Abstract
Pneumonia is a common infection that inflames the air sacs in the lungs, causing symptoms such as difficulty breathing and fever. Although pneumonia is not difficult to treat, prompt diagnosis is crucial. Without proper treatment, pneumonia can be fatal, especially in children and the elderly. Chest x-rays are an affordable way to diagnose pneumonia. Investigating an algorithmic model that can reliably and intelligently classify pneumonia based on chest X-ray images could greatly reduce the burden on physicians. The advantages and disadvantages of each of the four convolutional neural networks VGG16, ResNet50, DenseNet201, and DWA algorithm models are analyzed and given by comparing and investigating each model. The VGG16, ResNet50, and DenseNet201 network models are compared with the DWA model. When training the depthwise separable convolution with attention neural network (DWA), the training accuracy reaches 97.5%. The validation accuracy was 79% due to the model’s tendency to overfit, and the test dataset had 1175 X-ray images with a test accuracy of 96.1%. The experimental results illustrate the effectiveness of the attention mechanism and the reliability of the deeply separable convolutional neural network algorithm. The successful application of the deep learning algorithm proposed in this paper on pneumonia recognition will provide an objective, accurate, and fast solution for medical practitioners and can provide a fast and accurate pneumonia diagnosis system for doctors.
Highlights
Pneumonia is a common infection that inflames the air sacs in the lungs, causing symptoms such as difficulty breathing and fever
Investigating an algorithmic model that can reliably and intelligently classify pneumonia based on chest X-ray images could greatly reduce the burden on physicians. e advantages and disadvantages of each of the four convolutional neural networks VGG16, ResNet50, DenseNet201, and DWA algorithm models are analyzed and given by comparing and investigating each model. e VGG16, ResNet50, and DenseNet201 network models are compared with the DWA model
When training the depthwise separable convolution with attention neural network (DWA), the training accuracy reaches 97.5%. e validation accuracy was 79% due to the model’s tendency to overfit, and the test dataset had 1175 X-ray images with a test accuracy of 96.1%. e experimental results illustrate the effectiveness of the attention mechanism and the reliability of the deeply separable convolutional neural network algorithm. e successful application of the deep learning algorithm proposed in this paper on pneumonia recognition will provide an objective, accurate, and fast solution for medical practitioners and can provide a fast and accurate pneumonia diagnosis system for doctors
Summary
The structure of the deep convolutional neural network is used in this paper, of which there are four types of convolutional neural networks: VGG16, ResNet, DenseNet201, and DWA. ResNet won the championship in the 2015 ILSVRC (ImageNet Large Scale Visual Recognition Challenge) [6] He Kaiming proposed a residual learning method to solve the degradation problem [7]. A key design rule of ResNet is to double the number of feature maps (feature maps) when their size is reduced to half the size in order to maintain the complexity of the network layers. The short-circuiting mechanism is added between every two layers to form residual learning in RESNET compared to a normal neural network, where the dashed line indicates the change in the number of feature maps.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.