KEY POINTS
- This retrospective study evaluated 10 patients undergoing pancreatic SBRT. Treatment used 40 Gy in five fractions within an offline adaptive plan-library workflow combined with real-time fiducial tracking.
- EP-CBCT was acquired during repeated end-expiratory breath holds of approximately 15 seconds until sufficient projections had been collected over a full 360° rotation. Acquisition used 900 projections over approximately 60 seconds.
- Respiratory blurring was assessed using edge-response width at high-contrast abdominal interfaces. A total of 543 measurement points were evaluated across treatment fractions.
- Mean edge-response width was 2.04 mm for EP-CBCT versus 1.73 mm for daily CT. The absolute mean difference was approximately 0.31 mm, remaining inside the prespecified ±1 mm equivalence margin (p < 0.01).
- Structural similarity between EP-CBCT and daily CT was high, with a mean SSIM of 0.83.
- Blinded subjective assessment still rated daily diagnostic-quality CT significantly better than EP-CBCT. However, EP-CBCT consistently scored better than a historical end-expiratory 4D-CBCT dataset.
- The study did not perform a direct within-patient comparison against conventional free-breathing CBCT, and the historical 4D-CBCT comparison came from a separate dataset. Tumour and gastrointestinal-organ visualization could also not be evaluated using the same quantitative edge metric because of limited soft-tissue contrast.
CLINICAL TAKEAWAY
Repeated end-expiratory CBCT may provide a pragmatic way to improve pancreatic image guidance on conventional C-arm linacs without MR guidance or CT-on-rails. Its value for target/OAR localization and adaptive plan selection still needs direct comparison with standard free-breathing CBCT in larger cohorts.