Medical Image Detection Of Novel Coronavirus Pneumonia Based On Residual Attention Network

In early 2020,novel coronavirus pneumonia global challenges posed great challenges to medical resources all over the world.Hospitals around the world were overcrowded.During the novel coronavirus pneumonia treatment,novel coronavirus pneumonia patients’ pneumonia detection process has brought enormous burden to the medical institutions all over the place.As early as a few years ago,people have combined the most advanced advanced learning technology with medical treatment,which greatly improves the efficiency of medical diagnosis.For example,we know well that lymphoid cancer,brain cancer and other diseases can be identified and detected by computer-aided medical detection system.Novel coronavirus pneumonia novel coronavirus pneumonia diagnostic system is still scarce compared with other diagnostic systems.Therefore,it is urgent to establish a deep learning model for image detection of new crown pneumonia.A novel coronavirus pneumonia novel coronavirus pneumonia automatic diagnosis model based on residual network of recurrent attention mechanism is proposed.Because of the relatively large noise of medical images,the new algorithm is used to classify and recognize the new crown pneumonia medical images.The multi-scale MSR(Multi-Scale Retinex)algorithm is introduced to enhance the image features.The purpose of this algorithm is to make the details of lung focus image more obvious,and provide the key information of pneumonia image features for the training of the model.In the process of diagnosis of focus in medical image,data acquisition is a more difficult task.Due to the scarcity of related data and the privacy issues,the amount of data available in this paper is relatively small.Novel coronavirus pneumonia data novel coronavirus pneumonia novel coronavirus pneumonia novel coronavirus pneumonia image is trained to get an initial model.The data used in the experiment are all from GitHub website,and the accuracy rate of training based on the data set of this website is 0.966.Through a series of experiments,the following conclusions can be drawn:the detection of new crown patients by using deep learning model is feasible.