Purpose: To scale whole-body computed tomography (CT) based anatomy non-uniformly in 3D to provide age-specific phantoms for retrospective dosimetry in late-effects studies. This is necessary because the University of Florida/National Cancer Institute (UF/NCI) computational phantom series (the gold-standard pediatric phantoms) are limited to discrete ages which could result in uncertainty in dose reconstruction when discrete-aged phantoms are used instead of age-specific phantoms.
Methods: Fourteen UF/NCI phantoms (52 organs in each) were downscaled from their nearest older discrete age (1, 5, 10, 15, and 30 years old) using our in-house body-region-specific age scaling factors which accounts for non-uniform growth of head, neck, trunk, and legs in 3D. To validate the scaling, percent change in height, Dice Similarity Coefficients (DSC), mean distance to agreement (MDA), and Euclidean distance (ED) were calculated between the original and scaled anatomy for whole-body, brain, heart, liver, pancreas, and kidneys. Typical 6MV Wilms’ tumor plans were constructed on a downscaled 3.9-year-old (median age of Wilms’ tumor cohort from the Childhood Cancer Survivor Study) and 5-year-old male phantoms to quantify differences in V10, mean dose, D1, D50, and D95 for in-/near-beam organs.
Results: For whole-body, brain, and heart, the DSC, MDA, and ED were ≥0.61, ≤1.23cm, and ≤2.90cm, respectively. For liver, pancreas, and kidneys, the DSCs fell below 0.50 with MDA and ED up to 2.29cm and 4.09cm, respectively. Lower agreements were attributed to higher variation in organs’ shape across the age range. The heights of phantoms were within 5% of reference heights. In the dose study, a maximum V10 difference of 0.5% was obtained for the pancreas (near-beam organ). Absolute dose differences in organ-to-organ dose comparisons were within 20cGy for each metric.
Conclusion: CT based anatomy can be scaled to any arbitrary age with accurate anatomical shape, size, and position and accurate dose reconstruction for late-effect studies.