An Explainable AI System for Automated COVID-19 Assessment and Lesion\n Categorization from CT-scans

Abstract

COVID-19 infection caused by SARS-CoV-2 pathogen is a catastrophic pandemic\noutbreak all over the world with exponential increasing of confirmed cases and,\nunfortunately, deaths. In this work we propose an AI-powered pipeline, based on\nthe deep-learning paradigm, for automated COVID-19 detection and lesion\ncategorization from CT scans. We first propose a new segmentation module aimed\nat identifying automatically lung parenchyma and lobes. Next, we combined such\nsegmentation network with classification networks for COVID-19 identification\nand lesion categorization. We compare the obtained classification results with\nthose obtained by three expert radiologists on a dataset consisting of 162 CT\nscans. Results showed a sensitivity of 90\\% and a specificity of 93.5% for\nCOVID-19 detection, outperforming those yielded by the expert radiologists, and\nan average lesion categorization accuracy of over 84%. Results also show that a\nsignificant role is played by prior lung and lobe segmentation that allowed us\nto enhance performance by over 20 percent points. The interpretation of the\ntrained AI models, moreover, reveals that the most significant areas for\nsupporting the decision on COVID-19 identification are consistent with the\nlesions clinically associated to the virus, i.e., crazy paving, consolidation\nand ground glass. This means that the artificial models are able to\ndiscriminate a positive patient from a negative one (both controls and patients\nwith interstitial pneumonia tested negative to COVID) by evaluating the\npresence of those lesions into CT scans. Finally, the AI models are integrated\ninto a user-friendly GUI to support AI explainability for radiologists, which\nis publicly available at http://perceivelab.com/covid-ai.\n

Ähnliche Arbeiten