Modeling Life-span Brain Age from Large-scale Dataset based on Multi-level Information Fusion

This is a PyTorch implementation of the paper "Modeling Life-span Brain Age from Large-scale Dataset based on Multi-level Information Fusion", July, 2023.

1. Installation

(1) Create conda env and install pytorch

conda create -n brain python=3.9
conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia

(2) Install relevant libraries

pip install -r requirements.txt

2. Network Architecture

3. Data Distribution

Demographic information of 8 cohorts

Age Distribution on Healthy Controls	Age Distribution on Brain Disorders

4. Prediction Performance

Predictions on Healthy Controls	Predictions on Brain Disorders

5. Train the Model

To train the model, run train_threedim_3view_GAF.py file provided in the repository.

batch_size=8
learning_rate=0.001
weight_decay=0.0001
n_epochs=200
n_exps=1 # num of independent experiments

# ============= Training and Parameter Configuration ==============
python train_threedim_3view_GAF.py                       \
--batch_size        $batch_size                          \
--lr_s              $learning_rate                       \
--wd_s              $weight_decay                        \
--n_epochs          $n_epochs                            \
--n_exps            $n_exps                              \

6. Test on Brain Disorders

To test the model on brain disorders, run test_BDs.py.

python test_BDs.py