An Ensemble Learning Model to Detect COVID-19 Pneumonia from Chest CT Scan

The novel COVID-19, initially found in Wuhan (China), reached quickly around the globe and turned into a worldwide pandemic situation. It has set off a significant impact on daily life, general well-being, and global finance. It is crucial to diagnose predisposed patients rapidly. There are no exact tests for COVID-19 except RT-PCR which is costly and needs a huge time. Recent research acquired applying radiology imaging approaches recommend that such images comprise features about the COVID-19 infection. The use of machine learning techniques combined with chest imaging can be helpful in the precise recognition of this infection, and can likewise be assistive to beat the issue of an absence of specific doctors. This investigation developed a model for automatic identification of COVID-19 infection utilizing chest CT images. A convolutional neural network has been applied to extract the features from the chest CT images and Principle Component Analysis has been applied for feature selection to reduce computational effort. The proposed model (the ensemble of ML classifiers) has been developed to provide accurate diagnostics by considering the five classes (Normal, Mycoplasma pneumonia, Bacterial pneumonia, Viral pneumonia, and COVID-19). The proposed model reached an accuracy of 99.3%, precision of 99.3%, and recall of 99.2%. This can help clinicians invalidate their primary checkups and can be utilized promptly to check the patients’ infection rate.