Omni-Directional Shortest Distance Algorithm by Complete Parallel-Processing Based on GPU Cores

Abstract

Abstract:We propose a new algorithm for calculating the 3D omni-directional minimum distance from the cavitron ultrasonic surgical aspirator (CUSA) tip to blood vessels. The distance is selected from many shorter distances calculated by the GPU cores for many pixels. First, we use z-buffering (depth buffering) as the classic matured function of the GPU to effectively obtain depths (the distances to blood vessels) corresponding to many pixels. Second, we calculate the Euclidean distance from the scalpel tip to the closest z-values of the depths by multiple GPU cores for all pixels. The many pixels are prepared within a cubic region overlapped by six rectangular parallelepipeds along the +X, -X, +Y, -Y, +Z, and -Z axes centered at the CUSA tip. In this algorithm, all cores are not distinguished within six cubes. Finally, we evaluate the algorithm performance with regard to calculation time and visual reality using an inexpensive GPU (GTX950). Our experimental results show that the calculation time is twice as fast, and the visual reality is also improved.Key words: CUSA (cavitron ultrasonic surgical aspirator), DICOM (digital imaging and communication in medicine), GPGPU (general-purpose graphics processing unit), STL (STereo-lithographies), Z-buffering (depth buffering).