Comparison of organ and effective dose estimations from different Monte Carlo simulation‐based software methods in infant CT and comparison with direct phantom measurements

Abstract

Computational dosimetry software is routinely used to evaluate the organ and effective doses from computed tomography (CT) examinations. Studies have shown a significant variation in dose estimates between software in adult cohorts, and few studies have evaluated software for pediatric dose estimates. This study aims to compare the primary organ and effective doses estimated by four commercially available CT dosimetry software to thermoluminescent dosimeter (TLD) measurements in a 1-year-old phantom. Methods: One hundred fifteen calibrated LiF (Mg, Cu, P)-TLD 100-H chips were embedded within an anthropomorphic phantom representing a 1-year-old child at positions that matched the approximate location of organs within an infant. The phantom was scanned under three protocols, each with whole-body coverage. The mean absorbed doses from 25 radiosensitive organs and skeletal tissues were determined from the TLD readings. Effective doses for each of the protocols were subsequently calculated using ICRP 103 formalism. Dose estimates by the four Monte Carlo-based dose calculation systems were determined and compared to the directly measured doses. Results: Most organ doses determined by computation dosimetry software aligned to phantom measurements within 20%. Additionally, comparisons between effective doses are calculated using computational and direct measurement methods aligned within 20% across the three protocols. Significant variances were found in bone surface dose estimations among dosimetry methods, likely caused by differences in bone tissue modeling. Conclusion: All four-dosimetry software evaluated in this study provide adequate primary organ and effective dose estimations. Users should be aware, however, of the possible estimated uncertainty associated with each of the programs.

Cumulative radiation dose from medical imaging in paediatric congenital heart disease patients with epicardial cardiac implantable electronic devices

“… 16 Computational phantoms (such as VirtualDose, CT-Expo, and ImPACT) incorporate radiation transport codes with an anatomic model of the patient to account for dose distribution and are increasingly used for ED calculations. 17 CT doses calculated using the publicly available National Cancer Institute (NCI) dosimetry software have been shown to align with TLD values within 8–20%, particularly in infants and children, 18–20 and produce similar results to other virtual phantoms. 21 Therefore, for this retrospective study, all ED doses were calculated using the NCI dosimetry software.…”

Rapid estimation of patient‐specific organ doses using a deep learning network

“…In the past decade, Monte Carlo based dose estimation techniques have increasingly been applied as more computing power became available. [4][5][6][7] However, Monte Carlo computations are often time-consuming, laborious to properly setup, and often need specific knowledge of the Monte Carlo software at hand. This can be challenging to clinics, especially to those with high patient throughput, where a rapid workflow is required.…”

“…Although these methods clearly depict dose distributions from actual measurements, they fall short on doing so on real patients rather than phantoms. In the past decade, Monte Carlo based dose estimation techniques have increasingly been applied as more computing power became available 4–7 . However, Monte Carlo computations are often time‐consuming, laborious to properly setup, and often need specific knowledge of the Monte Carlo software at hand.…”