The doctors who normally investigate and report on CT scans (radiologists) now have access to an AI solution to help them with the diagnosis and to more accurately analyse the impact of COVID-19.
The AI software for COVID-19 triage is based on CT scans. The AI product, which is snappily named 'CAD4COVID-CT', builds on a previous development called CAD4COVID for X-ray images. Both solutions have been made available free of charge and are aimed at reducing the workload of clinicians and health workers.
CAD4COVID-CT assesses the suspicion of COVID-19 as well as the severity of the infection based on the percentage of the affected lung. The software aims to enable a quicker detection of suspected COVID-19 cases and allows healthcare specialists to track disease progression and recovery in COVID-19 patients.
The images (see below) show the differences between normal CT scans and those that have been enhanced using the AI solution.
“We believe imaging and AI can make a difference in the COVID-19 pandemic,” says Dr. Eva van Rikxoort, of the AI company (Thirona). “We developed two solutions to triage COVID-19 based on our existing AI software for diseases like COPD, asthma and tuberculosis.”
Besides being available in the Cloud, Thirona has partnered with German company Smart Reporting to make CAD4COVID available on a larger scale. Smart Reporting provides a platform for enhanced medical documentation and launched a free template at the end of March to provide guidance for the diagnosis of COVID-19 in computed tomography of the chest. CAD4COVID-CT is being integrated into Smart Reporting’s template, making it instantly accessible to the more than 10,000 radiologists who are registered on its platform.
Rapid detection and disease progression analysis
CAD4COVID-CT was developed in close collaboration with Radboud University Medical Center in the Netherlands and the Institute for Digital Medicine - Fraunhofer MEVIS in Germany. The product was developed and validated in over 500 CT scans and performs on the same level as a radiologist.
In this series of CT scan images, the first picture is a regular CT followed by the same scan using the AI heat map: