Mitral valve models reconstructor: a Python-based GUI software in a HPC environment for patient-specific FEM structural analysis

A new approach in the biomechanical analysis of the mitral valve (MV) focusing on patient-specific modelling has recently been pursued. The aim is to provide a useful tool to be used in clinic for hypotheses testing in pre-operative surgical planning and post-operative follow-up prediction. In particular, the integration of finite element models (FEMs) with 4D echocardiographic advanced images processing seems to be the key turn in patient-specific modelling. The development of this approach is quite slow and hard, due to three main limitations: i) the time needed for FEM preparation; ii) the high computational costs of FEM calculation; iii) the long learning curve needed to complete the analysis without a unified integrated tool which is not currently available. In this context, the purpose of this work is to present a novel Python-based graphic user interface (GUI) software working in a high performance computing (HPC) environment, implemented to overcome the above mentioned limitations. The Mitral Valve Models Reconstructor (MVMR) integrates all the steps needed to simulate the dynamic closure of a MV through a structural model based on human in vivo experimental data. MVMR enables the FEM reconstruction of the MV by means of efficient scientific routines, which ensure a very small time consuming and make the model easily maintainable. Results on a single case study reveal that both FEM building and structural computation are notably reduced with this new approach. The time needed for the FEM implementation is reduced by 1900% with respect to the previous manual procedure, while the time originally needed for the numerical simulation on a single CPU is decreased by 980% through parallel computing using 32 CPUs. Moreover the user-friendly graphic interface provides a great usability also for non-technical personnel like clinicians and bio-researchers, thus removing the need for a long learning curve.