EPISeg: Automated segmentation of the spinal cord on echo planar images using open-access multi-center data.

Area of Science:

• Neuroimaging
• Medical Image Analysis

Background:

• Spinal cord functional magnetic resonance imaging (fMRI) is crucial for understanding neurological functions.
• Current automated segmentation of spinal cord fMRI data is challenging due to low resolution and image artifacts.
• Manual segmentation is time-consuming and prone to user bias.

Purpose of the Study:

• To develop an automated deep learning model for spinal cord segmentation on gradient-echo EPI data.
• To create and share a multi-center dataset of spinal cord gradient-echo EPI with ground-truth segmentations.

Main Methods:

• Development of a deep learning model named EPISeg.
• Training and validation using a multi-center dataset of spinal cord gradient-echo EPI images.
• Comparison with existing spinal cord segmentation models.

Main Results:

• EPISeg demonstrates significant improvements in segmentation quality compared to existing models.
• The model shows resilience to various acquisition protocols and common fMRI artifacts.
• A multi-center dataset and training code were made publicly available.

Conclusions:

• EPISeg offers an effective automated solution for spinal cord segmentation in fMRI.
• The developed model and dataset will advance spinal cord fMRI research.
• The tool is integrated into the Spinal Cord Toolbox for wider accessibility.