Research Of DSA Vessel Subpixel Stenosis Detection And Quantification Based On Zernike Moments

With the acceleration of the pace of life and life stress,and unhealthy habits,the worldwide prevalence rate of cardiovascular disease has been presenting a trend of explosive growth.As an early manifestation of cardiovascular and cerebrovascular diseases,accurate detection and quantification of stenosis has become an important factor in prevention,diagnosis and treatment.DSA can clearly present the stenosis distribution of each major artery due to the characteristics of noise suppression and boundary enhancement;and it is the primary reference object for the detection and quantification of arterial stenosis.Referring to the characteristics of DSA in high precision and easy-to-measure,and integrating with clinical diagnosis requirements in high precision and high sensitivity,a subpixel detection and quantification method for DSA will be proposed in this paper.In this paper,requirements for the realization of subpixel stenosis detection are analyzed as follows: the precise positioning of vessel axis,the precise extraction of vessel boundary,and the diameter measurement method in accordance with the changes of vascular morphology.Therefore,the algorithm is separated into three steps for stenosis detection and quantification referring to abovementioned requirements.The phase of vascular axis extraction;by using the adaptive multiscale filtering based on Hessian matrix eigenvalues and eigenvectors,it effectively strengthen the tubular structure organization,restrain non-tissue tubular structure organization and improve the discrimination of arterial vessels and its surrounding muscle tissue.Regarding to the enhanced image object,a series of morphological operations are used to realize the accurate extraction of the axis.The phase of boundary extraction;by comparing and analyzing a variety of boundary extraction algorithms,this paper proposes the method of Zernike moments to realize the extraction of boundary pixels.The boundary extraction algorithm has better anti-noise performance,owing to the rotation invariant property of Zernike moments,and it can effectively reduce the computational complexity and improve the performance of the algorithm by using Zernike template.Through the comparison experiment,it verifies the influence of parameter value on the running result of algorithm.At the same time,the experimental results prove that algorithm has high performance from three aspects of noise immunity,extraction precision and operation time,which is the preferred algorithm within boundary extraction stage.The phase of vessel diameter measurement;with the enlightenment from conception of elastic sphere dynamic expansion and some improvement,the paper proposes an algorithm of blood vessel diameter based on the complex shape balance sphere.For the complex structure of blood vessels nearby stenosis,and the characteristics of large morphological variations,the algorithm constructs a balance ball under complex structure by using the characteristic of vascular tissue's morphological structure,which could receive a vascular diameter to be more in line with the actual trend and the structure through the angle of ball radius and its corresponding.By means of the comparison experiment of blood vessel diameter measurement algorithm,the proposed method in this paper is closer to the manual calibration value than traditional method.Finally,the paper provides comparison among typical artery stenosis patients' DSA samples.According to the experimental results,the proposed subpixel stenosis detection algorithm based on Zernike moments is in accordance with permissible error range of clinical actual detection,which has higher accuracy and much better robustness.