Efficient patient modeling for visuo-haptic VR simulation using a generic patient atlas.

Area of Science:

• Medical Simulation
• Medical Imaging
• Virtual Reality

Background:

• Existing visuo-haptic virtual reality (VR) systems for percutaneous transhepatic cholangio-drainage (PTCD) training require faster patient modeling.
• Current modeling processes rely on generic patient atlases and require optimization for individual patient data.

Purpose of the Study:

• To develop a time-saving virtual patient modeling system for VR-based PTCD training and planning.
• To generalize generic patient atlases for new patient data using advanced modeling techniques.

Main Methods:

• Utilized a generic patient atlas with organ-specific models, transfer functions, and intensity image data.
• Implemented patient-specific, locally-adaptive transfer functions, multi-atlas segmentation, vessel filtering, and spline-modeling.
• Developed a full image volume segmentation algorithm.

Main Results:

• Achieved high median DICE coefficients for soft-tissue (0.98), liver (0.93), bone (0.82), skin (0.74), blood vessels (0.51), and bile vessels (0.48).
• Demonstrated remarkable time savings compared to traditional slice-wise manual contouring.
• Validated the segmentation algorithm on ten test patients and three reference patients.

Conclusions:

• The segmentation process is efficient and robust for upper abdominal puncture simulation systems.
• This advancement represents a significant step towards patient-specific training and planning systems in clinical settings.
• The developed system facilitates more effective and personalized medical training.