ESTRO 2025 - Abstract Book

S3198

Physics - Intra-fraction motion management and real-time adaptive radiotherapy

ESTRO 2025

Phantom plans were created for each density insert with the corresponding CT image. A synchrony model was established to correlate the external surrogate (LED markers) motion with the internal target motion. The tracking accuracy was assessed by extracting the comprised vector of correlation errors in X, Y and Z directions from the log data analysis, indicating the discrepancy between the target locations identified in x-ray images and those predicted by the latest correlation model. CNR was quantified by dividing the CT value differences between each density plug and the surrounding lung tissue region of interest (ROI) by the standard deviation of lung tissue ROI. Pearson correlation tests were performed and the relationship between CNR and the correlation error was evaluated. P95% of the correlation error is defined as the 95 th percentile of absolute correlation errors. A clinically acceptable CNR was determined by assuming a P95% correlation error of 2mm.

Results:

The CNR threshold for tracking in this setup was 16.9 for sinusoidal motion and 16.0 for respiratory motion at a P95% correlation error of 2mm, as shown in Figure 1(b). Tracking errors increased significantly when the CNR fell below these thresholds, indicating model-building challenges and larger tracking discrepancies. A negative correlation between the CNR and the correlation error was observed, with correlation coefficients of -0.78 for respiratory motion and -0.81 for sinusoidal motion, as shown in Table 1.