Computer-assisted shape descriptors for skull morphology in craniosynostosis

Area of Science:

• Medical Imaging
• Radiology
• Pediatric Surgery

Background:

• Skull deformities, including craniosynostosis, require accurate diagnosis for timely intervention.
• Computed tomography (CT) imaging is crucial for assessing cranial morphology.
• Existing methods for characterizing skull deformities may lack sufficient sensitivity and specificity.

Purpose of the Study:

• To develop and validate a novel method for characterizing common skull deformities.
• Utilize two-dimensional (2D) shape descriptors derived from CT images.
• Achieve high sensitivity and specificity in differentiating normal and abnormal cranial shapes.

Main Methods:

• Analysis of CT images from 44 normal subjects and 39 infants with craniosynostosis (sagittal and bicoronal).
• Formulation of three 2D shape descriptors: cranial index (CI), cranial radius index (CR), and cranial extreme spot index (CES).
• Application of these descriptors to reference planes (sagittal and coronal) of CT images.

Main Results:

• Distinct patterns of CI and CES were observed for sagittal and bicoronal craniosynostosis.
• Sagittal craniosynostosis showed a mean S-CI < 1.0 and a specific CES pattern.
• Bicoronal craniosynostosis exhibited a mean CI > 1.0 and a different CES pattern.
• Joint analysis of CI and CES effectively discriminated between normal and deformed skulls.

Conclusions:

• Newly developed 2D shape descriptors accurately recognize common skull deformities in CT imaging.
• These descriptors demonstrate statistical significance in differentiating skull shapes.
• The findings may support the development of automated CT-based diagnosis for craniosynostosis.