RadVid-19 is an action among Brazilian radiologists supported by CBR to collect confirmed or suspected X-rays and tomography exams of COVID-19. The platform is a repository of COVID cases in Brazil, made by a joint action of Brazilian radiologists and will be open for the application of artificial intelligence to aid clinical decision, with the purpose of making this diagnosis more accurate and, in the case of tomography, automate quantitative data on disease involvement.
We will expand the base with the collection of cases at first, then annotate and classify images through a national initiative of radiologists. Then, the classified image base will be made available for the development and validation of Machine Learning algorithms.
The first algorithm already deployed on the platform analyzes the collected tomography images, generating a report that can be accessed by the radiologist, with the objective of assisting in the clinical decision. We reinforce that this initiative seeks support for clinical decision, and not replacement of a professional act.