Convolutional Neural Networks for Early Detection and Prediction of Viral Diseases through Analysis of CT Scan Images
Asadi Srinivasulu1* and Anupam Agrawal2
1Research Scholar, Indian Institute of Information Technology, Allahabad, Uttar Pradesh, India
2SM IEEE. Professor of IT, Indian Institute of Information Technology, Allahabad, Uttar Pradesh, India
*Corresponding Author: Asadi Srinivasulu, Research Scholar, Indian Institute of Information Technology, Allahabad, Uttar Pradesh, India.
Published: November 08, 2023
The escalating incidence of viral conditions such as COVID-19, flu, and various respiratory disorders presents a considerable obstacle for healthcare systems worldwide. Prompt identification and forecasting assessments are essential for efficacious treatments and control measures. Conventional diagnostic techniques like PCR assays and X-ray scans either take too long or lack the requisite precision, adding complexity to an already difficult situation. This study seeks to exploit the potential of Convolutional Neural Networks (CNN) for scrutinizing CT images in the early recognition and prediction of viral illnesses. We utilized an array of CNN designs, including ResNet, VGG, and Inception among others, and conditioned them on an extensive collection of CT images marked for several types of viral infections. We grappled with issues such as inconsistent data, unequal class distribution, and the demand for instantaneous analyses. Additionally, ethical questions tied to using patient information for algorithmic learning present further significant hurdles. Initial findings suggest that models based on CNN surpassed conventional methods in diagnostic efficacy, with accuracy levels exceeding 95%. This emerging technology offers a promising avenue to transform the way we diagnose viral diseases, providing a swifter, more reliable, and more precise approach.
Keywords: Convolutional Neural Networks (CNN); Viral Diseases; CT Scan Images; Early Detection; Prediction and Diagnostic Methods