Cloud-based Artificial Intelligence-assisted Diagnostic System For Multi Pathological Types Of Benign And Malignant Of Pulmonary Nodules Based On CT Images

The research of computer-aided diagnosis(CAD)is a rapidly growing and active field.Computer-aided diagnosis technology is used to process the computed tomography(CT)images of human body parts,which can provide practical reference information for doctors,assist doctors to make accurate diagnosis of diseases,and effectively reduce the occurrence of missed diagnosis and misdiagnosis.The purpose of this thesis is to design a Cloud-based Artificial Intelligence-assisted Diagnostic System for Multi Pathological Types of Benign and Malignant of pulmonary Nodules Based on CT Images.The aim of this study is to design a computer-aided diagnosis system that integrates the functions of CT display,pulmonary nodule extraction,pulmonary nodule classification diagnosis,etc.To assist radiologists in the diagnosis of multi pathological types of pulmonary nodules.A deep learning network model is constructed to classify and diagnose pulmonary nodules in lung CT images in the thesis.For the extraction of pulmonary nodules,in order to extract pulmonary nodules completely and not extract excessive background information when extracting small pulmonary nodules,a 3D multi-resolution data processing method is proposed.The multi-resolution 3D dual cpath network model is built based on the idea of joint dual path network(DPN).The area under the receiver operating characteristic curve(AUC)and classification accuracy(ACC)are used as the objective evaluation index to characterize the effect of the model.After the test set test,the ACC and AUC of the designed model reached 0.812 and 0.8757 respectively.On the basis of the multi-resolution 3D dual path network model,a cloud artificial intelligence aided diagnosis software system for pulmonary nodules is designed based on pyqt5.Considering the limited GPU resources of ordinary personal computer equipment,it is very difficult to realize deep learning operation under the condition of limited computing resources,so the software system is divided into two aspects:client and cloud server.Client realizes user login,lung CT images display,brightness and contrast adjustment,lung nodule coordinate selection and other functions.The GPU cloud server is built to complete the functions of pulmonary nodule extraction,diagnosis classification,file storage and so on,and the mysql server is built to store information and login verification.On the basis of secure shell(SSH)protocol,a simplified software structure of file and instruction interaction of world wide web(web)server is proposed to realize the design of user interface and cloud server function interaction.The cloud artificial intelligence aided diagnosis system of pulmonary nodules designed in this thesis not only encapsulates the complete function of CT images pulmonary nodule processing and diagnosis,but also expands the scope of application of the system through the design of cloud server.It has certain application value for computer aided diagnosis of benign and malignant pulmonary nodules in CT images.