Pneumoconiosis CT Image Recognition Of Migrant Workers Based On Capsule Neural Network

With the development of computer technology,computer aided diagnosis(CAD)is of great significance for pathological analysis and cancer diagnosis.It has been widely used in the early screening of lung cancer.Its general steps usually include images.Preprocessing,feature extraction and image processing.The distribution of lesions on the CT map of pneumonia and the appearance of shadows around the lesions are similar to those of lung cancer.The detection and detection are susceptible to the misdiagnosis caused by the size,shape and pulmonary tissue of the lungs.This paper proposes a Capsule neural network based on Capsule neural network.Pulmonary disease identification method.Considering that the difference of texture features of CT images of the lungs contains important medical information,in order to protect the texture features,the gray mean,entropy,fractal dimension and fractal intercept are combined to form the feature vector as the texture feature,and the context model is introduced to obtain the context.Information,using the LBG algorithm for context quantification,preprocessing the lung CT image,and completing the lung parenchymal segmentation by detecting and extracting the lung nodules.Moreover,because the existing lung CT image data is less,and the lung disease characteristics are small,in order to improve the network performance,data is enhanced on the existing pictures,and the data set is expanded.Therefore,this paper proposes a capsule neural network based method for classification and prediction of lung CT images.Firstly,the convolutional neural network and migration learning are introduced.Secondly,the Capsule algorithm is described in detail,and combined with migration learning,and then evaluated.The performance of the algorithm in the classification and prediction of digital pathological images is finally analyzed by comparing with other prediction methods.The application results show that the recognition accuracy of lung disease using Capsule neural network reaches 98.7%.Compared with the traditional convolutional network,Capsule neural network can better identify lung diseases.