Abstract
Objectives: To validate the performance of an automatic tool to estimate a patient's peak skin dose (PSD) and a skin dose map from data collected by a radiation dose management system (RDMS) during interventional procedures.Methods: In total, 288 eligible consecutive patients undergoing abdominopelvic embolisation or planned coronary angioplasty using radiochromic films were screened between June 2018 and March 2019. For 98 included patients, PSD was measured using radiochromic films (PSDFilm) and computed by RDMS (PSDRDMS) using one flat and two anthropomorphic phantoms. Statistical concordance between PSDFilm and PSDRDMS was computed with Lin's concordance correlation coefficient and clinical concordance with the Bland and Altman graphic; values were compared using the paired Mann-Whitney-Wilcoxon test.Results: In total, 190/288 patients were excluded and 98 patients were analysed (69 men, mean age 66 ± 14 years). The PSDFilm median (1st; 3rd quartile) was 0.59 Gy (0.40; 1.08). PSDRDMS was 0.62 Gy (0.43; 1.22) for the flat phantom and 0.62 Gy (0.42; 1.19) for anthropomorphic phantoms. The concordance between PSDFilm and PSDRDMS was good for both phantoms (flat: 0.94 [0.91; 0.95]; anthropomorphic 0.94 [0.91; 0.96]). Compared with the values of PSDFilm, the values of PSDRDMS were significantly increased by 5% (- 4%; 16%) for flat phantom (p = 0.001) and 7% (- 6%; 22%) for anthropomorphic phantoms (p = 0.002) for vascular procedures and 9% (- 4%; 26%, p = 0.01) and 6% (- 4%; 23%, p = 0.02) for cardiac procedures, respectively. Dose map representations matched for most patients. The gaps identified were due to table displacement during fluoroscopy events and the use of a wedge filter.Conclusions: The RDMS skin dose map tool allowed the computation of the PSD and skin dose distribution for all patients with fewer constraints than radiochromic films. However, the computed PSD was overestimated, increasing the number of patients requiring follow-up.