Purpose/Objective(s):A new Cone Beam CT platform installed on an O-ring linac, with large detector (86x43cm) and reduced acquisition time, is compared with standard CBCT systems installed on C-arm and O-ring linacs. The study characterizes image quality, Hounsfield Units (HU) accuracy and dosimetric performance of this new technology with high potential for offline and online adaptive and image guided radiotherapy.Materials/
Methods: Comparisons between CBCT and Fan Beam CT (FBCT) were performed with pelvic, lung and head anthropomorphic phantoms for default CBCT protocols available on a new CBCT device installed on an O-ring linac and standard imagers coupled with C-arm and O-ring linacs. Different CBCT reconstruction algorithms are compared, including Feldkamp-Davis-Kress (FDK), iterative CBCT (iCBCT) (w/ or w/o Acuros CTS) algorithms. Noise from different ROI on anthropomorphic phantoms is studied. Dose measurements are compared to manufacturer values and between linacs with CTDIw formalism. To compare imager performance in terms of image quality and exposition, an index is proposed: product noise-dose (PND). HU comparisons between FBCT (reference) and CBCT are provided by ME and MAE metrics for different CBCT default protocols (Pelvis, Large Pelvis, Thorax and Head) and reconstruction algorithms. The dosimetric comparison based on anthropomorphic phantoms reports dose deviations (D2% or Dmean) for organs at risk (OAR) selected and (D50%) for target volumes (TV). Results: New CBCT technology combined with iCBCT reconstruction algorithms have the lowest noise values per imager and per protocol, below 10HU for the pelvis and thorax phantoms, and 24HU for brain phantom. Dose measurements show a good agreement with manufacturers index (from -12% to +5% depending on the devices and protocols) and confirm dose reduction with new CBCT technology (around 30 to 50% compared to other CBCT technologies). PND, for the new CBCT and Acuros algorithm (HS+iCBCT Acuros), are from 1.5 to 4.5 lower than PND for other devices and iCBCT reconstruction algorithms. ME and MAE are systematically improved for HS+iCBCT Acuros images for all the protocols and phantoms imaged: ME (Body) <2HU except for Large Pelvis (ME=18HU) and MAE (Body) < 27HU except for brain phantom (MAE=51HU) and significantly for high densities (ME (Femoral Heads) <10HU and MAE (Femoral Heads) < 70HU). Dose deviations are improved or equal (except for Large Pelvis protocol) with new CBCT technology (deviations<0,5% for pelvis and lung phantoms (OAR and TV) and head (only OAR) and <1% for head (TV)) in comparison to standard imagers (range 0.1% up to 2%). Conclusion: Dose and noise are reduced with the new CBCT platform compared to standard imagers and confirmed by PND metric. HU accuracy and dose comparison on anthropomorphic phantoms highlight the clinical potential to use this technology for offline and online adaptive strategies.
