Image classification and discrimination of COVID pneumonia based on convolutional neural network

Abstract

To address the negative quality of RT-PCR and the time taken to obtain results, this paper reports on the classification and discrimination of COVID pneumonia images by developing a neural network adaptively. Given the limited number of published COVID-19 CT data currently available, we suggest using the proposed negative process to refine the data to obtain more CT data to reduce matching risk; Experimental results show that compared to network models such as AlexNet and GoogleNet, the proposed BUF-Net network model has the best performance with an accuracy of 93%. Visualization of systemic findings using Grad-CAM techniques may further elucidate the critical role of CT imaging in the diagnosis of COVID-19.Conclusion: The use of deep learning tools in clinical practice can help radiologists make better diagnoses.