Research On Image Segmentation Algorithm Of Pulmonary Nodule Based On Fuzzy C-Mean

According to reports,lung cancer is gradually becoming the first cancer that threatens human life.Early lung cancer is manifested in the form of pulmonary nodules.Early detection and treatment are the most effective methods for treating lung cancer.The Computer Aided Diagnosis(CAD)system of lung tumors assists in the detection of lesions through medical image processing technology,and improves the accuracy of diagnosis.The key issue of CAD is to achieve correct and rapid segmentation of diseased tissue in medical imaging.The main method of collecting lung images is to take a computed tomography(CT)image.Due to the imaging principle of CT images,there is a large amount of noise and gray unevenness in the CT images of the lungs.Accurate segmentation of lung CT images becomes a key issue.At present,image segmentation based on fuzzy C-means clustering is widely used,but there are still problems of incomplete segmentation and low segmentation efficiency in the application of lung nodule segmentation.This paper mainly studies the CT image segmentation algorithm model of pulmonary nodules based on fuzzy C-means clustering.The main contributions are as follows:1)For the vascular adhesion type pulmonary nodules segmentation algorithm,there are the characteristics of pulmonary nodules and vascular segmentation.An adaptive fuzzy C-mean clustering(ARFCM)lung nodule CT image segmentation algorithm model is proposed.Since the neighborhood pixels may be noise points or edge pixels,not all neighborhood information will have a positive correlation effect on the central pixel points,so a reference mechanism for redefining the neighborhood window pixel information needs to be redefined.In this paper,according to the characteristics of the pixel and the gray fluctuation of the eight neighborhood pixels of the central pixel,the weighting factor is introduced into the objective function,the weighting factor is adaptively updated,and the different fuzzy items are selected by the weighting factor.If the central pixel is a noise point or an edge point,the objective function selects a blur factor suitable for the situation based on the weighting factor.The experimental results show that the method is superior to other typical algorithms in segmenting lung nodules.2)For the lung nodule segmentation algorithm to ignore the inherent structural information of the whole CT map,a fuzzy C-means clustering lung nodule CT image segmentation algorithm model(U-MRF and Fuzzy C-mean,U-MFCM)is proposed.Clustering.The method mainly uses the Markov random field to integrate the internal structure information of the pixel into the membership matrix of the fuzzy C-means cluster.Firstly,the CT image of the lung is initially segmented by fuzzy C-means clustering to obtain the initial membership matrix of the central pixel.Then,the neighborhood pixels of the central pixel are processed by the Markov random field to obtain the membership category label of the neighborhood pixel,and the membership degree of the central pixel is calculated by using the spatial neighborhood relationship between the central pixel and the neighboring pixel.matrix.Finally,the fuzzy membership matrix and the membership matrix of Markov random field are superimposed and updated to obtain the final segmentation result of lung nodules.The experimental results show that the method solves the problem of over-segmentation of images by traditional segmentation methods and improves the segmentation efficiency of lung nodules.