Aortic valve leaflets motion trajectories tracking by using echocardiography data

A new approach is proposed for evaluating the motion trajectories of aortic valve leaflets throughout the cardiac cycle based on echocardiography data, aimed at subsequently determining their mechanical properties. The non-invasive nature of this method presents a significant advantage over existing experimental techniques for assessing the mechanical properties of soft tissues, which often involve invasive procedures and are limited by sample availability and ethical considerations. In this study, a software module was developed to track the movement of control points marked on the aortic valve leaflet, using computer vision algorithms applied to echocardiography video data. The program utilizes a neural network based on the YOLOv8 library for semantic segmentation and object detection, enabling precise identification and real-time tracking of the valve leaflets. The software was initially validated on model data to ensure the accuracy and reliability of the tracking algorithm. It was then applied to real echocardiography data, successfully reconstructing the trajectories of the aortic valve leaflets throughout the cardiac cycle. These trajectories can be used in conjunction with mathematical models to determine the mechanical properties of the valve leaflets using optimization methods. The proposed method represents a non-invasive, efficient, and accurate means of assessing the dynamic behavior of aortic valve leaflets, potentially contributing to improved diagnosis and treatment planning for patients with aortic valve diseases. Further development of this approach may lead to a deeper understanding of the valve's mechanical characteristics and enhance surgical outcomes.