A parallel computing approach for tracking of neuronal fibers

Abstract

A parallel computational accelerator based on the Cell Broadband Engine® Architecture was utilized to process medical image data of magnetic resonance scanners. Information on the orientation of white matter fiber tracts within the human brain was calculated. These results are visualized in three dimensions to help physicians involved with diagnostics for brain tumor surgeries. This parallel implementation demonstrates the potential of computational accelerators to significantly reduce the time needed for computationally demanding tasks in medical imaging.