Skull Segmentation and Reconstruction From Newborn CT Images Using Coupled Level Sets.

Area of Science:

• Medical Imaging
• Computational Anatomy

Background:

• Accurate segmentation of neonatal skulls is crucial for diagnosing congenital abnormalities.
• Existing methods struggle with the complex structures of fontanels and sutures in CT images.

Purpose of the Study:

• To develop and validate a new computational method for segmenting and reconstructing neonatal skulls from CT data.
• To accurately delineate cranial bones, fontanels, and sutures.

Main Methods:

• Utilizes geodesic active regions with interacting smooth surfaces propagating inwards and outwards.
• Incorporates specific constraints for coupled interfaces to handle non-detectable features like fontanels and sutures.
• Employs a level set initialization algorithm for accurate surface conformity.

Main Results:

• The method successfully segmented bones, fontanels, and sutures in 18 neonatal CT images.
• Quantitative evaluation using Dice similarity coefficient and modified Hausdorff distance demonstrated satisfactory results compared to manual segmentation.

Conclusions:

• The proposed surface propagation method offers a robust and accurate approach for neonatal skull segmentation and reconstruction.
• This technique has the potential to enhance the diagnosis and management of pediatric craniofacial conditions.