In the European 4-IN-THE-LUNG-RUN (4ITLR) trial, AI integration in lung cancer screening is evaluated. Participants, aged 60-79 years with a significant smoking history, underwent low-dose chest CT scans. These scans were initially analyzed by radiologists and then by the AI-based Lung Screen Software (AVIEW) developed by Coreline Soft. Discrepancies between the two assessments led to a third arbitration read.
Results showed a 78.7% agreement rate between human and AI readings. Around 21% of arbitration cases required a different categorization from both human and AI reads. Most participants were classified as 'negative,' with fewer in 'indeterminate' or 'positive' categories.
The study concludes AI as a concurrent reader is feasible, reducing the need for a second human reader and limiting cases requiring arbitration. Less than 5% of participants needed specialist referral after a baseline scan. The effectiveness of this AI-based approach will be further evaluated in future research.