Publication

Abstract : In recent times, the most dangerous and rapidly spreading lung infection disease is COVID-19 infection. Diagnosis of COVID-19 infection from radiographic images is one of the biggest challenges. For COVID-19 infection detection and segmentation, chest CT scans and X-rays are widely used by clinicians. Out of these, Computed Tomography (CT) scans of chest images are one the leading and promising methods to detect COVID-19 patches in the lungs of an infected person. The proposed work aims to automate the process of detection of infection patches with the help of a neural network model, called the COVSeg-Net. The model is trained and validated on a dataset consisting of 2581 CT scan images of volumetric CT scans of 10 patients. The proposed work obtained a Dice coefficient score of 0.783 and a Jaccard index of 0.729 on 381 testing CT scan slices.