Segmentation of 3D vasculatures for interventional radiology simulation

Area of Science:

• Medical Simulation
• Medical Image Analysis
• Radiology

Background:

• Interventional radiology training is transitioning to simulation for risk-free practice.
• Accurate segmentation of anatomical structures is crucial for realistic surgical simulations.
• Current methods may be limited by human error and artifacts.

Purpose of the Study:

• To develop a generic, fast, and precise segmentation framework for virtual anatomies.
• To create patient-specific simulations from CT/MRA images, including various pathologies.
• To reduce subjective human influence in the segmentation process.

Main Methods:

• Developed a segmentation framework with two novel methods: region model-based level set and hierarchical segmentation.
• Compared proposed methods against the open-source ITK-SNAP application.
• Analyzed subjective human influences like inter-observer variability and aliasing artifacts.

Main Results:

• Successfully applied the segmentation techniques to generate a diverse database of anatomies with pathologies.
• Demonstrated the framework's capability for creating patient-specific virtual anatomies.
• The developed methods offer a robust approach to anatomical segmentation for simulation.

Conclusions:

• The proposed segmentation framework provides fast and precise virtual anatomies for interventional radiology simulation.
• This approach enhances the realism and effectiveness of computer-based training.
• It addresses limitations of subjective human influence in segmentation for medical simulation.