Parallel processing approach for finite difference modelling of human thorax as a volume conductor

The finite difference element method (FDM) provides a straightforward and easy way to construct and solve accurate, patient-tailored models of the human thorax as a volume conductor. However, the usability of the method is decreased by a computationally slow iterative method to solve the source-field problem. Here, the authors present a method to implement FDM thorax model into a MASPAR massively parallel array processor. FDM elements are based on a rectangular grid that is used throughout the volume. This grid was mapped directly into the parallel processing array in such a way that each processor acts as a computational element. The stack of elements at a particular. FDM grid position in all slices were loaded into the memory of a processor at the same grid point in the processor array. Thus the iterative process was mapped from the single-element mode of the conventional computers into a single-slice mode. Each slice can have 16000 elements, which are solved simultaneously, decreasing the calculation time accordingly. The array processor will be used to obtain solutions for a very accurate thorax model featuring the anatomy of the Visible Human Man.