Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Panoptic Segmentation and Labelling of Lumbar Spine Vertebrae using Modified Attention Unet

Rikathi Pal, Priya Saha, Somoballi Ghoshal, Amlan Chakrabarti, Susmita Sur-Kolay

TL;DR

This work addresses the challenge of accurate segmentation and labeling of lumbar spine vertebrae in MRI, proposing a Modified Attention U-Net that integrates centroid-based, multi-label panoptic masking to enable 3D panoptic segmentation from 2D sagittal slices. The method expands the network depth, balances depth with spatial resolution, and employs a dual loss to optimize semantic and instance segmentation, achieving a reported vertebra tagging accuracy of 99.5% and IoU > 0.9. Evaluation on multiple lumbar MRI datasets demonstrates robust performance and potential clinical impact for diagnosis, treatment planning, and surgical guidance. The approach advances spine imaging analytics by providing precise 3D vertebral delineation and automated labeling, with future prospects in robotic-assisted planning and broader MRI-domain applicability.

Abstract

Segmentation and labeling of vertebrae in MRI images of the spine are critical for the diagnosis of illnesses and abnormalities. These steps are indispensable as MRI technology provides detailed information about the tissue structure of the spine. Both supervised and unsupervised segmentation methods exist, yet acquiring sufficient data remains challenging for achieving high accuracy. In this study, we propose an enhancing approach based on modified attention U-Net architecture for panoptic segmentation of 3D sliced MRI data of the lumbar spine. Our method achieves an impressive accuracy of 99.5\% by incorporating novel masking logic, thus significantly advancing the state-of-the-art in vertebral segmentation and labeling. This contributes to more precise and reliable diagnosis and treatment planning.

Panoptic Segmentation and Labelling of Lumbar Spine Vertebrae using Modified Attention Unet

TL;DR

This work addresses the challenge of accurate segmentation and labeling of lumbar spine vertebrae in MRI, proposing a Modified Attention U-Net that integrates centroid-based, multi-label panoptic masking to enable 3D panoptic segmentation from 2D sagittal slices. The method expands the network depth, balances depth with spatial resolution, and employs a dual loss to optimize semantic and instance segmentation, achieving a reported vertebra tagging accuracy of 99.5% and IoU > 0.9. Evaluation on multiple lumbar MRI datasets demonstrates robust performance and potential clinical impact for diagnosis, treatment planning, and surgical guidance. The approach advances spine imaging analytics by providing precise 3D vertebral delineation and automated labeling, with future prospects in robotic-assisted planning and broader MRI-domain applicability.

Abstract

Segmentation and labeling of vertebrae in MRI images of the spine are critical for the diagnosis of illnesses and abnormalities. These steps are indispensable as MRI technology provides detailed information about the tissue structure of the spine. Both supervised and unsupervised segmentation methods exist, yet acquiring sufficient data remains challenging for achieving high accuracy. In this study, we propose an enhancing approach based on modified attention U-Net architecture for panoptic segmentation of 3D sliced MRI data of the lumbar spine. Our method achieves an impressive accuracy of 99.5\% by incorporating novel masking logic, thus significantly advancing the state-of-the-art in vertebral segmentation and labeling. This contributes to more precise and reliable diagnosis and treatment planning.
Paper Structure (17 sections, 9 equations, 10 figures, 1 table)

This paper contains 17 sections, 9 equations, 10 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Preliminaries
  4. Proposed Methodology
  5. Data Preparation and Masking
  6. Model Training for Panoptic Segmentation
  7. Input Preprocessing:
  8. Proposed Modified Attention UNet Architecture:
  9. Training and Evaluation:
  10. Output Generation:
  11. Testing:
  12. Results
  13. Dataset Description
  14. Experimental Configuration and Parameterization
  15. Performance Evaluation Metrics
  16. ...and 2 more sections

Figures (10)

  • Figure 1: Generating centroid and corner points from reconstructed slices
  • Figure 2: Proposed Masking Process
  • Figure 3: Block diagram of the proposed method (a) Training Phase (b) Testing Phase
  • Figure 4: Attention Unet Model
  • Figure 5: Example 1 for Vertebra labeling in 2D lumbar sagittal slice -- (a) Input image, (b) Output labeled vertebra
  • ...and 5 more figures