Development of a statistical shape model of the human scapula: Development of a statistical shape model of the human scapula for improvement in vehicle safety aspects

Abstract

The human scapula is the fundamental structure of the shoulder and an anatomical landmark for upper limb movements. Understanding variations in scapula shape can influence biomechanical modeling, clinical diagnosis, and safety simulation, particularly in vehicle crash testing. The aim of this study is to develop a Statistical Shape Model (SSM) of the human scapula bone from computed tomography (CT) image data segmentations to analyze and describe the variability in the population based on sex, height, age, and BMI. A dataset comprising 70 segmented scapulae derived from an original set of 230 segmented scapulae, including healthy samples and samples with Hill-Sachs lesions, some of which are considered unhealthy, was used for this study. The landmarking set-up consisted of 45 landmarks in total, including 20 anatomical landmarks and 25 pseudo-landmarks placed along splines, which provide reliable reference points for shape analysis. Morphing techniques, performed primarily in Python code, were used to standardize the mesh structures of scapulae by mapping every sample onto a common average model. For ease of comparison, General Procrustes Analysis (GPA) was used first to remove position, orientation, and scale variance, and then Principal Component Analysis (PCA) to capture the main modes of shape variation in the dataset. Then, a regression model is used on the principal components of the PCA to study variations based on population data. The resulting SSM provides valuable information on the variation of the anatomical scapula of the population and how demographic factors influence bone shape and size. The results have potential applications to personalized biomechanical modeling, clinical diagnostics, and improved occupant protection in vehicle crash simulations.