S2858
RTT - RTT education, training, and advanced practice
ESTRO 2026
Digital Poster 1940
Optimizing Prostate SGRT Workflow Using Lean Principles: Improving Efficiency in Clinical Practice Matthias Hermans 1 , An Sprangers 1 , Ali Dabach 1 , Jana Decraene 1 , Hannelore Van Dyck 1 , Jolien Cleijman 2 , Klaartje Steensels 2 , Dirk Verellen 1 1 Physics, Iridium, Antwerp, Belgium. 2 Radiotherapy, Iridium, Antwerp, Belgium Purpose/Objective: This project aimed to optimize the prostate SGRT workflow using lean management principles, emphasizing multidisciplinary collaboration between RTTs, physicists, and quality managers. A brownpaper session identified two key challenges: reference surface cropping and pitch/roll overrides. Three studies addressed these: (1) development of an automatic cropping method, (2) retrospective comparison of SGRT versus markerline positioning, and (3) evaluation of C-RAD’s predictive accuracy for pitch, roll, and rotation. These studies guided targeted workflow improvements. Material/Methods: Retrospective analyses used the C-RAD database of 29 prostate cancer patients treated with SGRT (230 fractions) and 25 patients treated using markerlines (150 fractions). An in-house script calculated differences between C-RAD surface positioning and online CBCT matching. Automatic cropping of reference surfaces was implemented in RayStation and validated with test groups of 10 patients. Bladder contours were auto-generated on CBCTs using RayStation’s segmentation tools to assess the impact of bladder filling on pitch alignment differences between surface-based positioning and CBCT matching. Results: SGRT-based positioning showed improved accuracy compared to markerlines, especially in the longitudinal direction (mean: 0.14 mm vs. -2.34 mm; stdev: 3.1 mm vs. 18.3 mm) and laterally, where variability was reduced (stdev: 1.5 mm vs. 3.5 mm). Vertical deviations were comparable for both methods.Figures 1 and 2 show cropping comparison. Manual and automatic cropping achieved similar translational accuracy, though rotational variability was slightly higher with auto-crop. Non-cropped surfaces performed comparably and showed slightly better pitch values, suggesting workflow simplification.Correlation analysis revealed strong agreement between C-RAD and CBCT for roll (r = 0.63) and rotation (r = -0.61), but poor predictive accuracy for pitch (r = 0.10). A moderate correlation between pitch deviations and bladder filling (r = 0.23) suggests that bladder volume changes partly explain pitch variability.
Figure 1 Translational alignment between C-RAD and CBCT for manual, automatic, and no cropping.
Figure 2 Rotational alignment between C-RAD and CBCT for manual, automatic, and no cropping. Conclusion: Automatic cropping streamlined the SGRT workflow, reducing manual workload for RTTs and improving efficiency. SGRT positioning tolerance margins were tightened for lateral translation, rotation, and roll, while pitch tolerance was increased based on bladder analysis. The observed variability in pitch was confirmed to correlate moderately with bladder filling, explaining why pitch deviations occur despite accurate surface matching. Overall, automatic cropping was successfully integrated into the workflow, and findings suggest that non-cropping may be a viable future option, further simplifying processes without compromising accuracy. Keywords: SGRT, Automation, Prostate
Made with FlippingBook - Share PDF online