Monte Carlo simulation of skin dose distribution from interventional cardiology procedures

Abstract

Purpose: To determine the accuracy of skin dose maps calculated by the openSkin open source software contained within version 0.10.0 of the OpenREM patient dose management system (OpenREM 0.10.0). The maps are designed to be representative of the distribution of absorbed dose to the patient skin from x-ray fluoroscopy procedures. Methods: TRADER, a Tool for RADiology Education and Research, has been developed using the Geant4 Monte Carlo toolkit and validated using American Association of Physicists in Medicine Task Group 195 test case data and physical measurements of air kerma using a Siemens Axiom Artis zee interventional cardiology x-ray system. The composition of the TRADER table and mattress have been reverse-engineered to match the transmission properties of the Artis zee table and mattress using physical transmission measurements. TRADER was used to determine the geometric and dosimetic accuracy of OpenREM skin dose maps generated from ten synthetic and seventeen clinical Artis zee studies. Peak skin dose values in OpenREM and TRADER skin dose maps were compared. For three synthetic and clinical studies TRADER was used to determine skin dose to the International Commission on Radiological Protection Publication 110 (ICRP110) male and female reference computational phantoms and the results compared to those obtained using the openSkin phantom. Results: OpenREM 0.10.0 skin dose maps contain geometric and dosimetric errors. The geometric error affects exposures with a non-zero caudocranial angle: these exposures have incorrect field size and shape at the phantom surface. The dosimetric error causes all exposures to have a 5 by 5 cm field size backscatter factor applied whatever the actual field size at the phantom surface. An updated version of the openSkin code has been produced which addresses these issues (OpenREM-up) and revised skin dose maps calculated. For the synthetic studies the differences in peak skin dose between OpenREM and TRADER maps are -27.2 - 5.1% for OpenREM 0.10.0 and -11.6 - 6.2% for OpenREM-up. For the clinical study skin dose maps the differences in peak skin dose compared with TRADER are -37.6 - 28.5% for OpenREM 0.10.0 and -3.7 - 19.1% for OpenREM-up. Peak skin dose values in the ICRP110 phantoms are within 6% of openSkin phantom values for all three clinical studies and two of the synthetic studies. For the third synthetic study containing lateral exposures differences of up to 37% in peak skin dose are seen between the ICRP110 and openSkin phantom maps which are shown to be partly due to the different physical size and shape of the phantoms. When the different focus to skin distances are accounted for the maximum difference in PSD was 11.9%. Conclusions: OpenREM 0.10.0 skin dose maps contain geometric and dosimetric inaccuracies. Peak skin dose values for OpenREM 0.10.0 differ from TRADER by between -37.6 - 28.5%. Caution should be exercised when using skin dose map results from OpenREM 0.10.0 and earlier versions of OpenREM. The geometric and dosimetric OpenREM 0.10.0 errors have been addressed in OpenREM-up. Skin dose maps calculated with this updated code differ from TRADER by -11.6 - 19.1% and are suitable to guide clinical decisions on patient skin care and follow up. The simple openSkin phantom is sufficient to estimate peak skin dose from clinical interventional cardiology x-ray studies.

Date of Award	1 Aug 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Daniel Shaw (Supervisor) & Michael Taylor (Supervisor)