Graph CNNs for image segmentation of thoracic images for automating COVID-19 diagnosis

1. Designing a framework to convert thoracic CT images into a graph-representation, specifically for lung-region based segmentation. 2. Designing a Graph-CNN to perform semantic segmentation to separate different Regions of Interest as a prerequisite to the classifier 3. Designing a classifier based on medical statistics of lung-appearances in COVID affected and healthy patients to decide whether a patient is infected or not.