Artificial Intelligence-Assisted Lung Nodule Evaluation on Low-Dose Chest CT in Asymptomatic Individuals: A...

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Healthcare
• Clinical Trials

Background:

• Previous studies on AI for lung nodule evaluation used retrospective designs outside clinical practice.
• Real-world clinical workflow integration of AI tools for low-dose CT (LDCT) requires evaluation.

Purpose of the Study:

• To compare interpretation times and lung nodule detection rates with and without an AI tool in LDCT examinations.
• To assess the impact of AI on clinical decision-making for asymptomatic individuals undergoing LDCT.

Main Methods:

• Prospective, single-center, parallel, open-label clinical trial with 1:1 randomization.
• Intervention group used AI-assisted interpretation; control group used standard interpretation.
• Radiologists reported nodules ≥4 mm; primary outcome was interpretation time, secondary outcomes included nodule detection rates and follow-up recommendations.

Main Results:

• No significant difference in interpretation time (187s vs 172s).
• AI group showed significantly higher detection rates for Lung-RADS-positive nodules (16.9% vs 10.3%) and all nodules (52.9% vs 32.6%).
• AI group had more follow-up LDCT recommendations (15.3% vs 7.4%).

Conclusions:

• AI tools integrated into PACS do not increase interpretation time for LDCT.
• AI-assisted interpretation significantly improves the detection of clinically actionable lung nodules.
• This trial provides pragmatic evidence for AI's role in LDCT interpretation.