Enhancing Low Back Pain Assessment with Diffusion Models for Lumbar Spine MRI Segmentation

2026-08-05

AI Summary

Accurate semantic segmentation of degenerative structures such as intervertebral discs in lumbar MRI remains challenging for patients with low back pain. This work proposes SpineSegDiff, the first framework to apply diffusion models to lumbar spine MRI segmentation, enabling robust performance on both T1- and T2-weighted images. Evaluated on the SPIDER dataset, SpineSegDiff achieves overall segmentation accuracy comparable to state-of-the-art non-diffusion approaches like nnU-Net, while significantly improving the detection accuracy of degenerated discs. Furthermore, the method enhances clinical interpretability by generating uncertainty maps that provide insight into segmentation reliability, thereby supporting radiological assessment.

Abstract

This study introduces a diffusion-based framework for robust and accurate semantic segmentation of lumbar spine MRI scans from patients with low back pain (LBP), regardless of whether the scans are T1- or T2-weighted. We compared with advanced models for segmenting vertebrae, intervertebral discs (IVDs), and spinal canal using the SPIDER dataset. The results showed that SpineSegDiff achieved a segmentation performance comparable to that of the state-of-the-art non-diffusion nnUnet, particularly in improving the identification of degenerated IVDs. In addition, the uncertainty maps generated by our model provide valuable insights for clinical review, enhancing the robustness and reliability of the segmentation results. The potential of diffusion models to enhance the diagnosis and management of LBP through more precise analysis of pathological spine MRI is underscored by our findings.

Problem

• low back pain
• lumbar spine MRI
• semantic segmentation
• intervertebral disc degeneration
• medical image analysis

Innovation

• diffusion models
• lumbar spine MRI segmentation
• low back pain
• uncertainty quantification
• semantic segmentation