The Background
In a recent study presentation1 from the New York Proton Center, MIM Software®’s AI auto-contouring solution Contour ProtégéAI was evaluated for its clinical application in complex head and neck cancer. According to the study’s authors, AI-based auto-segmentation is particularly critical for these types of treatment sites, and accurately defined organ-at-risks (OARs) are crucial to successful radiotherapy. Even though various auto-contouring systems are available on the market today, validation on real-world clinical cases has been lacking.
The Method
New York Proton Center researchers created seventeen OARs for each of their sample head and neck cases using Contour ProtégéAI. This data was compared against twenty-two benchmark datasets, and the performance of the AI contours was evaluated using Dice Similarity Coefficient and Hausdorff distance (HD). The clinical staff also visually evaluated the quality of each of the AI-generated OARs.
The Conclusion
The majority of the Contour ProtégéAI contours showed no notable difference from the benchmark contours. The researchers found that Contour ProtégéAI accurately delineated the OAR for auto-segmentation tasks on CT images without user intervention. Additionally, segmentation of the Contour ProtégéAI contours took under five minutes, which is significantly less time than it takes to manually draw similar head and neck contours.
These results demonstrate Contour ProtégéAI’s accurate and efficient application in clinical use. For more information on how MIM®’s flexible and timesaving auto-contouring solution can seamlessly integrate into your department’s workflow, email us at info@mimsoftware.com.
Find the full AAPM presentation from the New York Proton Center here.
1. Pingfang T. et al., “Validation of a Commercial Artificial Intelligence Auto-Segmentation System for Head and Neck Treatment Site,” (presentation, American Association of Physicists in Medicine 64th Annual Meeting and Exhibition, Washington DC, July 11, 2022).
