Ultrasound registration: A review

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computer Vision

Background:

• Ultrasound imaging is safe, cost-effective, and real-time, widely used in clinical and scientific settings.
• Ultrasound images present unique challenges due to noise and artifacts, complicating image analysis and registration.
• Accurate image registration is crucial for various applications, including surgical guidance and image-guided interventions.

Purpose of the Study:

• To provide a comprehensive review of monomodal image-based ultrasound registration algorithms.
• To introduce the theoretical and practical aspects of ultrasound-to-ultrasound image registration.
• To summarize the existing literature and validated registration systems for specific anatomical regions.

Main Methods:

• Theoretical framework: organized by registration components (preprocessing, similarity metrics, optimizers).
• Classification of methods: elastic (non-rigid) versus inelastic (matrix) transformations.
• Practical review: organized by anatomical regions, discussing validated registration systems.

Main Results:

• Detailed overview of theoretical components in ultrasound image registration.
• Categorization of registration methods based on transformation types.
• Presentation of practical, anatomy-specific ultrasound registration systems and their validation.

Conclusions:

• Ultrasound-to-ultrasound registration is a critical yet challenging field.
• A thorough understanding of theoretical components and practical systems is essential for advancing ultrasound image analysis.
• This review serves as a foundational resource for researchers and practitioners in medical imaging and biomedical engineering.