Skip to content
/ DFC Public

"DFC: Anatomically Informed Fiber Clustering with Self-supervised Deep Learning for Fast and Effective Tractography Parcellation".

Notifications You must be signed in to change notification settings

SlicerDMRI/DFC

Repository files navigation

DFC (Deep Fiber Clustering)

Deep fiber clustering: Anatomically informed fiber clustering with self-supervised deep learning for fast and effective tractography parcellation

This code implements a deep learning method for white matter fiber clustering using diffusion MRI data, as described in the following paper:

Chen Y, Zhang C, Xue T, Song Y, Makris N, Rathi Y, Cai W, Zhang F, O'Donnell LJ. Deep Fiber Clustering: Anatomically Informed Fiber Clustering with Self-supervised Deep Learning for Fast and Effective Tractography Parcellation. NeuroImage. 2023 Apr 3:120086.

fig1

Installation

The code has been tested with Python 3.7, Pytorch 1.7.1, CUDA 10.1 on Ubuntu 18.04.
whitematteranalysis
scikit-learn

Usage

To train a model for fiber clustering with tractography data:

python train.py -indir <path of training data>

To evaluate the model with testing data:

python test.py -indir <path of testing data> -modeldir <path of training model>

Fast and effective fiber clustering was achieved with the proposed method. Below is a visualization of the obtained clusters.

fig2

The training model and testing dataset are available here: https://github.com/SlicerDMRI/DFC/releases

See our project page https://deepfiberclustering.github.io/ for more details.

Reference

Chen Y, Zhang C, Xue T, Song Y, Makris N, Rathi Y, Cai W, Zhang F, O'Donnell LJ. Deep Fiber Clustering: Anatomically Informed Fiber Clustering with Self-supervised Deep Learning for Fast and Effective Tractography Parcellation. NeuroImage. 2023 Apr 3:120086.

About

"DFC: Anatomically Informed Fiber Clustering with Self-supervised Deep Learning for Fast and Effective Tractography Parcellation".

Resources

Stars

Watchers

Forks

Packages

No packages published