Kidney Disease Screening Based On Deep Neural Network

Kidney disease has become the fourth most important disease in the world of modern society.In China,more than 11% of the population suffer from different kidney diseases.Every year,more than one million kidney patients worldwide develop into kidney failure,kidney tumor and other malignant diseases.Therefore,the early screening of kidney disease plays an important role in the timely diagnosis and the control of the disease development.The clinical diagnosis of kidney disease is very dependent on the Color Doppler Ultrasound and Computer Tomography medical imaging approach and the doctor’s professional level.However,the regional distribution of medical resources such as doctors and equipment are seriously imbalanced in our country.At the same time,doctors need to analyze and deal with a large amount of image data,and manual reading is not only inefficient but also has a heavy workload.Therefore,the current situation of kidney disease prevention and control in China is not optimistic.The computer aided diagnosis system based on advanced computer hardware and software technology can effectively help hospitals and doctors to carry out the screening and diagnosis of kidney diseases.With the rapid development of deep learning in recent years,in the context of medical big data,combined with artificial intelligence technology for the analysis and diagnosis of medical images,the computer aided diagnosis system becomes more intelligent and accurate.The system has the advantages of short time consumption,high efficiency and free from human subjective judgment when processing a large number of medical images.It can effectively assist doctors in diagnosis,greatly shorten the time of reading images and reduce the burden on doctors.In this paper,we analyzed and studied the intelligent aided system for kidney disease,and proposed the algorithm of kidney disease classification and kidney segmentation based on deep neural network.The main content of this paper is as follows:1.This paper uses kidney color ultrasound data,proposes a kidney disease classification model Kid-Net based on deep neural network.In the model,the deep residual network is used as the basic layer of feature extraction,and the spatial and channel attention mechanism is integrated into the residual learning.The spatial pyramid pooling structure is used as the multiscale feature extraction layer.Finally,the full connected layer and Softmax are used to complete the classification.In this paper,the model is verified on the self-constructed kidney color doppler ultrasound classification data set,and the experimental results show that the model can effectively screen out the images of suspected disease from kidney data.2.This paper uses abdominal CT data,proposes a kidney organ and tumor segmentation model Kid SegNet based on deep neural network.The model uses the end-to-end encoderdecoder structure as the main architecture,integrates the attention mechanism through skip connection,increases the receptive field of the network by using atrous convolution,and makes full use of the multi-scale feature of the underlying space by combining with spatial pyramid pooling.The model is validated on open data sets,and the results show that the proposed method can effectively achieve automatic semantic segmentation of kidney organ and tumor.3.In this paper,the kidney classification and segmentation algorithms mentioned above are integrated into the computer system to build an intelligent aided system Deepkidney,which is used for kidney disease screening and kidney segmentation.The system can automatically process the uploaded kidney color ultrasound and CT data and return the diagnosis results,thus effectively assisting doctors to carry out kidney color ultrasound screening and automatic semantic segmentation.