GPU-Accelerated High Performance Computing System: Application to Real-Time Functional Magnetic Resonance Imaging

• Main Authors: Wei-jhong Siao, 蕭為中
• Other Authors: Teng-yi Huang
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/e448bc

碩士 === 國立臺灣科技大學 === 電機工程系 === 99 === This study attempts to build a real-time functional magnetic resonance imaging system (rtfMRI) to monitor blood-oxygen-level-dependent (BOLD) signal during fMRI experiment. To detect the BOLD signal change, a Gaussian filter and a general linear model analysis were performed on MRI images immediately subsequent to image acquisitions. A graphic processing unit (GPU) with massively parallel computation kernels was used to accelerate the image processing [i.e. gaussian filter and general linear model analysis (GLM)]. The GPU program was compatible to MATLAB environment through a communication interface of MATLAB and C language. Using GPU computation, the analysis of rtfMRI could be accomplished in less than 1 second in a conventional personal computer.