 512 --hidden_layers 2 --merge_layers 2 --num_heads 4 --lr 1e-4 --factor 0.5 --patience 20 --recon FalseIf you find this work useful in your research, please consider citing:
@inproceedings{ye2023bidirectional,
title={Bidirectional Mapping with Contrastive Learning on Multimodal Neuroimaging Data},
author={Ye, Kai and Tang, Haoteng and Dai, Siyuan and Guo, Lei and Liu, Johnny Yuehan and Wang, Yalin and Leow, Alex and Thompson, Paul M and Huang, Heng and Zhan, Liang},
booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
pages={138--148},
year={2023},
organization={Springer}
}