CAD4COVID: Artificial Intelligence software for COVID-19 triage

Although SARS-CoV-2 diagnostic tests such as RT-PCR are the gold standard to confirm COVID-19, demand for PCR test-kits is extremely high and many countries face capacity limitations in the resources involved, thus timely access to rapid testing and diagnosis is critical. CT imaging has shown to be a highly sensitive method to detect COVID-19, however, availability of CT imaging is limited in developing countries. Alternatively, X-ray is also used among patients presenting COVID-19 symptoms, and although it is not as sensitive as CT, X-ray uses less radiation and is often less expensive, faster, and easier to use than CT, and more widely available to patients in developing countries. Therefore, Delft Imaging has developed CAD4COVID (Computer Aided Detection for COVID-19), an artificial intelligence software that triages COVID-19 suspects on chest X-ray images to respond to the COVID-19 emergency. CAD4COVID is developed with the intention to support triaging COVID-19 suspects and help determining the next step in patient’s care, particularly in resource-constrained settings and high prevalence areas. As a social enterprise, Delft Imaging believes that we need to act fast and have the biggest impact possible; therefore CAD4COVID is available free-of-charge. Also, Delft Imaging developed CAD4COVID-CT, for COVID-19 triage on CT scans.

Key features of CAD4COVID are: 1) generate a score between 0 and 100 indicating the extent of COVID-19 related abnormalities; 2) display such lung abnormalities through a heatmap; 3) quantify the percentage of visible lung tissue that is affected; 4) show result within 20 seconds; and 5) offer cloud solution (with stable internet connection) and offline box solution (without internet connection). CAD4COVID can bring added value particularly to facilities with overwhelmed/limited capacity because it can help alleviate the burdens related to pathology tests, reduce the workload of health workforce, and enhance diagnostic infrastructure. Also, it enables quicker detection and allows for higher throughput, thereby helps prevent and control the spread of novel coronavirus. Additionally, CAD4COVID quantifies disease progression and recovery thus allows effective tracking. Finally, CAD4COVID is being integrated into Delft’s mobile clinic and can be deployed to communities or workplaces with high prevalence.