多模态虚拟心脏可视化方法分析-模式识别与智能系统专业论文.docxVIP

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 -II -II- Abstract Contraction and relaxation of the heart which provide the impetus for the bodys blood circulation is the bodys source of power. However, cardiovascular diseases have been the threat to health of human being. Due to the high cost and low efficiency of clinical study on the heart, the research of the virtual heart can develop rapidly. The research of the virtual heart has recently focused on the visualization of the cardiac anatomical data and the modeling of cardiac electrophysiological principle. The rendering speed of the CPU-based visualization is slow when using volume rendering to visualize the cardiac anatomical data as the amount of the cardiac anatomical data is huge, so it is difficult for the users real-time interaction. Compared to CPU, there are more ALU on GPU and GPU has also achieved the basic operation of the image, so GPU is faster than CPU on the graphics and image process. Visualization of the cardiac anatomical data using GPU-based ray casting algorithm is fast and can achieve real-time interaction. The design of 1D histogram based transfer function is firstly introduced to make the heart tissue distinguish. Opacity modulation based on statistics of the specific gray level is then presented to observe the interior subtle cardi ac tissues clearly. 2D histogram based transfer function using gray level and gradient magnitude is implemented to get clear contour of heart and to enhance boundaries of cardiac tissues. Transfer function using the mean and standard deviation is presented for the extraction of the boundary of heart tissues. 4D visualization is a difficulty in the field of scientific computing visualization, time histogram is used to divide the 4D data of cardiac electrophysiology into different time periods according to the characteristics of 4D data of cardiac electrophysiology. The adaptive transfer function is designed for each time period and GPU-based ray casting algorithm is improved