Using Anatomic Intelligence to Localize Mitral Valve Prolapse on Three-Dimensional Echocardiography.

Area of Science:

• Cardiovascular Imaging
• Medical Ultrasound Technology
• Surgical Planning

Background:

• Accurate mitral valve prolapse (MVP) localization is critical for effective surgical planning.
• Three-dimensional transesophageal echocardiography (3D-TEE) improves visualization but interpretation is expertise-dependent.
• Manual topographic mapping of the mitral valve is accurate but time-consuming and labor-intensive.

Purpose of the Study:

• To evaluate the efficacy of Anatomical Intelligence in Ultrasound (AIUS), a novel computer-learning technique, in improving MVP localization.
• To assess whether AIUS enhances accuracy and efficiency in MVP localization across different operator experience levels.
• To compare AIUS performance against manual segmentation methods.

Main Methods:

• Retrospective analysis of 3D-TEE images from 90 degenerative MVP patients.
• Echocardiographers (2 experts, 4 intermediate) performed both AIUS and manual segmentation.
• Comparison of AIUS and manual segmentation results with surgical findings.

Main Results:

• AIUS significantly improved nonexpert accuracy (83% to 89%) and sensitivity (60% to 90%) compared to manual segmentation.
• AIUS enhanced accuracy in specific segments (A3, P1) challenging for manual tracking.
• AIUS substantially reduced image analysis time for both experts (9.9 to 1.9 min) and nonexperts (13 to 5.0 min).

Conclusions:

• Echocardiographic assessment of mitral valve pathology is influenced by operator experience.
• AIUS, a semiautomated algorithm, demonstrably improves the accuracy and efficiency of MVP localization.
• AIUS is particularly beneficial for less experienced operators, standardizing diagnostic capabilities.