MRI-SEG Tool (Cross-Platform)

MRI-SEG, Spine MRI Segmentation and Classification Tool (Cross-Platform)

I developed MRI-SEG, a cross-platform software package for automatic classification and segmentation of Spine MRI images. The tool features a custom GUI designed for ease of use on both Windows and Linux systems. MRI-SEG integrates a deep learning model that I personally built and trained, enabling accurate segmentation and diagnosis support for spinal structures in MRI scans.

Data Source: Spine MRI data was provided by AIIMS Raebareli, Department of Neurosurgery, under the guidance of Dr. Suyash Singh. This project was a collaborative effort between NIT Meghalaya and AIIMS Raebareli.

Features

  • Custom-built deep learning architecture for spine MRI segmentation.
  • GUI-based application with real-time visualization and result output.
  • Compatible with Windows and Linux platforms.
  • Designed for ease of integration into clinical workflows and research settings.
Overview of the performance of a customized nnU-Net version (used in the initial training step).

Note: The source code is currently private due to permission constraints and will be released as open-source once approved.

Check nnU-Net here.

Note: More information about this project will be added soon.

If you’re looking for MRI-SEG Lite, a lightweight version of MRI-SEG specifically developed for resource-constrained devices, please refer to this page.