A Study On ASM-Based Lung Edge Extraction

Lung cancer,according to the information statistics,is one of the most common cancers worldwide,but also a higher cancer mortality rate.The cancer incidence rates continue to rise in our country since the 1970s,and lung cancer has borne the brunt of cancer death.To diagnosis of cancer and improve the survival rate of lung cancer patients accurately,easily and quickly is the focus of international studies but also a long-standing medical profession.CT examination of lung cancer has become an indispensable means with the development of medical imaging techniques such as CT.The lung physiology to structure information of patients could be observed directly through the CT sweep surface data.The information of CT slices increased greatly as the increasing accuracy of CT and the workload of doctors' increased significantly by the way.As the result of that condition,the likelihood of misdiagnosis increased,diagnostic results are simply not the same because of the the ability and experience from medicals.The excellent image processing tools,therefore,to accurately determine the lesion is very important.Medicals could work better with the help of Computer-aided detection and diagnosis of medical images(CAD),reducing the workload of doctors,meanwhile,can assist doctors to make a final judgment which was needed evolved.To quickly and accurately automatically segmenting lung area,the crucial significance method is one of computer-aided diagnosis of pulmonary critical technology.Our topic "Lung research based on contour extraction algorithm ASM" followed the several study steps:The first step,image enhancement,morphology,threshold segmentation,and etc.are used to prepare the work for the ASM training set to get the parenchyma profile from healthy human lung.In order to make this work easy and fast,we set up a GUI platform to read DICOM images and stored in the appropriate image feature points.Afterwards,depth study in the active shape.Selection of a group of typical lung parenchyma contour shape as the training set,using the outline feature point to express the lung parenchyma boundary alignment training set shape,get training set statistics.Use the gray scale information around each feature point,to establish the image local gray model.In the ASM image search,local gray model use to find the best mobile location feature points,then calculates the shape parameters,attitude parameters,and update these parameters.Iteration continues until the position of the feature point did not change significantly,to obtain the target shape..Combined with knowledge of lung physiology and medical CT image characteristics establish ASM model,to achieve adhesion lung contour completion.Finally,the study results were analyzed and summarized.