Robust model-based 3d/3D fusion using sparse matching for minimally invasive surgery.

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Image Fusion

Background:

• Minimally invasive and transcatheter procedures are replacing classical surgery, necessitating advanced intra-operative guidance.
• Current intra-operative imaging (3D C-arm CT, C-arm fluoroscopy) offers limited soft-tissue visualization and often requires harmful contrast agents for reliable cardiac anatomy assessment.
• Pre-operative CT provides high-quality anatomical data but cannot be directly mapped to the dynamic intra-operative environment.

Purpose of the Study:

• To develop a novel sparse matching approach for fusing pre-operative CT with non-contrasted, non-gated intra-operative C-arm CT.
• To enable the creation of high-quality, patient-specific anatomical models for intra-operative guidance in minimally invasive procedures.
• To improve the visualization of cardiac anatomy during procedures without the need for contrast agents.

Main Methods:

• Utilized robust machine learning and numerical optimization techniques for image fusion.
• Developed a sparse matching approach to map pre-operative CT data to the intra-operative C-arm CT imaging environment.
• Validated the method through extensive quantitative experiments.

Main Results:

• The proposed model-based fusion approach achieved an average execution time of 2.9 seconds.
• The accuracy of the fused images falls within expert user confidence intervals.
• Successfully generated high-quality patient-specific models from pre-operative CT and mapped them to the intra-operative setting.

Conclusions:

• The novel sparse matching approach effectively fuses pre-operative CT and intra-operative C-arm CT data.
• This method provides accurate, high-quality anatomical models for guiding minimally invasive procedures without contrast agents.
• The technique offers a faster and safer alternative for intra-operative cardiac anatomy assessment.