Spinal ligaments detection on vertebrae meshes using registration and 3D edge detection

TL;DR

The paper addresses automatic detection of spinal ligament attachment points on vertebrae meshes to enable accurate biomechanical spine models. It introduces a pipeline using registration based on fifteen Points of Interest (PoIs), followed by edge-detection refinement to project ligaments onto patient-specific vertebrae, resulting in 66 landmark detections. Experimental results on VerSe 2021 vertebrae show favorable accuracy for anterior ligaments (e.g., PLL ~1.68 mm, ALL ~2.24 mm) with an overall average error of 3.64 mm and RMSE of 3.98 mm, and a runtime of ~3.0 seconds per vertebra, outperforming slower baselines. The authors release two annotated datasets and a ready-to-use 3D Slicer plugin, highlighting the method's practicality for rapid, automated ligament localization in biomechanical spine models.

Abstract

Spinal ligaments are crucial elements in the complex biomechanical simulation models as they transfer forces on the bony structure, guide and limit movements and stabilize the spine. The spinal ligaments encompass seven major groups being responsible for maintaining functional interrelationships among the other spinal components. Determination of the ligament origin and insertion points on the 3D vertebrae models is an essential step in building accurate and complex spine biomechanical models. In our paper, we propose a pipeline that is able to detect 66 spinal ligament attachment points by using a step-wise approach. Our method incorporates a fast vertebra registration that strategically extracts only 15 3D points to compute the transformation, and edge detection for a precise projection of the registered ligaments onto any given patient-specific vertebra model. Our method shows high accuracy, particularly in identifying landmarks on the anterior part of the vertebra with an average distance of 2.24 mm for anterior longitudinal ligament and 1.26 mm for posterior longitudinal ligament landmarks. The landmark detection requires approximately 3.0 seconds per vertebra, providing a substantial improvement over existing methods. Clinical relevance: using the proposed method, the required landmarks that represent origin and insertion points for forces in the biomechanical spine models can be localized automatically in an accurate and time-efficient manner.

Figures (5)

• Figure 1: Anatomically grouped 3D ligament landmarks annotated on the artificial L3 vertebra model panjabi1980basic. The ligaments PLL, ALL, ITL, FL, and CL, which run over a particular surface area, are divided into different ligament bundles.
• Figure 2: Workflow of the proposed method: Starting with annotated ligament landmarks on an artificial vertebra and the model of a patient-specific vertebra (left) we first detect 15 Points of Interest (PoIs) as the anatomical extrema/most outer points (middle). Using these points we register the annotated ligament landmarks to the patient specific vertebra (3). Since the landmarks might still be off due to the patient specific vertebra geometry, we now project them to the surface using edge detection (4). The result is (5), in which all landmarks are aligned with the patient specific vertebra geometry.
• Figure 3: Workflow of the ligament landmark projection: (a) We calculate the center of mass of the initial landmarks and determine a plane oriented as in the picture intersecting the center of mass, (b) the intersection between the vertebra and the plane is detected, (c) within a fixed radius around each ligament landmark we determine so-called edge values in a similar manner to ahmed2018edge, (d) the intersection point with the highest edge value is the most promising match for the ligament landmark.
• Figure 4: The ligament landmarks (red) are detected on the fractured vertebra model by (a) our method and (b) ALPACA Porto2021ALP in comparison to the ground truth (green).
• Figure 5: The ligament landmarks in the anterior (a) and posterior (b) parts of the healthy L1-vertebra are detected by our method (red) and compared to the ground truth (green).