Automatic Recognition System For Lung Cancer Based On Deep Learning

With the continuous development of artificial intelligence technology,deep learning algorithms have many important applications and research significance in the field of intelligent medicine,it had excellent performance,especially in the recognition,classification,detection and segmentation of medical images.Replacing doctors with artificial intelligence to handle some simple and complex repetitive things can reduce a lot of time cost,at the same time,complex illnesses can be easily dealt when the artificial intelligence assisting doctors to interpret these cases.The intelligence of medical devices has brought convenient and rapid diagnostic assistance to doctors in the detection,diagnosis,evaluation,image segmentation,registration and quantification of diseases,and has made great progress,reducing the pressure on doctors and improving the accuracy of the diagnosis of the lesion.Lung cancer is the most harmful and the highest rate of death in cancer,among all diseases.However,in the treatment of lung cancer,the survival rate of lung cancer patients found early is at least twice than advanced carcinoma,so early detection of lung cancer can greatly improve the cure rate of cancer patients.The early lung cancer imaging features are not obvious,doctors only have difficulty by the naked eye and are easily missed diagnosis,but through artificial intelligence automatic screening can quickly and efficiently identify smaller early tumors.Therefore,this paper will study the automatic identification of lung cancer and develop an automatic identification system.In this thesis will used DeepLesion public data as a training data,for the pathological features such as the leaf sign and burr sign of lung cancer in the data,and the images of most lesions in the data set will be 5*5 to 20*20.Based on the VGG-16 network model,combined with the advantages of the ImageNet competition network and Googlenet's inception structure,the deep learning algorithm is studied and improved on the convolutional neural network structure and parameters.At the same time,combined with the characteristics of neural network recognition and the characteristics of the data,the data will be preprocessed to highlight the features of the image and improve the accuracy of the deep learning algorithm.And the auxiliary recognition system is developed around the improved algorithm,so that the deep learning algorithm can express the automatic recognition of lung cancer intuitively through the recognition system after a large amount of data training.The automatic identification system has certain forward-looking characteristics.At present,the application of artificial intelligence in the health field mainly includes: auxiliary medical treatment,medical imaging,drug mining,and health management.In recent centuries,many smart medical companies have emerged,and many landing applications have been developed from these directions.This thesis will use Python as the development language,based on the PyQt function library for system development.Through the layout of the visual interface,the main programs such as training,recognition,image preprocessing and image display are put into the corresponding interfaces in each module framework to realize the overall design of the automatic lung cancer detection and detection system.By setting the system training and identification parameters to optimize the performance of the system,the learning accuracy rate reached 93.82%.Finally,public data and hospital data will be predicted and analyzed on this system.Through identifying two different data,the accuracy and missed diagnosis rate of the system in both cases are up to the expected design effect.In the hospital clinical data identification,the rate of missed diagnosis is 15% to 20% compared with the doctor.The missed diagnosis rate of this system was 14.16%,which was higher than doctor`s.Through this system,the CT image of the lung can be interpreted,the scope of the doctor's reading is narrowed,and the efficiency of the treatment is improved.Although the system has obtained a good result,there are still many problems to be studied and further improved.With the further development of artificial intelligence,smart medical systems will be as popular as hospital visits.Accuracy will be higher and higher,and it can better assist medical care.