INESC TEC tests the use of Artificial Intelligence in the diagnosis of COVID-19

INESC TEC is the coordinator of project THOR (Computer Assisted Thoracic Assessment using POCUS), which will test the use of Artificial Intelligence (AI) in the diagnosis of COVID-19, by analysing chest ultrasounds using computer vision techniques.

Lung ultrasound enables the fast diagnosis and monitoring of cases, namely in situations where performing other tests is not suitable. The use of chest ultrasounds (Point-of-Care Ultrasound – POCUS) is particularly interesting in the context of COVID-19, since the ultrasound system is portable and easy to use, allowing the diagnosis of viral pneumonia associated with most severe cases, by identifying pleural thickening and the presence of characteristic patterns of B lines (vertical lines on the periphery of the lung).

“However, the widespread use of this technique is limited by the operational complexity of an ultrasound exam. The correct positioning of the probe and image evaluation are two interrelated and complex tasks, which require specialised knowledge and specific training, in order to be used effectively”, explained Miguel Coimbra, coordinator of INESC TEC’s TEC4Health, and lecturer at the Faculty of Sciences of the University of Porto (FCUP).

The computer-aided diagnosis focuses not only on guiding the expert during the examination, but also, after the examination, on automatically quantifying the presence of B lines in the lungs, indicators of viral pneumonia.

The prototype will be installed in Garcia de Orta hospital, which already uses ultrasounds to study patients with COVID-19, for clinical research purposes. “The HGO team, made up of two elements with experience in chest ultrasound (Rui Gomes and Jacobo Bacariza), will recruit patients with COVID-19, in order to collect, annotate and process images and videos. These will act as a substrate for the algorithm and computer vision project. Later, HGO will also be the location for installing the prototype for the final stages of the project “, said Filipe Gonzalez, medical coordinator of the project at HGO.

This project is a multidisciplinary collaboration between INESC TEC (computer vision), the Garcia da Orta hospital (medicine), FC.ID (computer vision) and Nevarotech (interoperability, prototyping and technology transfer).

THOR (Computer Assisted Thoracic Assessment using POCUS) is one of 12 projects funded by the programme “AI 4 COVID-19: Data Science and Artificial Intelligence in Public Administration to fight against COVID-19 and future pandemics – 2020”, promoted by the Foundation for Science and Technology. The project received €240K of funding and the agreed duration is 24 to 36 months.

INESC TEC researcher mentioned in this article is associated with UP-FCUP.