Severity assessment of Covid-19 disease: radiological Visual Score versus automated quantitative CT parameters using a pneumonia analysis algorithm

Dario Baldi¹Bruna Punzo¹Andrea Colacino¹Marco Aiello¹Monica Franzese¹Ludovico La Grutta²Luca Saba³Eduardo Bossone⁴Emanuela Passaro¹Cesare Mantini⁵Carlo Cavaliere^1*Sergio Berti⁶Simona Celi⁷Antonella Meloni⁷Filippo Cademartiri⁸Erica Maffei¹

• ¹IRCCS SYNLAB SDN, Naples, Italy
• ²University of Palermo, Palermo, Sicily, Italy
• ³University of Cagliari, Cagliari, Sardinia, Italy
• ⁴University of Naples Federico II, Naples, Campania, Italy
• ⁵University of Studies G. d'Annunzio Chieti and Pescara, Chieti, Abruzzo, Italy
• ⁶Department of Interventional Cardiology, BioCardioLab, Gabriele Monasterio Tuscany Foundation (CNR), Massa, Italy
• ⁷Bioengineering Unit, Toscana Gabriele Monasterio Foundation, Pisa, Tuscany, Italy
• ⁸Gabriele Monasterio Tuscany Foundation (CNR), Pisa, Tuscany, Italy

This study aligns strongly with the objectives of the Research Topic, particularly in its exploration of AI-assisted quantitative tools in the diagnosis and management of lung diseases. By comparing a machine learning-based CT analysis tool with traditional radiologist-driven visual scoring in patients with suspected COVID-19 pneumonia, the research demonstrates how AI can enhance diagnostic accuracy, reproducibility, and clinical decision-making. The study addresses a key challenge in lung disease management-stratification of severity in acute presentations-by showcasing that AI-based approaches can match human assessments while offering consistent, scalable insights, especially in severe cases. Moreover, the work highlights how AI integration into imaging workflows can improve clinician efficiency and reliability in real-world settings. Although focused on imaging, the methodology and findings speak directly to the call's themes of precision medicine, data-driven decision support, an