Research And Implementation Of The Cardiovascular Image Edge Detection,

Cardiovascular diseases have a high occurrence among the aged people. The angiogram image is a method of early diagnosis and treatment, which can eliminate the pain of the patient. Guided by the radiography, catheterization into the heart through femoral, contract medium is injected at the entrance of the coronary artery, and the heart and vessel are imaged. There is lots of information about vessel in the cardiovascular image. The images are captured, stored and processed for accurate diagnosis and guidance to clinic therapy. Three main areas of interest in the dissertation are: Cardiovascular Image Processing, Extraction of vascular edge-line and Measurement of the vascular diameter.Cardiovascular image is obtained under X-ray irradiation. The nonlinear transform and noise are taken into these images after X-ray attenuation, image enhancement, optical focus, amplification and A/D conversion. Image Techniques including smoothness, enhancement and sharpening are adopted to process original image in order to reflect the vascular information.Threshold smoothness, median filter and lowpass filter are used in the dissertation. Noise in the image is eliminated with the low-frequency filter; meanwhile the vascular edges are blurred. However, threshold smoothness and median filter are suitable to both elimination of noise and smoothness of image edge.The methods of image enhancement include histogram equalization, logarithm enlargement, grayscale transformation and space gray-level mapping s = T(r). Image enhancement not only improves contrast and visual effect of the vascular image, but also brings out image details and features.Direct Current parts in the image could be filtered out after different image sharpening methods, including space convolution, difference, space convolution sharpening and highpass filtering, thus the vessels and tiny tissues are extruded from the dark background.In order to extract vessels from the image and get the pure vascular image, we put forward the grayscale minimization of circle template and differential edge detection. At first, the vascular path is outlined with the grayscale minimization of circle template. After calculation of differential direction of vascular edge according to its local slope and Processing grayscale in its neighbor with differentiation, the edge points are fixed according to maximization of the differential value. Finally, the edge lines are determined after smoothing to the edge points using B-Spline function.At first, after thinning the binarized vascular image and extraction of the vessel and its conjunctions layer by layer with template, the centerline of vessel is obtained to calculate the diameter of the vessel. According to the centerline data, vertical line of simulating line to the local centerline is also determined. Finally, the distance, which is the diameter of vessel at one certain point, is calculated from one point to another that are the cross-points between the vertical line and two edge lines. However, diameter (Di) and diameter difference (Dci) could not be applied to measurement of the degree of vascular straitnessand clinic diagnosis, therefore, the relative difference of vessel ( Dpi ) defined as Dpi = (Dci D) 100 % ( D is the locally average diameter) is expressed to be one clinic diagnosis criterion.