IntraCross: Cross-modality graph matching for intravascular sequence registration.

Area of Science:

• Cardiovascular Imaging
• Medical Image Analysis
• Biomedical Engineering

Background:

• Intravascular ultrasound (IVUS) and optical coherence tomography (OCT) are key imaging modalities for in vivo atherosclerosis assessment.
• Combining IVUS and OCT enhances the characterization of vulnerable plaques, crucial for predicting acute coronary events.
• Challenges in aligning IVUS and OCT data due to differing sensitivities and acquisition protocols hinder clinical application.

Purpose of the Study:

• To develop a novel framework, IntraCross, for robust and accurate registration of IVUS and OCT imaging sequences.
• To overcome limitations of existing methods, such as frame-level correspondences and multi-step processes, which lead to errors and instability.
• To enable simultaneous temporal and rotational registration, aligning with clinical workflows.

Main Methods:

• Proposed IntraCross, a graph matching framework utilizing partial landmark assignments for flexible, non-rigid registration.
• Extended 2D partial matching techniques to 3D sequences and incorporated a temporal prior for regularization.
• Developed a method for simultaneous temporal and rotational registration of IVUS and OCT data.

Main Results:

• IntraCross demonstrated high agreement with expert assessments in 77 vessels from 22 patients (Williams Index=1.1).
• The method significantly outperformed existing approaches in circumferential registration accuracy.
• Achieved stable and accurate alignment even in regions with limited visual information.

Conclusions:

• IntraCross offers a significant advancement in registering multimodal intravascular imaging data.
• The framework's ability to perform simultaneous temporal and rotational alignment improves clinical utility for atherosclerosis assessment.
• This approach enhances the characterization of vulnerable plaques, potentially improving patient outcomes in cardiovascular disease.