Detection and prioritization of COVID-19 infected patients from CXR images: Analysis of AI-assisted diagnosis in clinical settings

In this paper, we present the significant results from the Covid Radiographic imaging System based on AI (Co.R.S.A.) project, which took place in Italy. This project aims to develop a state-of-the-art AI-based system for diagnosing Covid-19 pneumonia from Chest X-ray (CXR) images. The contributions of this work are manifold: the release of the public CORDA dataset, a deep learning pipeline for Covid-19 detection and prioritization, the clinical validation of the developed solution by expert radiologists, and an in-depth analysis of possible biases embedded in the data and in the models, in order to build more trust in our AI-based pipeline. The proposed detection model is based on a two-step approach that provides reliable results based on objective radiological findings. Our prioritization scheme ensures the ordering of the patients so that severe cases are presented first. We showcase the impact of our pipeline on radiologists' workflow with a clinical study, allowing us to assess the real benefits in terms of accuracy and time efficiency. Project homepage: https://corsa.di.unito.it/.