Research On Edge Detection Method For Medical Images Using 3D Canny Operator

Medical image processing is an emerging interdisciplinary subject that integrates medicine and image processing,it plays a key role in medical research and diagnosis.Edge detection is the key to parallel boundary technology and its importance is self-evident.The complexity of human anatomy and the uncertainty of individual patho-physiology differences all add up new difficulties to edge detection.Canny operator,as the core operator of edge detection,is popular due to its excellent performance in edge detection.The purpose of the paper was to break down the idea that Canny operator was usually used in two-dimensional images and to achieve three-dimensional edge detection for MRI breast images.Taking three-dimensional Canny operator as an experimental reference and referring to the knowledge of the two-dimensional Canny operator and the 3D edge detection,the work was done in the following two aspects:(1)Based on the magnitude and direction formula of the three-dimensional gradient,the idea of three-dimensional non-maximum suppression was constructed by using vector summation and vector difference.Based on the assumption of non-maximum suppression,the other seven cases were proposed to construct a non-maximum suppression model.After the non-maximum suppression,the distribution of gradient values was plotted with a two-dimensional histogram.Finally,three adaptive threshold methods were used to complete the entire edge detection.(2)Subjective evaluation and numerical evaluation methods were used to evaluate the edge detection performance after the edge detection,subjective evaluation included displaying the result of edge detection in two and three dimensions.Based on the good image de-noising performance of the median filter,the Gaussian filter was replaced by a three-dimensional median filter and the edge detection results of the two filters were compared.In order to better serve the clinic,the method of the region growing wasintegrated with the improved 3D canny algorithm,the tumor regions of the breast were extracted and the tumor volumes were calculated.The fusion of the two algorithms additionally verified the performance of the improved algorithm and provided help for the diagnosis of the breast.