Detecting Presence of COVID-19 with ResNet-18 using PyTorch

Abstract

COVID-19 virus can be detected with the help of medical images such as X-Ray and CT Scan of human lungs. The unavailability of proper resources might result in the delay of future proceedings. Henceforth, this paper analyzed and comprehended the medical images for COVID-19 lungs and tried to figure out the practical application of that procedure with the overall ecosystem. The whole paper is split into major categories such as: analyzing medical images, classifying and detecting the presence of COVID-19, and building the ResNet-18 model. The present paper creates a base for these terminologies and presents the implementation of a Classification Model on the Kaggle dataset for the X-Ray images. The current work serves the purpose of identification and detection of the Coronavirus, Pneumonia, and normal lungs with the help of ResNet-18. The model was implemented with the Transfer Learning technique. In brief, the work addresses the identification or detection of the COVID-19 reports and differentiates them from reports having symptoms synonymous with the task at hand. The accuracy obtained was 97.78% with 96.33% average sensitivity rate and 98.21% average specificity rate. The conclusion of this report ideally focuses on delineating the result coherently.

Ähnliche Arbeiten