Efficient explainable deep learning technique for COVID-19 diagnosis based on computed Tomography scan images of lungs

Abstract

The entire human race is currently facing a huge disruption of everyday life due to the rapid spread of the novel Corona Virus disease 2019 (COVID-19). It is essential to develop a tool or model for fast diagnosis of the disease which is pandemic and also the model should be able to justify the result for trustworthy in the field of medicine. Machine learning (ML) and Deep Learning (DL) models play a vital role in identifying COVID-19 patients by visually analyzing their Computed Tomography (CT) scan images. In this paper, few publicly available convolutional neural network models (CNN) were analyzed to classify the CT scan images of lungs into two classes, COVID-19 positive and negative cases. In addition to that, Local Interpretable Model-agnostic Explanation (LIME) framework is used as an explanation technique for interpretability. The pixel of relevancy responsible for the outcome of classification is visually explained through LIME technique which gives trustworthiness in the field of healthcare. © 2022 Author(s).