Enhancement Of Renal Artery Based On Matched Filter

With the increasing incidence of renal cell carcinoma, renal tumor resection operation has become the preferred option for treatment of localized renal cell carcinoma. In order to ensure the quality of life of patients after operation, partial nephrectomy surgery of segmental renal artery clamping has caused great interest in the clinical and research fields. Due to the clinical demand of the preoperative plan, renal artery, especially those renal artery branch terminals near or into the renal tumor with smaller diameter, need to be clearly displayed, which can help clinician accurately judge the renal artery branches that supply blood for tumor area and rationally choice the blocked renal artery branches during operation. However, commonly-used display imaging tools can not meet the above requirements. Therefore, the renal artery and its branches in renal CTA(Computed Tomography Angiography) images were effectively enhaced, which will become the key pre-processing step before direct volume rendering.To meet the above clinical needs, an enhancement method of renal artery based on matched filter has been investigated in this thesis. Firstly, the filter with three-dimensional vascular tubular feature is designed, and is extended to the three-dimensional filter bank with tunable scale and orientation. Meanwhile, the selection rules of parameters in matchced filter bank are given. Then, we conceive vessel enhancement as a filter process and realize the enhancement of renal artery in CTA image using the matched filter bank. Meanwhile, the basic enhancement method of renal artery is proposed.Considering that the renal cortex of CTA has a high gray value and will have a strong response in the process of vessel enhancement using basic enhancement method, we improve the response after filtering by means of principal component analysis and ellipse fitting and design the vascular description operator, to effectively suppress the error response generated by the renal cortex and make renal artery and its branch terminals with small diameter be clearly visualized after enhancing. From the two point of view of reducing the number of voxels to be processed and optimizing the program, the strategy of improving the efficiency of the algorithm is given.In this thesis, we firstly give a quantitative analysis using synthetic data, compared with the commonly applied Frangi vessel filter. On the basis of the above research, we further test the clinical renal CTA images and give the comparison of the results of direct volume rendering to initially illustrate the effectiveness of the method proposed in this paper.