Research Of The Extraction Of Lung Tumor Based On CT Image

Lung cancer,known as primary bronchogenic carcinoma,is a very common primary malignant tumor occurring in the lungs.The diagnosis and treatment of lung cancer has become the most important thing in medical domain,and it is also very difficult to research.At present,medical image are playing a more and more important role in practical clinical treatment.The C T scan image of the lung is an important reference for the doctor to make a diagnosis and give the treatment plan.Especially to the patient who is being operated,it is very important to recognize the size,location and count of tumor before accurate operation.Therefore,computer aided diagnosis system(CAD)which analyses and process a large number of CT images of the lung has come into being.Based on this background,this paper focuses on the extraction of large tumors of the lung CT images in the computer aided system.The Lung C T images in this paper are all from the Lung Imaging Database Consortium library,which was published by the National Cancer Institute.Firstly,noise reduction and enhancement of the original lung C T image are needed to remove the noise information from the image and enhance the contrast of the information in the image at the same time.Then,extract the lung parenchyma from the C T image of the lung.There are some background areas such as heart,mediastinum,trachea and bronchus in lung CT images.These regions and tissues will interfere with the subsequent extraction of lung tumors,so it is necessary to remove them and get the mask images of lung parenchyma.Because of CT scan is a kind of scanning image,some C T images of lung may have le ft and right lung adhesion,so it is necessary to detect whether the mask image is adhesive or not.If adhesions,separate,otherwise proceed directly to the next step.In the process of obtaining pulmonary parenchyma mask,the information of pulmonary parenchyma boundary is usually lost,which is manifested as pulmonary parenchyma boundary depression.There are many details about the boundary of lung parenchyma,which is usually the information that doctors need to refer to in the course of diagnosis and treatment,so the boundary of lung parenchyma needs to be repaired.In the process of repair,this paper proposes a new algorithm based on diffusion algorithm,and compares the results with those of common algorithms.Finally,extract the lug tumor from the pulmonary parenchyma of the CT image.This paper is to improve the traditional random walk algorithm.During the process of extracting the lung tumor,the gradient information of the image is added to better preserve the edge information of the lung tumor.At the same time,the backup seed points were added to solve the problem that the final extraction of lung tumors was influenced by seed points.Compared with the traditional random walk algorithm,this method improves the accuracy of lung tumor extraction,and provides a better reference for doctors to make correct diagnosis and treatment.