Solid-water extensions restore HyperSight CBCT dose-calculation accuracy with a thin calibration phantom

Adding solid-water extensions restored PTV gamma agreement to 98–100% at 1%/1 mm, whereas unextended wide-field calibration performed poorly.

KEY POINTS

  • This methodological study investigated three CT-number-to-electron/material-density (CT-ED/MD) calibration strategies for the Varian Ethos HyperSight CBCT system using a standard CIRS Model 062 M phantom. Dosimetric performance was evaluated using anthropomorphic head, thorax, and pelvic phantoms, with representative VMAT/IMRT plans recalculated using Acuros XB and simulation CT serving as the reference.
  • Three calibration configurations were compared: Method A, using a narrow 4-cm superior-inferior blade aperture; Method B, using the full 24.2-cm aperture without additional phantom material; and Method C, using the same 24.2-cm aperture with 10-cm solid-water extensions at both ends of the thin phantom to provide more representative scatter conditions.
  • The calibration setup substantially affected Hounsfield unit accuracy. Mean central-water HU was −7.8 on simulation CT, compared with 3.1 for Method A, 131.1 for Method B, and −2.7 for Method C. Mean absolute HU differences from simulation CT across calibration inserts were 35.5, 96.9, and 28.4 HU, respectively, showing the pronounced distortion produced by wide-field scanning without sufficient scattering material.
  • At the stringent 1%/1 mm gamma criterion, direct wide-field calibration without solid water produced PTV passing rates of only 40.4% for head, 58.9% for thorax, and 7.0% for pelvis. Even at 2%/2 mm, thoracic and pelvic PTV passing rates remained 84.6% and 68.0%, respectively, demonstrating that the calibration error was not limited to an unusually strict gamma evaluation.
  • The narrow-aperture Method A substantially improved agreement, achieving 1%/1 mm body gamma passing rates of 96.6%, 99.5%, and 99.9% for head, thorax, and pelvis. Corresponding PTV rates were 99.0%, 85.7%, and 99.9%; at 2%/2 mm, all body and PTV passing rates exceeded 99.8%, supporting the approach as a practical alternative when solid-water extensions are unavailable.
  • Method C provided the most consistent dosimetric agreement, with 1%/1 mm PTV gamma passing rates of 98.2% for head, 99.5% for thorax, and 99.9% for pelvis. Body passing rates were 99.8–100%, while PTV D95 deviations relative to simulation CT remained within ±0.8% for both Methods A and C, demonstrating that representative scatter geometry improves density calibration and subsequent dose calculation.
  • An exploratory analysis applied the 125-kV Method C calibration curve to 140-kV HyperSight images, producing 1%/1 mm body gamma passing rates of 97.97% for head, 96.93% for thorax, and 94.03% for pelvis. No separate 140-kV calibration curve was generated, and the study used phantoms rather than patients, without independent point-dose measurements or respiratory-motion evaluation; therefore, clinical use at other acquisition settings remains unvalidated.

CLINICAL TAKEAWAY

HyperSight CBCT dose-calculation accuracy depends not only on image reconstruction but also on how the CT-density calibration phantom is scanned, particularly when a thin phantom creates nonrepresentative scatter conditions. Solid-water extensions provided the most consistent results in this study, while narrow blade collimation offered a practical alternative; both approaches require site-specific validation before use in clinical adaptive dose calculation.

SOURCE

Radiation Oncology

Browse more research Suggest a correction