Quantitative analysis and automatic classification of skull deformity based on combined two- and three-dimensional shape indices

For early diagnosis and evaluation of skull deformities, quantification of the severity of skull deformity and automatic classification between the skull deformities are necessary. 3D CT images, which are commonly used to diagnose skull deformities, are prone to inter-observer variability depending on clinical experience. In this paper, we propose 2D and 3D shape indices by considering prominent areas such as restricted growth or compensative growth in the skull of the representative planes and by considering the degree of regional bone deformation of deformed skull relative to the normal skull. In experimental results, the 2D and 3D shape indices showed 13.89% and 16.67% improvement in the overall accuracy compared to TCI, respectively. The combined 2D and 3D indices showed best performance with 95.83% overall accuracy.