Smoothing Splines on Unit Ball Domains with Application to Corneal Topography

Area of Science:

• Biomedical Imaging
• Medical Image Analysis
• Computational Anatomy

Background:

• Optical Coherence Tomography (OCT) is crucial for non-invasively imaging biological tissues, particularly the human eye's anterior chamber.
• Accurate reconstruction of tissue shape from OCT data is challenging, requiring both data fidelity and inherent smoothness.
• Similar shape reconstruction problems exist in other medical imaging modalities like Magnetic Resonance Imaging (MRI).

Purpose of the Study:

• To develop a novel algorithm for reconstructing biological tissue shapes from OCT images.
• To ensure reconstructed surfaces accurately fit imaging data while preserving observed smoothness properties.
• To unify existing ad-hoc approaches for surface reconstruction in medical imaging.

Main Methods:

• The problem is formulated as penalized weighted least squares regression.
• A penalty is applied to the magnitude of the second derivative (Laplacian) of the surface.
• A novel algorithm is presented to construct the Kimeldorf-Wahba solution for unit ball domains.

Main Results:

• The proposed method unifies various ad-hoc surface reconstruction approaches.
• The algorithm was applied to data from an anterior segment OCT.
• A detailed comparison of reconstructed surfaces using different methods was performed, demonstrating the efficacy of the novel approach.

Conclusions:

• The developed algorithm provides a unified and effective method for reconstructing biological tissue shapes from OCT data.
• This approach enhances the accuracy and smoothness of reconstructed surfaces, improving the analysis of internal biological structures.
• The findings have implications for advancing medical imaging analysis and understanding tissue morphology.