Research On Automation Evaluation Of Cerebrovascular Disease

As the largest threat to human's heath nowadays, cardiovascular diseases have drawn worldwide attention. Along with CAD widely used in the field of medicine, Scholars devote to the research on application of CAD in diagnosis of vascular lesions which mainly focus on vascular enhance, vascular segmentation and recognition. Mostly, it is needed the doctors to review the vascular image after processing.It is devoted to implement the automation evaluation of vessel disease. Thesis analyzed vascular enhance which based on hessian matrix in multi-scale with different resemblance measurement function, extracted the orientation field use Gabor filter and compared with which was extracted by hessian matrix eigenvalue. The thesis improved the traditional recognition algorithm of vascular node, the improved algorithm did not have restrict request one pixel wide on skeleton which was just needed to be in good continuity, many thin algorithm such as Zhang, Hilidtch.etc worked well with new recognition algorithm. Thesis measured the vascular diameter by distance transfer technique which was taken as the most important concept to evaluate cerebrovascular diseases. The diameter is also used as the self-adaptive Gabor filter's scale parameter.After the research on a plenty of cerebrovascular disease samples, thesis proposed a suite of cerebrovascular disease evaluation system which could be separated into two subsystem. One is the evaluation principal to cerebrovascular disease, consists of Minus Diameter, Stenosis, Variation, Incomplete, the other is the assess to the evaluation principal including Precision and Recall. Thesis designed a new cerebrovascular disease auto evaluation algorithm based on skeleton segmentation under the evaluation system, and had a good performance. The propose of the cerebrovascular disease evaluation system established very important basis of cerebrovascular disease automation evaluation.