A hybrid framework for registration of carotid ultrasound images combining iconic and geometric features

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Cardiovascular Research

Background:

• Stroke is a leading cause of death, with carotid atherosclerosis being a primary cause of ischemic stroke.
• Accurate monitoring of carotid plaque size is crucial for preventing stroke complications.
• Patient head movement during imaging introduces variability, hindering accurate serial monitoring.

Purpose of the Study:

• To develop and evaluate a novel hybrid image registration technique for carotid ultrasound.
• To improve the accuracy of monitoring carotid atherosclerosis progression.
• To address the challenge of patient head movement during serial imaging.

Main Methods:

• A hybrid registration method combining feature-based (Iterative Closest Point) and intensity-based algorithms.
• Utilized a rigid transformation model with Mutual Information (MI) similarity and Powell optimizer.
• Quantitative evaluation using visual information fidelity, structural similarity index, correlation coefficient, and MI.

Main Results:

• The proposed hybrid registration effectively suppresses global patient movements in time-series carotid ultrasound images.
• Quantitative metrics demonstrated the accuracy and robustness of the novel registration technique.
• Qualitative visual examination confirmed the successful alignment of registered image pairs.

Conclusions:

• The novel hybrid registration technique offers a significant advancement for monitoring carotid atherosclerosis.
• Accurate image registration is vital for precise disease characterization and quantification.
• This method has the potential to enhance stroke prevention strategies through improved imaging analysis.