Flip-avoiding interpolating surface registration for skull reconstruction.

Area of Science:

• Medical Imaging
• Geometric Modeling
• Computer-Aided Surgery

Background:

• Skull reconstruction is critical for craniofacial surgery, forensics, and anthropology.
• Current methods use soft constraints, leading to significant reconstruction errors.
• Existing techniques struggle with defective skulls and outliers.

Purpose of the Study:

• To develop a novel geometric skull reconstruction method using hard constraints.
• To improve accuracy and robustness compared to existing skull reconstruction techniques.
• To investigate the impact of reference model selection on reconstruction accuracy.

Main Methods:

• Non-rigid registration of an interpolating reference surface with hard constraints.
• Implementation of a flip-avoiding algorithm to manage conflicting constraints.
• Incorporation of symmetry constraints for reconstructing defective skulls.
• Evaluation of robustness against outliers like CT artifacts.

Main Results:

• The proposed method achieves significantly lower reconstruction error than existing approaches.
• Demonstrated superior accuracy and robustness, especially for defective skulls.
• Showcased effectiveness against severe outliers and improved performance over thin-plate spline for resampling.
• Analysis of reference model selection implications for reconstruction accuracy.

Conclusions:

• The novel geometric method offers a more accurate and robust solution for skull reconstruction.
• The approach effectively handles complex cases, including large defects and imaging artifacts.
• This work advances skull reconstruction for applications in surgery, forensics, and anthropology.