S772
Clinical - Lung
ESTRO 2026
registration (a demons method by Matlab) between DI1 and DI2/DI3 generated deformation vector fields (DVFs) from which motion vectors were extracted for the gross tumor volume (GTV) center and left/right diaphragm apices. Patients were categorized by breath-hold consistency. Ten patients were excluded because of incomplete traces or discrepancy between RGSC and CT. Group A included 140 patients (169 tumors) who had reasonable RGSC and CT agreement; Group B, as a subset, had 58 patients (with 76 tumors) who had all three breath holds within a same 3 mm window. Correlations and root mean square errors (RMSE) were analyzed between RGSC, diaphragm, and tumor motions. Results: For all Group A patients, absolute 3D tumor displacement reached 4.9 mm at the 90th percentile and 6.4 mm at the 95th percentile, with maxima exceeding 10 mm. Even among great breathers (Group B), 3.4% had mean motion >5 mm and 15.5% had a range >5 mm across all three DI scans. Tumor–RGSC correlations were poor for both AP (r=0.28; RMSE=1.5mm) and superior-inferior (SI) (r=0.26; RMSE=2.6 mm). By contrast, tumor SI variation correlated strongly with ipsilateral diaphragm position (r=0.71, RMSE=1.7 mm) and moderately in AP (r=0.56, RMSE=1.0 mm). Diaphragm variability correlated moderately with RGSC amplitude: AP r ≈ 0.50 (RMSE ≈ 2.0 mm) and SI r ≈ 0.49 (RMSE ≈ 4.1 mm), with modest RMSE improvement ( ≈ 0.2–0.4 mm) when breath holds remained within ±3 mm.
verify the results. Keywords: Lung cancer, Curative RT, Methylated ctDNA
Digital Poster 1628
anatomic uncertainty in deep inspiration breath hold radiotherapy for lung tumor treatments monitored by external surrogates Weixing Cai 1 , Bohong Huang 1 , Wei Lu 1 , Wendy Harris 1 , Lakshmi Santanam 1 , Tianfang Li 1 , Yabo Fu 1 , Daphna Gelblum 2 , John Cuaron 2 , Laura Cervino 1 , Jean Moran 1 , Xiang Li 1 1 Medical Physics, Memorial Sloan Kettering Cancer Center, New York, USA. 2 Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, USA Purpose/Objective: This study aims to quantify inter–breath-hold variability of lung tumor and thoracic anatomy during deep inspiration breath hold (DIBH) and to evaluate how well external surrogates and the ipsilateral diaphragm predict tumor position, with implications for treatment planning and motion management.
Conclusion: Internal anatomy during DIBH can vary by several millimeters between breath holds even when RGSC traces are nearly identical; About 15% of all cases approach or exceed typical SBRT margins (5 mm), and two cases (1.5%) exceed 1 cm. RGSC amplitude is an unreliable predictor of tumor position, whereas ipsilateral diaphragm position strongly predicts tumor SI displacement in most patients. References: S Han-Oh et al, “Geometric Reproducibility of Fiducial
Material/Methods: This single-institution retrospective study included 150 lung radiotherapy simulations (2023–2025) using a standardized DIBH protocol, including one free- breathing (FB) helical CT and three sequential DIBH scans (DI1–DI3) within 2-3 minutes. DI1 served as the planning baseline. The Varian Respiratory Gating for Scanners (RGSC) system recorded anterior–posterior (AP) amplitude; the mean AP level per DIBH scan quantified RGSC drift relative to DI1. Deformable
Made with FlippingBook - Share PDF online