Characterizing Shape Variations in Infant Hips Using Statistical Shape Modeling for Ultrasonic Diagnosis of Hip Dysplasia with Graf's Method.

Abstract

Purpose: Through the measurement of Graf's alpha (α) angle, the Graf method uses two-dimensional ultrasound (US) to diagnose developmental dysplasia of the hip (DDH) in infants. However, this unidimensional index cannot fully reflect anatomic shape features and variations of iliac wing and bony acetabular roof (IW-AR) coronal outlines that may influence DDH. This study aimed to analyze the shapes of IW-AR outlines by revealing their mean shape, possible shape variations, and the impact of these variations on the α angle variability.

Materials and methods: By segmenting US images of 510 infant hips, IW-AR outlines in Graf's standard plane were obtained from a mixed screening population. A statistical shape model (SSM) was then developed to analyze the outline shapes.

Results: In the IW-AR outlines, shape variations were described by linear combinations of six global and local shape modes. A global mode, dominantly causing an entire outline to bend about the vicinity of its apex, could affect the α angle in a large range (38°-70°). Although a local mode produced bending patterns that had a lesser impact on the α angle (up to 6°), it may relate to DDH diagnosis and clinical outcome.

Conclusion: Shape variations in IW-AR outlines can be effectively modeled by using a compact SSM representative of the variations as linear combinations of a few global and local modes. The shape variations and the angle variability by the local modes should not be underestimated even though the effect of global modes on the α angle is dominant.