NON-INVASIVE METHOD FOR PATIENT-SPECIFIC VIRTUAL HEART BASED ON FIBER-FLUID MODEL

Abstract

In this paper, we present a non-invasive methodology in constructing a patient-specific virtual human heart based on fiber-fluid model. We applied digital image processing techniques on patient-specific MR images for obtaining the geometry information of human heart. The techniques include: acquisition, image visualization, image enhancement and segmentation. We incorporate cubic Hermite basis functions in our epicardium surface interpolation algorithm. We formularized a three-dimensional rule based fiber reconstruction mechanism to reconstruct the cardiac fiber architecture. A fiber model has been constructed which consists of 1,038 fibers with 371,239 fiber points. This model describes the ventricular threedimensional geometry. Immersed Boundary Methods is used in constructing fiber-fluid cardiac simulation. The simulation of early ejection (from 0ms to 0.5ms) for the Left Ventricle (LV) has been implemented in SGI workstation. Simulation results for cardiac fiber and blood flow are presented in three-dimensional (3- D). Open GL-based animated visualization programs are developed to serve three purposes: (1) to demonstrate the interpolation and rule-based fiber reconstruction process. (2) To visualize the simulation results of cardiac fiber and blood flow; (3) To analyse the dynamics of the epicardium fibers as well as the blood flow in the cavity of the LV.