Biomechanically Constrained Surface Registration: Application to MR-TRUS Fusion for Prostate Interventions

Area of Science:

• Medical image analysis
• Computer-assisted interventions
• Computational biomechanics

Background:

• Surface-based registration is crucial for image-guided interventions.
• Missing data in medical images, common in real-time modalities like ultrasound, presents challenges in establishing correspondences and extrapolating deformation fields.
• Existing methods struggle with data incompleteness, impacting registration accuracy.

Purpose of the Study:

• To develop a novel non-rigid registration method to address challenges posed by missing data in surface-based registration.
• To improve the accuracy and reliability of image registration in the presence of partial or incomplete surface data.
• To validate the proposed method in the context of prostate interventions.

Main Methods:

• A novel non-rigid registration algorithm, termed GMM-FEM, was developed.
• A probabilistic framework using a Gaussian mixture model (GMM) was employed for establishing correspondences with partial surface observations.
• Biomechanical prior knowledge, integrated via a finite element model (FEM), was used to extrapolate and constrain the deformation field.

Main Results:

• The GMM-FEM method demonstrated a significant reduction in target registration error (TRE) for missing data up to 30%, achieving a mean TRE of 2.6 mm.
• The algorithm performed comparably to or better than state-of-the-art surface-based registration techniques when full segmentations were available.
• Robustness analysis confirmed GMM-FEM as a practical and reliable solution for surface-based registration.

Conclusions:

• The GMM-FEM approach effectively handles missing data in surface-based registration, a common issue in image-guided interventions.
• This method offers improved accuracy and robustness compared to existing techniques, particularly in scenarios with incomplete imaging data.
• GMM-FEM presents a viable and reliable solution for enhancing precision in medical interventions requiring accurate image registration.