| Medical ultrasound imaging as an irreplaceable important means in modernmedicine clinical diagnosis, compare with CT and MIR, medical ultrasound imaginghas no radiation and side effect as well as safety, cheap, portable and can repeatedMedical ultrasound imaging as an irreplaceable important means in modernmedicine clinical diagnosis, compare with CT and MIR, medical ultrasound imaginghas no radiation and side effect as well as safety, cheap, portable and can repeatedinspection. The ultrasound elastography mode is a relatively new technology, whichusing ultrasonic imaging method for making estimating and real-time displaying localtissues’ elastic modulus distribution. Because many sign of disease is closely related totissues’ stiffness, and that the traditional medical imaging methods can not reflect thecharacteristics of physics tissues’ stiffness, so medical ultrasound elastographytechnology is widely attended and researched by the medical imaging field.This dissertation research real-time elastic imaging based on B mode’s grayscaleimages under the applied force by hand. For the input of continuous movement relatedB model’s grayscale images to preprocess by stacking frames before and after; Usestandard histogram template to dispose the full developed speckle noise; Use blockmatching algorithm to tracking the motions in energy structure images and use meshingto suppress the error, and get the tissues displacement distribution matrix of images;Use strain calculation to the tissues displacement distribution matrix and it will get thematrix of tissues’ strain distribution; And then process strain distribution of tissues withnatural3-spline interpolation to obtain smooth strain distribution; Process straindistribution image with false-color mapping, then use alpha mixing to stack it withoriginal B mode grayscale image and use median filter to make post-processing, finallyobtained the E model image output.The dissertation has analyzed and discussed the processes that were, for some ofthe algorithm of efficiency first, use the optimization design, in order to make algorithmcan get better imaging effect, such as: grayscale mapping of histogram, block matchingsimilarity estimate, the interpolation algorithm and post-processing method, etc. And propose to use CUDA paralleling algorithm in order to ensure the real-time performanceof the system imaging. Refresh, reanalysis and redesign the existing experimentalsystem, and implement the serial module of total business logic control on CPUmulti-threading and the parallel module of image processing on GPU. Finally theexecution’s efficiency and effect of system algorithm are improved, and the availabilityof improved elastic imaging methods in this paper is proved, and feasibility andadvantage of parallelism algorithm of real-time elastic image processing are proved. |
