MDM

By Junyan Lyu, Perry F. Bartlett, Fatima A. Nasrallah, Xiaoying Tang.

This repository contains an official implementation of MDM for the IEEE Transactions on Medical Imaging paper "Masked Deformation Modeling for Volumetric Brain MRI Self-supervised Pre-training".

Quick start

Environment

This code is developed using on Python 3.9.13 and Pytorch 2.0.1 on CentOS 7 with NVIDIA GPUs. Training and testing are performed using 4 NVIDIA Tesla A100 or RTX A6000 GPU with CUDA 11.8. Other platforms or GPUs are not fully tested.

Install

conda create -n mdm python=3.9.13
conda activate mdm
pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

Data

1. Select and download the isotropic (spacing < 1.1mm) T1-weighted brain MRIs from ADNI.
2. Perform skull-stripping and affine transformation to MNI152 for all the MRIs using ANTs.
3. Create a file list for preprocessed MRI paths using

cd MDM
ls path/to/mri/*.gz > datasets/adni_mni152_affine_clean.txt

Pre-training

Modify --data_path and --atlas_path into your actual paths

bash scripts/pretrain.sh mdm_run1

Citation

If you find this repository useful, please consider citing TMI paper:

@ARTICLE{10777582,
  author={Lyu, Junyan and Bartlett, Perry F. and Nasrallah, Fatima A. and Tang, Xiaoying},
  journal={IEEE Transactions on Medical Imaging}, 
  title={Masked Deformation Modeling for Volumetric Brain MRI Self-supervised Pre-training}, 
  year={2024},
  volume={},
  number={},
  pages={1-1},
  keywords={Brain;Magnetic resonance imaging;Deformation;Brain modeling;Image segmentation;Image restoration;Biomedical imaging;Annotations;Feature extraction;Lesions;Self-supervised learning;masked deformation modeling;brain segmentation},
  doi={10.1109/TMI.2024.3510922}}