SOLID: a novel similarity metric for mono-modal and multi-modal deformable image registration

Area of Science:

• Medical Imaging
• Computer Vision
• Computational Anatomy

Background:

• Medical image registration is crucial for clinical applications like image guidance and diagnosis.
• Existing methods face challenges in mono-modal and multi-modal scenarios, especially with varying contrasts.

Purpose of the Study:

• To introduce a novel, robust approach for mono-modal and multi-modal medical image registration.
• To present the Shape Operator based Local Image Distance (SOLID) similarity metric.

Main Methods:

• Proposed the SOLID metric, comparing second-order curvature information for image similarity.
• Utilized higher-order shape information for accurate local feature extraction.
• Implemented a variational image registration algorithm based on curvature matching.

Main Results:

• SOLID demonstrated favorable accuracy, precision, and robustness compared to state-of-the-art methods.
• Experiments covered mono-modal, multi-modal, and cross-contrast co-registration in diverse anatomical regions.
• The method excelled in highly challenging registration tasks.

Conclusions:

• The SOLID metric offers a robust solution for medical image registration across different modalities and contrasts.
• The proposed approach enhances the accuracy of registering smaller anatomical structures.
• SOLID shows significant potential for improving clinical applications reliant on precise image alignment.