Skip to content

linum-uqam/ViT-OCM-WMSegmentation

Repository files navigation

Brain Segmentation Repository

This repository contains three different experiments related to brain image analysis. Here's a brief overview of each experiment:

1. Allend_data_Backbone

This experiment focuses on training a SWIN model using brain section images obtained from the Allen Institute dataset. The goal is to perform image classification on various brain section classes.

2. Allen_Structure_ROIs_dataset

In this experiment, we generate brain section image data by querying specific regions of interest (ROIs) across thousands of full brain acquisitions conducted by the Allen Institute. This data can be used for various brain image analysis tasks.

3. Self-supervised_segmentation

The Self-supervised_segmentation folder contains the core work of the repository. This section represents the culmination of our Master's thesis and has been published as a scientific paper at MLMI 2023. It introduces a novel architecture for unsupervised segmentation of white matter images using self-supervision techniques.

Key Features

  • Cutting-edge self-supervised segmentation methodology.
  • Codebase for replicating the experiments mentioned in the MLMI 2023 paper.
  • Detailed documentation for running the experiments and reproducing the results.

Getting Started

To get started with each experiment, please refer to the respective README files in the corresponding experiment folders.

Citation

If you find this work useful for your research, please consider citing our paper:

@inproceedings{hawchar2023leveraging,
  title={Leveraging Self-attention Mechanism in Vision Transformers for Unsupervised Segmentation of Optical Coherence Microscopy White Matter Images},
  author={Hawchar, Mohamad and Lefebvre, Jo{\"e}l},
  booktitle={International Workshop on Machine Learning in Medical Imaging},
  pages={247--256},
  year={2023},
  organization={Springer}
}

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors