Quantitative identification and segmentation repeatability of thoracic spinal muscle morphology.

Area of Science:

• Radiology and Imaging Science
• Spinal Anatomy and Biomechanics
• Medical Imaging Analysis

Background:

• Spinal muscle morphology is key for diagnosing and treating spinal conditions.
• Current MRI quantification of spinal muscles is limited to cervical and lumbar regions.
• Thoracic spinal muscle morphology influences kyphosis severity in older adults.

Purpose of the Study:

• To develop and validate a quantitative method for identifying thoracic spinal muscle regions of interest (ROIs) in axial MRI.
• To assess the repeatability of measurements for thoracic spinal muscle morphometry.

Main Methods:

• A segmentation methodology was developed for trapezius, erector spinae, and transversospinalis muscles in thoracic MRI.
• Guidelines were based on anatomical landmarks, size, shape, and position.
• 2D (cross-sectional area, position) and 3D muscle geometries were computed.
• Intra- and inter-rater repeatability was assessed using ICC and Dice Coefficient.

Main Results:

• The developed method demonstrated good to excellent intra- and inter-rater repeatability for both 2D and 3D thoracic spinal muscle parameters.
• Average ICC (3,1) was 0.9 (range 0.56-0.98) for 2D parameters.
• Average Dice Coefficient (DC) was 0.92 (range 0.85-0.95) for 3D parameters.

Conclusions:

• The proposed guidelines enable reliable MRI-based morphometric measurements of thoracic spinal muscles.
• This facilitates consistent comparisons of muscle data across global research.
• High segmentation repeatability supports further investigation into posture and spinal curvature effects on thoracic muscle morphology.