Abstract
Tomographic computational models, based on regional segmentation of CT or MRI medical images, have increasingly been proposed as replacements for current stylized mathematical models of human anatomy used in radiation dosimetry studies. While much effort has been devoted towards the creation of adult models, few research studies have been initiated to address the need for models supporting paediatric radiology dosimetry. In this study, two tomographic models were created using a combination of automatic and manual segmentation via a program created in-house using IDL version 5.5. The first model is of a normal 6 day female newborn, and consists of a 512 × 512 × 485 data array. The CT slices of this model were obtained every 1 mm, and 66 different anatomic regions were defined. The second model is of a patient developmentally equivalent to a 2 month male, and was saved as a 512 × 512 × 438 data array. This subject had severe tissue oedema within the gut, kidneys, liver and spleen. The model is thus considered representative of a critically ill child, from a patient subpopulation expected to receive a larger than normal number of diagnostic x-ray exams. The voxel volumes for the two models are 0.35 mm3 and 0.30 mm3, respectively, thus making these models the most detailed in existence for paediatric dosimetry applications. Ratios of organ masses for the UF newborn model to those within the reference ORNL/MIRD model range from highs of 1.2 to 1.7 for the oesophagus and small intestine/colon, respectively, to lows of 0.18 to 0.27 for the mandible and humeri, respectively. For the UF 2 month model, ratios of organ masses in the UF model to those in the 8 week GSF BABY model ranged from highs of 3.7 to 5.2 for the clavicles and spleen, respectively, to lows of 0.2 to 0.3 for the adrenals and scapulae, respectively.