Deep Learning Assisted Covid-19 Detection using full CT-scans

The ongoing pandemic of COVID-19 has shownthe limitations of our current medical institutions. Thereis a need for research in the field of automated diagnosisfor speeding up the process while maintaining accuracyand reducing computational requirements. In this work, anautomatic diagnosis of COVID-19 infection from CT scansof the patients using Deep Learning technique is proposed.The proposed model, ReCOV-101 uses full chest CT scans todetect varying degrees of COVID-19 infection, and requiresless computational power. Moreover, in order to improvethe detection accuracy the CT-scans were preprocessed byemploying segmentation and interpolation. The proposedscheme is based on the residual network, taking advantageof skip connection, allowing the model to go deeper.Moreover, the model was trained on a single enterpriselevelGPU such that it can easily be provided on the edge ofthe network, reducing communication with the cloud oftenrequired for processing the data. The objective of this workis to demonstrate a less hardware-intensive approach for COVID-19 detection with excellent performance that canbe combined with medical equipment and help ease theexamination procedure. Moreover, with the proposed modelan accuracy of 94.9% was achieved.