Thresholding Segmentation Errors and Uncertainty with Patient-Specific Geometries.

Area of Science:

• Medical simulation
• Computational geometry
• Medical imaging

Background:

• Computer simulations offer virtual training when hands-on experience is limited.
• Accurate medical simulations require patient-specific geometries derived from medical imaging.
• Processing complex patient geometries often involves simplifications, compromising specificity.

Purpose of the Study:

• To investigate the impact of user choices on patient-specific 3D model generation.
• To assess the variability in geometric outcomes based on user input.
• To explore methods for maintaining geometric fidelity in medical simulations.

Main Methods:

• Utilized medical imaging data to create complex, patient-specific geometries.
• Employed numerical algorithms for 3D surface processing.
• Assessed the influence of multiple users' processing decisions on the final geometry.

Main Results:

• The creation of patient-specific geometries from medical images is sensitive to user input.
• Geometric dimensions of the 3D models significantly vary based on user choices.
• Simplifications, while common, reduce the patient-specificity of simulation models.

Conclusions:

• User interaction critically influences the accuracy and specificity of patient-derived 3D models.
• Standardized protocols may be needed to reduce variability in geometry processing.
• Further research is needed to achieve truly patient-specific simulations without compromising fidelity.