Deep Learning Improves Sensitivity to Change in Sinus Computed Tomography: Evidence from Two Randomized Controlled Trials

International Forum of Allergy & Rhinology, July 2026

This research evaluates a deep learning-based Sinus Severity Score (SSS) as an objective imaging endpoint in chronic rhinosinusitis with nasal polyps (CRSwNP) clinical trials. Using CT data from two Phase 3 randomized controlled trials of benralizumab (OSTRO and ORCHID), the authors compared the automated SSS with the traditional Lund-Mackay Score (LMS).

The findings demonstrate that the AI-driven SSS was more sensitive to treatment-related changes while maintaining similar correlations with clinical outcomes, suggesting that automated quantitative CT analysis may improve the ability to detect therapeutic effects and support more efficient clinical trial designs.

Why Read This Article?

• See real-world evidence for AI-powered imaging endpoints from two large Phase 3 clinical trials.
• Learn how deep learning improved sensitivity to treatment response compared with the widely used Lund-Mackay scoring system.
• Understand the potential of automated CT analysis to provide more objective, reproducible, and scalable assessments in clinical research
• Explore how advanced quantitative imaging could enable smaller and more efficient trials by detecting subtle treatment effects.
• Gain insights into the future of AI-enabled imaging biomarkers and their role in respiratory and inflammatory disease drug development