Automated cardiac time interval measurement for Modified Myocardial Performance Index calculation of right ventricle

摘要

The Modified Myocardial Performance Index (Mod-MPI) has sparked great interest as a parameter for fetal cardiac function assessment. However, measurement of this index requires expertise and its clinical application might be limited, owing to its poor repeatability. Research groups have been investigating left Mod-MPI (that is, Mod-MPI from left ventricle valve events), and an automated algorithm has been developed for left Mod-MPI calculation in our previous study. Right MPI is also important as it becomes abnormal earlier than left MPI in some pathologies; however, for use across the gestational age spectrum, it requires two-image acquisition. This paper presents an automated method to detect valve movements during atrioventricular outflow and ventricular inflow and to further calculate the time intervals required for right MPI calculation. Ninety pulsed-wave Doppler ultrasound images of the right ventricle in fetuses, forty-five showing outflow and forty-five inflow, were analyzed to automatically detect the valve clicks generated by tricuspid valve movement in inflow waves, and pulmonary valve movement in outflow waves. The morphological characteristics of waves were combined with intensity information to locate clicks. This automated method can detect valve movement events with a high positive predictive value (96.20-98.96%) and sensitivity (97.95-100.00%), using manual annotation from an expert ultrasonographer as the gold standard for evaluation.