Computer Aided Diagnostic Method Study Of Lung Nodules In Low-dose Thoracic CT Lung Image

In recent years,lung cancer is known as the number one cancer in the world,the incidence rate is still rising on a global scale,especially obvious in China.Studies have shown that earlier diagnosis and treatment are important and effective to improve the cure rate of lung cancer.Recently,along with the widespread usage of various medical imaging equipment,clinical diagnosis and related research in the field of medicine have a good promotion and development,all kinds of advanced technologies of medical imaging equipment are also getting more and more attention by the medical workers.Location and characterization of lung nodules depend on the application of these techniques.So,more and more researchers and doctors have focused on the study of lung nodules in the chest CT scan.Because the medical image itself has the influence of unevenness of gray value,individual difference,artifact,noise,edge blur and other factors,it is difficult to achieve high sensitivity and precision with the relevant algorithm.In this thesis,based on chest CT images,many experiments and researches have carried out on the aspect of the original CT image preprocessing,pulmonary parenchyma segmentation,suspected lung nodule extraction,lung nodules detection.This thesis has proposed a lung nodule detection method based on a low dose of chest CT.Firstly,the raw data were transformated and a fast adaptive weighted median filter was introduced to deal with the noise problem in CT images.Secondly,since the target region area is lung,in order to narrow the target range and the existence of abnormal picture in the original data,it is difficult to obtain better segmentation results based on the single slice processing.So a pulmonary parenchyma extracting method based on multiple successive preprocessed CT slices was proposed.Then,since the final region of interest is lung nodule,so the area of the suspected nodules were divided by the morphological method on the preprocessed CT frame sequence.Finally,according to the rule of CT truly positive nodules and spatial location,the minimum distance algorithm based on the distance of the lung margin was proposed,it finally achieves the classification of pulmonary nodules,as well as the initial location of suspected nodules,and improves the accuracy and efficiency of early diagnosis of lung cancer,reduces the low false positive rate and the workload of radiologists.