Top Next: Basics of Heart Anatomy

1. Introduction

By exploiting the power of the parallel technologies currently available, it is possible to simulate both natural phenomena and experiments that would cost vast amounts of money, or those which are ecologically problematic or dangerous for humans. In this paper we shall focus on the verification and practical application of our research work in the area of the simulation of heart cooling during surgery. Computer simulations in medicine are less expensive and faster than experimental studies. Performing in vivo experiments and measurements is often difficult, dangerous or even impossible, while simulation can provide insights into physiological processes without any harm. High performance parallel computers could lead to the improved analyses of various methods of heart cooling during the hypothermic cardiac arrest induced during open heart surgery, the prediction of temperature elevation following coronary artery occlusion, the interpretation of electrical cardiac signals and many other medical applications.

High performance parallel computers provide the computational rates necessary for advanced biomedical computing [1]. During the induced hypothermic cardioplegic cardiac arrest in open heart surgery the human body and the heart have to be cooled appropriately in order to slow down their vital functions [2]. Different levels of systemic-body hypothermia are used to lower its metabolic requirements; for even better cardiac cooling a method of topical cooling is used [3]. The human heart is an irregularly shaped three dimensional object. In vivo temperature measurements are limited to a few test points. Some initial results on the simulation of heart cooling have been presented in [4]. The EFD (Explicit Finite Difference) method that imposes a regular grid on the physical domain is being used [5]. In most scientific computing applications a physical system is represented by mathematical model. The continuous physical domain has to be replaced by a discrete representation that is suitable for a numerical solution. Usually, the physical domain is partitioned into many small subdomains by imposing a grid. Solving the mathematical model over a discretised domain involves obtaining the values of certain physical quantity at every grid point for each time interval. A grid point is influenced only by the surrounding grid points, usually with a simple local rule. Each calculation step gives new values of the physical quantity for the next interval of the real time. The derivative of an unknown temperature T is approximated by the ratio of the difference in T at adjacent grid points, to the distance between the grid points. Results of the similar work, but in the area of the spatial and temporal electrical activity of the heart, were presented in [6].

To get the three-dimensional heart model, 156 slices of VHD (Visible Human Dataset, National Library of Medicine) [7] were used in the region of heart position (the X-Y plane in Z direction representing the heart axis, and originating from the apex toward base. The spatial mesh was thus discretised into millimetre range resulting in 146x152x156 grid points. It was necessary to guarantee the exact overlapping of all slices because of the integrity of the model. Before digitalisation two reference points on every picture were marked in order to adjust all the data. Besides that, several impurities and other errors in the pictures(photographs) had to be discarded. Finally, the edges between different substances had to be determined. To perform all the above actions, a standard software package for digital picture processing was used, and more specialised custom programs were also implemented.

Some basic data on heart anatomy are given in the next section. Then, in the next two sections, a procedure for the spatial-heart model derived from VHD, including slice preparation and heart tissue distinction, is described. Finally, the detailed description of the 3-D editor is given and some final model results are shown. In the concluding section, some practical implementation problems and directions for future work are given.


Top Next: Basics of Heart Anatomy