Subject-specific shoulder muscle attachment region prediction using statistical shape models: A validity study.

Area of Science:

• Biomechanics
• Medical Imaging
• Computational Anatomy

Background:

• Accurate subject-specific musculoskeletal models are crucial for clinical biomechanics.
• Current models lack precise subject-specific muscle parameters, limiting predictive accuracy.
• Shoulder joint modeling is complex due to intricate muscle attachments.

Purpose of the Study:

• To develop and validate a statistical shape model (SSM) for predicting subject-specific muscle attachment regions on scapula and humerus.
• To assess the accuracy of predicted muscle attachment regions using similarity measures.
• To demonstrate the utility of this method in enhancing musculoskeletal modeling pipelines.

Main Methods:

• Augmented statistical shape models (SSMs) were constructed using bone meshes and key shoulder muscle attachment regions (Subscapularis, Supraspinatus, Infraspinatus, Teres Major, Teres Minor).
• Muscle attachment regions were defined by vertex subsets on bone meshes, tracked via identifiers.
• Subject-specific regions were predicted using independent bone datasets, with validity assessed through visual inspection and quantitative similarity measures.

Main Results:

• The SSM-based method demonstrated excellent concurrent validity in predicting subject-specific muscle attachment regions.
• Quantitative similarity measures confirmed the high accuracy of the predicted regions compared to manual segmentations.
• The approach showed strong potential for integration into automated medical image segmentation and modeling workflows.

Conclusions:

• The developed statistical shape model effectively predicts subject-specific shoulder muscle attachment regions with high accuracy.
• This method offers a significant improvement over current strategies lacking precise subject-specific muscle parameters.
• Further validation across all shoulder complex muscles is recommended for broader clinical application.