Study Of Lung Parenchyma Segmentation Algorithm Based On The Lung CT Image

Object:Lung cancer is the first killer that threatens the health of human.The lung CT image is the best method to diagnose lung cancer.But because the existence of the uneven gray level,the interference of noise and the partial volume cause the image fuzzy phenomenon,leading to errors in diagnosing and treatmenting.Computer aided diagnosis of lung diseases can improve the accuracy of diagnosis and efficiency of medical working.Computer aided diagnosis of lung diseases can improve the accuracy of diagnosis and efficiency of medical working.The lung parenchyma area is the important basis of clinical diagnosis and lung diseases.Segment the lung parenchyma accurately and fast in lung CT image can assist doctors to diagnose the lung diseases efficiently.Method:The proposed algorithm made full use of the local characteristics of the image.This paper put forward the SVM,Adaboost,Decision Tree by integrating the three classifiers and then used the largest voting principle to get the final segmentation results of lung parenchyma segmentation based on the detailed segmentation in pixels.To solve more complex lung CT images,this paper improved the segmentation algorithm on the basis of the above refined segmentation algorithm,which algorithm combined the fuzzy c-means clustering algorithm.Then labelled and merged the region pixels,finally obtained the lung parenchyma segmentation results.Results:This paper reduced the noise and bias field of lung parenchyma segmentation results,made full use of the local characteristics of image.The refined segmentation algorithm based on the SVM,Adaboost,Decision Tree classifier in superpixels for integration,finally realized the segmentation of lung parenchyma.Conclusion:The algorithm can effectively weaken the noise and the bias field of lung parenchyma segmentation results,reduce the complexity of image processing.And segmentation accuracy in the segmentation algorithm of integrated classifiers is higher than single classifier.To solve more complex lung CT images,this paper improved the superpixel refined segmentation algorithm and combined with the fuzzy c-means clustering algorithm.Finally the algorithm realized the lung parenchyma segmentation accurately and fast.The algorithm made full use of the gray and texture feature of images.The edge of image is well reserved.The algorithm realized the automatic,accurate lung parenchyma segmentation.The proposed algorithm can make full use of the local characteristics of lung CT images,decrease the complexity of the image processing,and finally realize the rapid and accurate segmentation of the lung parenchyma.The computer aided diagnosis of lung diseases is of great significance.