A Computational Pipeline for Patient-specific Prediction of the Post-operative Mitral Valve Functional State.

Abstract

In the present work, we established a patient-specific mitral valve (MV) computational pipeline based strictly on standard-of-care preoperative imaging data to quantitatively predict the post-repair MV functional state. First, we developed a finite-element model of the full patient-specific MV apparatus by quantifying the MV chordae tendinae (MVCT) distributions from five CT-imaged excised human hearts and segmenting the MV leaflet geometry and identifiable MVCT origins from three patients' preoperative 3D echocardiography images. To functionally tune the patient-specific MV mechanical behavior, we simulated preoperative MV closure and iteratively updated the leaflet and MVCT pre-strains to minimize the mismatch between the simulated and target end-systolic geometries. With this fully calibrated MV model, we simulated undersized ring annuloplasty (URA) by defining the annular displacement using the ring dimensions. In all cases, the postoperative geometries were predicted to within 1 mm of the target, and the MV leaflet strain fields demonstrated very good global and local correspondence. Moreover, our model predicted increased posterior leaflet tethering after URA in a recurrent patient, which is the likely driver of long-term MV repair failure. Such studies are crucial since MV repair outcomes remain unpredictable, yet patient-specific optimization techniques are profoundly limited. With this pipeline, we are able to predict postoperative outcomes from preoperative clinical data alone, and thus lay the foundation for quantitative surgical planning, tailored patient selection, and ultimately, a more durable repair.