MIA-COV19D is an international competition organized by the University of Lincoln (UK), aimed at developing new solutions for identifying COVID-19-related diseases through the analysis of lung CT images. The competition was held as part of ICCV 2021, one of the world’s leading scientific conferences dedicated to computer vision.
At synbrAIn, our data scientists have long been working on this specific challenge, developing AI-powered solutions that are already on the market, integrated into the MS humanAId platform (as mentioned in a previous article on our magazine).
This effort was carried out thanks to our close and ongoing collaboration with our partner SenticLab, which participated in the MIA-COV19D competition and achieved outstanding results.
SenticLab’s AI-Powered COVID-19 Detection Solution
SenticLab developed a custom AI solution for detecting COVID-19 from CT scans, leveraging deep learning algorithms. Their model achieved a macro F1-score of over 90%, significantly surpassing the 70% threshold required for the competition. While there was a slight difference compared to the winning team’s results, the overall approach is highly promising. Both SenticLab and synbrAIn teams are now working together to further refine the solution with new methodologies that could enhance its performance even more.
We are extremely proud of this achievement, especially given the global impact that these advancements can bring to humanity. For this reason, the SenticLab team has documented the entire approach in a scientific article now available on arXiv and accessible via the following link.
Advancing AI-Powered Medical Diagnostics
These results reaffirm the position of our group, together with our partners, as a leading force in AI-assisted medical diagnostics. We believe that these technologies have the potential to deliver exceptional and tangible benefits to society, becoming essential tools for human-machine cooperation, which lies at the core of our company’s philosophy.
