Hospitals were hurled into a health crisis that was still poorly understood during the outbreak of Severe Acute Respiratory Syndrome Coronavirus 2 that surfaced in late 2019. Due to the massive quantity of data collected during the epidemic, machine-learning algorithms were trained on it to assist clinicians in making better, more informed decisions and saving lives. Hundreds of prediction tools were developed as a consequence. The use of transfer learning-based deep learning models for recognizing COVID-19 from a Computed Tomography (CT) scan, is one of the most commonly recommended approaches. The interpretability of machine learning models is critical in human-centric sectors like healthcare. It is not enough for end users to have an accurate model; they must also be able to accept the model’s validity and correctness, as well as comprehend how it works. In this research, we focus on a few issues with existing basic transfer learning approaches for identifying COVID-19 from CT images by interpreting their outputs using Local Interpretable Model-Agnostic Explanations (LIME) and propose training the models with only lung images from chest CT scans rather than the entire scan. The models developed were capable of correctly interpreting scans to diagnose infections with high accuracy.