Design And Implementat- Ion Of Stroke Diagnosis System

With the increasing living pressure, the rhythm of our life is getting faster and faster, the incidence of stroke is higher. It becomes Greater threat to human beings. How to diagnose a stroke and make a project to treat it in short time is very important. Traditional diagnosis depends on the experience of doctors. This diagnosis may cause large error. There is a software to calculate perfusion-diffusion mismatch in stroke designed by Stanford University. But doctors who use the software have no chance to change the empirical constants which may be different in the person in different race or region,because it is a black box. And it cost too much time to calculate the result in clinical diagnosis. All in all, doctors need a software which can change parameters and generate results for research and computing time is short enough for clinical diagnosis. Based on the foregoing overview and analysis, And according to use perfusion-diffusion mismatch principle,we design and implement a software system which is a white box software according to the function of the black box. And make foundation for us to accelerate our software with GPU for real diagnosis, I design and implement some modules, such as pretreatment of images, changing parameters, selection of arterial input function in stroke, etc. In the paper, I use motion correction and time correction for pretreatment, improve AIF selection algorithm to find AIF location, calculate the volume to compute the perfusion-diffusion mismatch.In this paper, some real cases are used to test the stroke diagnosis system, the result is positive;...