Validation of a TOPAS cone beam computed tomography (CBCT) Monte Carlo model towards personalised CBCT dosimetry in...

Accurately estimating patient-specific radiation dose from cone beam CT (CBCT) in interventional radiology (IR) is essential for assessing exposure to radiosensitive organs. This study aimed to evaluate the accuracy of Monte Carlo (MC) dosimetry models for two CBCT IR systems, the Philips Azurion and the Siemens Artis Q, using a realistic anthropomorphic phantom and its CT images.

MC simulations of abdominal CBCT protocols were performed using the CIRS® ATOM phantom. For the Philips system, manufacturer-supplied exposure parameters were used. For the Siemens system, a previously developed AECScorer module was implemented to account for automatic exposure control (AEC), modulating simulated projections based on phantom attenuation. Experimental organ doses were measured using thermoluminescent dosimeters (TLDs) and compared against MC results.

The Philips MC model showed good agreement with TLD data, with an average difference of 14 % for organs in the central field of view (FOV). Simulated organ doses were consistent with published literature, but underestimated empirical values in some cases. The Siemens model, while reasonably accurate, overestimated organ doses by an average of 11 %. For both systems, the largest discrepancies were observed in the peripheral FOV. On average, Siemens experimental doses were 3.45 times higher than those from the Philips system.

Results of this study suggest that MC models, informed by manufacturer-provided CBCT acquisition parameters or appropriate AEC MC simulation, can reliably replicate experimental dosimetry in the irradiated volume with good accuracy. This supports the feasibility of scanner-, region-, and exam-specific dose estimations using the proposed MC methodologies.