Voxel-Based Monte Carlo Estimation of Absorbed and Effective Radiation

Voxel-Based Monte Carlo Estimation of Absorbed and Effective Radiation Dose in Lung Tissue from ¹⁸F-FDG PET/CT Using OpenGATE

Farag Mahmoud Ali                            ¹ Aya Alhadi Ermeelah²

farag.mahmoud@sabu.edu.ly             ¹ aya.armeelah@sabu.edu.ly²

https://orcid.org/0009-0004-0358-1194                 ¹ https://orcid.org/0009-0008-2125-3679²

Faculty of Pharmacy Sabratha University¹                Faculty of Science Sabratha University²  

https://doi.org/10.5281/zenodo.19801043 

Abstract:

Background: Working out precise, truly patient-specific radiation doses from ¹⁸F-FDG PET/CT scans remains a genuine headache in practice. Anatomy differs so much between individuals, and the standard reference phantoms we often turn to simply don't account for that level of variation very well. In our case, we wanted to try something more grounded: a voxel-by-voxel Monte Carlo approach built in OpenGATE, drawing directly from real clinical CT data to estimate doses for a routine 370 MBq ¹⁸F-FDG injection in a thoracic region that actually reflects a patient's body.

Methods: We pulled an anonymized chest CT from The Cancer Imaging Archive (TCIA), specifically the NSCLC-Radiomics set [24]. Converting it to a voxel phantom involved segmenting into six tissue classes (air, lung, fat, soft tissue, cartilage, bone) using Hounsfield Unit cutoffs following Schneider's method [27].

To define the source, we kept the ¹⁸F-FDG activity uniform but restricted it inside an eroded body outline this helped cut down on positron escape artifacts that can mess up boundary results. Simulations used one million primary positrons with the QGSP_BIC_EMY physics list. Scaling by 33,208 brought it up to clinical activity levels, and we layered on a rough correction factor of ~250× to roughly compensate for photon escape and the non-uniform uptake seen in real scans.

We then extracted organ absorbed doses, computed effective dose via ICRP 103 weighting factors [18], and looked at dose-volume histograms (DVHs) to assess how evenly (or unevenly) the dose spreads through tissues.

Results: With corrections applied, effective dose landed at 5.85 mSv comfortably inside the 5–9 mSv ballpark from ICRP Publication 128 for standard ¹⁸F-FDG exams [21]. It felt reassuring to see it align so closely. Tissue-specific means came out to about 22.58 mGy in lung, 24.97 mGy in soft tissue, and 18.39 mGy in compact bone. For the lung DVH, D95 was 3.45 mGy and V20 reached 24.9%, suggesting reasonable coverage but with the kind of heterogeneity you'd naturally expect in real anatomy.

The erosion step worked well only 1.12% of dose leaked outside the body afterward. Statistical uncertainties stayed below 1% across major tissues, which gave us confidence in the numbers.

Conclusion: In summary, this work presents a reproducible Monte Carlo workflow for ¹⁸F-FDG PET dosimetry using Open GATE and real patient CT data. The calculated effective dose of 5.85 mSv and lung absorbed dose of 22.58 mGy align well with established ICRP reference ranges, validating the methodology and correction approach. This provides a reliable educational baseline with clear potential for refinement in future patient-specific studies.

Keywords: ¹⁸F-FDG; PET/CT; Monte Carlo simulation; Open GATE; patient-specific dosimetry; effective dose; dose-volume histogram; voxelized phantom; lung dosimetry