The Feature Recognition And Aided Diagnosis Of Chronic Obstructive Pulmonary Disease CT Images

Modern life has rapidly improved people's living standard and quality of life.People are more concerned about health and pay more attention to the prevention and treatment of major diseases.The incidence of chronic lung diseases is increasing year by year,showing a trend of younger age.It has become the focus and difficulty of clinical diagnosis for medical staff.In order to effectively treat and prevent chronic obstructive pulmonary disease(COPD),the paper relies on the research project of CT imaging assistant diagnosis of chronic obstructive pulmonary disease in Xinqiao Hospital.This paper mainly aims at the doctor's clinical diagnosis of chronic obstructive pulmonary disease by random selection of CT images,and it is difficult to extract the features of the morphological changes of the lesion.Based on a large number of image data in the CT image database system of chronic obstructive pulmonary disease in Xinqiao Hospital,the research content and results are as follows:(1)to extract the doctor's experience in the clinical diagnosis of the prominent characteristics of the lung lesions in the patients with chronic obstructive pulmonary disease,as a priori knowledge,the CT images with obvious focus features were extracted,the features of the lesions were classified and the training data set was set up.By using a convolution neural network(convolution neural network)algorithm for multi-layer training and testing,an improved feature extraction algorithm for slow resistance lung focus in convolution neural network is proposed by comparing learning,determining the number of algorithms and initialization parameters.We use different rules of convolution kernel operation to remove a large number of invalid interference pixels,and summarize the characteristics of lesions.Experimental results show that the algorithm is effective and fast,and it is accurate to extract lesion features from CT images.(2)by analyzing the distribution of the lesions on the left and right lobe of the CT image of the patients with chronic obstructive pulmonary disease and the intensity of the focus of the focus,the number of pixels of the focus of the CT image was calculated.Based on medical lung function data PFT,we extract parameter variation data for inhalable air volume due to changes in lung tissue.Fitting the trend curve of the feature change data with the trend curve of the parameter change data,using the data fusion result to judge the development trend of the patient during the treatment,provide the treatment plan for the CT image auxiliary diagnosis of the slow resistance lung,and realize the effective application of the CT image characteristic data of the lung.