Diffusion-Based Semantic Segmentation of Lumbar Spine MRI Scans of Lower Back Pain Patients

TL;DR

This paper presents SpineSegDiff, a diffusion-based framework that enhances the accuracy and robustness of segmenting lumbar spine structures in MRI scans, outperforming existing models especially in identifying degenerated intervertebral discs.

Contribution

The study introduces a novel diffusion-based segmentation method for lumbar spine MRI scans that works across T1w and T2-weighted images, improving accuracy over prior models.

Findings

SpineSegDiff outperforms non-diffusion models in segmenting degenerated IVDs.

The framework is effective on both T1w and T2-weighted MRI scans.

Diffusion models show promise for improving LBP diagnosis.

Abstract

This study introduces a diffusion-based framework for robust and accurate segmenton of vertebrae, intervertebral discs (IVDs), and spinal canal from Magnetic Resonance Imaging~(MRI) scans of patients with low back pain (LBP), regardless of whether the scans are T1w or T2-weighted. The results showed that SpineSegDiff achieved comparable outperformed non-diffusion state-of-the-art models in the identification of degenerated IVDs. Our findings highlight the potential of diffusion models to improve LBP diagnosis and management through precise spine MRI analysis.