Research Of Lung Nodules Detection Algorithms Based On The Lung Immune Segmentation

In all cancers,the incidence and mortality of the lung cancer have been at the forefront,it has become one of the most important cancers that are harmful for our health.In the medical CT images,the lung cancer was shown in the form of pulmonary nodules in the early stage,if we can detect the pathological changes in the lung and take the timely treatment,the patient's mortality will be greatly reduced.But in the reality of the diagnosis,doctors have to read lots of CT images every day,high reading and diagnostic accuracy rate are always conflict with each other.With the development of the artificial intelligence,medical image diagnosis technology and image recognition technology,computer aided detection system can help doctors to diagnose the diseases.So it is very important and valuable to study the algorithm of computer aided detection system based on lung nodules.According to the authority of the United States lung cancer image research database,we analyzed the shortcomings of existing algorithms,and the improved algorithms were proposed,ultimately we realized the segmentation of lung parenchyma and the detection of small pulmonary nodules.Firstly,this article carried out the median filter algorithm as pretreatment to remove the image noise.Secondly,facing the deficiencies of the Otsu algorithm,we used the improved Otsu algorithm based on immune genetic for the lung image segmentation.And we combined with the region growing algorithm and morphology algorithms to produce the lung parenchyma mask,then multiplied this parenchyma mask with the original CT image,we got the lung parenchyma image.Finally,the rolling ball algorithm was used to repair the lung parenchyma image edge to form the final image of lung parenchyma.On the basis of the final image of lung parenchyma,we had explored a series of image edge detection algorithms,Susan algorithm and FCM algorithms in the detection of the lung nodules,and the improved Fast algorithm was used to detect the nodules in the lung parenchyma images eventually,meanwhile the detected lung nodules were marked out with colorful points,thus we had researched the algorithms of the pulmonary nodules detection process.In the end of this article,we analyzed the whole algorithms with its stability,efficiency and accuracy.We did lots of experiments,and the results showed that the algorithms were feasible,those algorithms can be used to segment lung parenchyma image and detect the ROI areas of the lung nodules.