PubMedJournal of applied clinical medical physics2026-09-10
Towards adaptive radiotherapy: Comparing HyperSight and standard CBCT reconstructions on a C-Arm Linac.
Clarke Niamh L NL, Hughes Jeremy L JL, Wanigaratne Derrick D, Yeo Adam U AU
Cone-beam computed tomography (CBCT) is integral to image-guided radiation therapy and increasingly used for adaptive radiotherapy and CBCT-based dose calculation. Conventional kV-CBCT systems on C-arm TrueBeam linacs with Feldkamp-Davis-Kress (FDK) reconstruction have historically been limited by reduced Hounsfield Unit (HU) accuracy, scatter contamination, image noise and reconstruction artefacts. Iterative CBCT (iCBCT) reconstruction has improved image quality and HU consistency relative to the conventional FDK reconstruction. HyperSight represents a further evolution through software and hardware development with an enhanced detector design, advanced reconstruction algorithms and gantry rotation speeds up to 9°/s.
This study reports commissioning and evaluation of HyperSight iterative CBCT (HS-iCBCT) acquired on a Varian TrueBeam (v4.1), to support adaptive workflows in comparison with standard (Std-FDK) and iterative CBCT (Std-iCBCT) on a TrueBeam (v2.7).
CBCT images were acquired on a Varian TrueBeam using the conventional kV On-board Imager (OBI) for Std-FDK and Std-iCBCT reconstructions and the HyperSight kV-imaging system for HS-iCBCT. Commissioning included CBCT dose calculation protocol optimization, CT-to-electron density (CT-ED) calibration and validation using STEEV, Rando and the CIRS Dynamic Thorax phantoms. CBCT images of anthropomorphic Head, Thorax/Spine, Rib and Lung phantoms were rigidly registered to the planning CT (pCT) and resampled using Velocity AI (v4.2). Voxel-wise HU difference histograms were generated, with peak position used to quantify systematic HU bias. Treatment plans (20 Gy/1# for Brain SRS/Spine, 24 Gy/2# for Rib and 54 Gy/3# for Lung) were optimized on the pCT and recalculated on the CBCT datasets. Dose-volume histogram (DVH) metrics were assessed and 3D-gamma analysis (2%/1 mm to 1%/1 mm; 10%/70% dose thresholds) was performed and compared with Std-FDK and Std-iCBCT.
HS-iCBCT showed improved HU accuracy compared with Std-iCBCT and Std-FDK when benchmarked against the pCT. For materials ≤ 1.08 g/cm3, the mean HU difference was lowest for HS-iCBCT at 7.7 ± 4.1 HU, compared with 23.6 ± 13.5 HU for Std-iCBCT and 38.0 ± 19.7 HU for Std-FDK (Friedman p < 0.05). For higher-density materials, HS-iCBCT demonstrated mean HU differences of 33.5 ± 22.9 HU, compared with 46.2 ± 22.2 HU for Std-iCBCT and 93.4 ± 57.6 HU for Std-FDK. HS-iCBCT improved image quality with HU difference histograms demonstrating peak values closest to zero (∆HU within 15HU). An optimized CBCT acquisition protocol (140 kV) combined with a single baseline 120 kV pCT HU-ED calibration curve proved feasible for CBCT-based dose calculation. Dose calculation agreement was strongest for HS-iCBCT (p < 0.05) with median target dose differences measuring < 0.3%, < 0.6%, 0.6% and < 0.4% for the Brain, Thorax/Spine, Rib, and Lung anatomies, respectively. Median (min-max) gamma passing rates across all anatomical sites were 98.8% (97.8-99.8), 98.7% (96.7-99.6), and 96.0% (94.5-99.5) with 2%/1 mm passing criteria for HS-iCBCT, Std-iCBCT, and Std-FDK, respectively.
HyperSight reduces HU variability, improves image quality, and enhances CBCT-based dose calculation relative to conventional CBCT implementations. Use of a single optimized CBCT dose calculation protocol and baseline HU-ED calibration curve proved feasible across multiple anatomical sites, supporting implementation of HS-iCBCT for adaptive radiotherapy workflows on conventional C-arm TrueBeam linacs.