Lung Nodules CAD System Based On Chest CT Image

CT image based CAD (Computer-aided Diagnosis) technology is currently anadvanced medical imaging diagnostic clinical application technology in the world.This technology can effectively identify suspicious lesions in CT images by usingthe computer image processing technology. CAD technology can improve theaccuracy of recognition of lesions of the physician, also can reduce a lot of work ofdoctors which improves the efficiency of diagnosis.The main work of this paper is developed a CT image based CAD (ComputeraidedDiagnosis) system for the detection of cancerous lung nodules. First thesystem reprocess the original CT image, by which the noises of the digital imagewas removed. Then the segmentation algorithm was used on the CT image, afterwhich the chest area was segmented from the original CT image. The systempreliminary segment the lung region based on the chest area. Then, the repairalgorithm is used in the lung region obtained to fix the lose part in the preliminarysegmentation and gets a more complete lung region, so that the next step oftreatment will be more convenient. Then the Distance Transform algorithms is usedon the lung region to segment the lung nodules, and then compute the feature valuesof lung nodules. Each nodule will be calculated and get a set of feature values, thesystem filter, analysis and class the nodules with those feature values, and thencompare and analyse the final classification results to verify its accuracy. In theprocess of image processing, this paper has proposed a new region filling algorithm,that is the Inside and Outside Boundary fill algorithm, and in practical applicationsit has obtained a good result.The system can effectively complete the detection and analysis of lung nodule,in practice there are good results, and has a very important value.