Classification and Identification of COVID-19 in CT Images Using the DL Approach

Globally, COVID-19 has had major impacts on both the public and the healthcare organizations. Polymerase chain reaction screening coupled with transcription therapy can be precisely reinforced by computed tomography imaging. Convolutional neural networks have been utilized for this investigation's COVID-19 testing. We evaluated at the ways various pre-trained models performed on CT testing and determined that the algorithms' testing capacity is enhanced by more extensive, out-of-field samples. The 2019 COVID-19 pandemic had major consequences on the global economy and people's lives. Deep learning networks have been developed to obtain medical information from chest CT pictures, serving as an evolving diagnostic approach in healthcare. COVID-19 CT images show widespread and localized lesions. The work aims to employ CT scans to develop a deep learning technique using Convolutional Neural Network (CNN) to identify and recognize Covid-19 illness. The hybrid technique employing an optimization method (genetic search) in fusion with pre-trained CNN model (MobileNetV2) is presented in this paper. An accuracy, precision, sensitivity, specificity and AUC of 87.64 %, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 1 \%, 8 8 \%, 9 1 \%}$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{0. 9 2}$</tex> respectively is attained using the fusion methodology for class 0. Whereas an accuracy, precision, sensitivity, specificity and AUC of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$87.64 \%, 82 \%, 86 \%, 82 \%$</tex> and 0.95 respectively is attained using the fusion methodology for class 1.