Precision TAVR quantification- AI-accelerated TAVR planning reduces assessment time by 80% in bicuspid aortic...

Area of Science:

• Cardiovascular Imaging
• Artificial Intelligence in Medicine
• Medical Device Technology

Background:

• Bicuspid aortic valve (BAV) stenosis presents unique challenges in transcatheter aortic valve replacement (TAVR) planning.
• Current methods lack validated automated algorithms for precise BAV anatomical assessment.

Purpose of the Study:

• To develop and validate Cardioverse, the first fully automated deep learning algorithm for BAV anatomical assessment in TAVR planning.
• To evaluate the algorithm's accuracy, efficiency, and performance across diverse BAV phenotypes.

Main Methods:

• A large-scale, multicenter retrospective study of 1,147 patients with BAV undergoing TAVR.
• Training and internal/external validation of the Cardioverse algorithm on patient cohorts.
• Assessment of segmentation performance, workflow efficiency (time, interactions, manual effort), and measurement accuracy.

Main Results:

• Cardioverse achieved exceptional segmentation performance (Dice coefficients >0.97).
• Significant workflow efficiency gains: 80% reduction in assessment time, 85% in user interactions, and 87% in manual effort (P < 0.001).
• High accuracy maintained across all BAV phenotypes (correlation coefficients >0.91 for critical measurements).

Conclusions:

• Cardioverse offers a validated, accurate, and efficient solution for pre-TAVR assessment in BAV patients.
• The algorithm reduces assessment time to under 5 minutes, addressing the need for rapid and reliable BAV evaluation.
• Its robust performance enhances TAVR planning consistency for diverse BAV phenotypes.