S538
Clinical – Head & neck
ESTRO 2026
of nasopharyngeal cancer." Journal of applied clinical medical physics 25.9 (2024): e14474.2.Ng, Wai Tong, et al. "Application of artificial intelligence for nasopharyngeal carcinoma management–a systematic review." Cancer management and research (2022): 339-366.3.Hu, Lin, et al. "Semi-supervised NPC segmentation with uncertainty and attention guided consistency." Knowledge-Based Systems 239 (2022): 108021.4.Wong, Jordan, et al. "Comparing deep learning-based auto-segmentation of organs at risk and clinical target volumes to expert inter-observer variability in radiotherapy planning." Radiotherapy and Oncology 144 (2020): 152-158. Keywords: Headneck,AI contour,Training reliability strategy Digital Poster 409 Definitive Chemoradiotherapy Outcomes in Patients with Locally Advanced Laryngeal Cancer Asma Daneshvar, Candan Abakay radiation oncology, uludag University, Bursa, Turkey Purpose/Objective: This study aimed to assess the effectiveness and toxicity of definitive chemoradiotherapy (CRT) in patients with locally advanced laryngeal squamous cell carcinoma (SCC), with a focus on treatment response, survival outcomes, and associated toxicities. Material/Methods: A retrospective analysis was performed on 68 patients with stage IIIA–IVB laryngeal SCC treated with definitive CRT at Uluda ğ University Hospital between January 2010 and January 2022. Patient demographics, chemotherapy regimens, radiotherapy parameters, and treatment responses were documented. Survival analyses were conducted using SPSS version 29. Results: The median overall survival (OS) was 46.0 months, with disease-free survival (DFS) of 33.8 months. Local control (LC) was significantly associated with T stage (p = 0.004) and complete response (p < 0.001). Grade 3–4 dysphagia occurred in 14.7% of patients, xerostomia in 4.4%, and hematologic toxicities in 11.8%. Most patients were treated with cisplatin-based chemotherapy regimens and intensity-modulated radiotherapy (IMRT). Conclusion: Definitive CRT is a promising organ-preserving approach for locally advanced laryngeal cancer. Tumor stage and complete response are critical prognostic factors. Despite treatment-related toxicities, CRT results in favorable survival and local control outcomes. References: Zhang Q, Wang H, Zhao Q, et al. Evaluation of risk
Figure 1 ： Training loss decomposition showing the three learning phases (initial–trade-off–convergence)
Figure 2 ： Comparison between early stopping vs. delay-stopping under imbalanced dataset Conclusion: This study provides a clinically reliable solution to the blackout problem in limited-data radiotherapy segmentation. The proposed delay-stopping framework stabilizes training, prevents tumor- recognition collapse, and enhances contour accuracy across physicians. The integration of a semi- supervised EfficientNetV2 further improves generalization and contour consistency for CTV delineation. Together, these advances establish a foundation for personalized, data-efficient, and clinically deployable AI contouring systems, improving precision and efficiency in radiotherapy planning. References: 1.Sjogreen, Carlos, et al. "Landmark - based auto - contouring of clinical target volumes for radiotherapy
Made with FlippingBook - Share PDF online