Enhancing Low Back Pain Assessment with Diffusion Models for Lumbar Spine MRI Segmentation
Keywords: Diffusion Models, Lumbar Spine MRI, pre-segmentation
TL;DR: A diffusion-based model, achieves state-of-the-art performance in pathological IVD disc semantic segmentation of lumbar spine MRI scans for low back pain patients, providing valuable variability maps for clinical decision-making
Abstract: This study introduces a diffusion based framework for robust and accurate se4 mantic segmentation of lumbar spine MRI scans from patients with low back pain (LBP), regardless of whether the scans are T1w 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 state-of-the-art performance, particularly in 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.
Primary Subject Area: Segmentation
Secondary Subject Area: Generative Models
Paper Type: Validation or Application
Registration Requirement: Yes
Reproducibility: https://gitlab.ethz.ch/BMDSlab/publications/ low-back/diffusion-models-for-lumbar-spine-mri-segmentation
Submission Number: 156
Loading