Room: ePoster Forums
Purpose: For over two decades, we have used an in-house computational human phantom for large cohort radiotherapy-related late effects studies, hereafter referred to as 2D phantom. Here, we aimed to (1) develop a 3D model of our phantom within a commercial treatment planning system (TPS), (2) scale the 3D phantomâ€™s body regions and organs to different ages, (3) develop organ auto-contouring for the age-scaled phantoms, and (4) compare the volumetric and geometric information with pediatric NCI reference phantoms.
Methods: A program was written to generate a 3D model of our phantom in RayStation TPS. Our phantom is generically defined as an 18-year old that can be scaled to any age between newborn and 19 years using non-uniform growth age-scaling functions (based on Society of Automotive Engineers data). We translated the age-scaling functions to the 3D phantom using python scripting within the TPS. A Convex Hull algorithm was used to convert age-scaled phantomsâ€™ organs from grids of points to 3D contours. Lastly, we calculated the normalized mean square distance between the organs in our 2D and 3D phantoms for ages 1, 5, 10, and 15 years. Our 3D phantom models were registered with NCI reference phantoms of the same age.
Results: Between our 2D and 3D phantom models, the average error for 4 organs of 4 ages was Â±3.15 mm (Â±0 to Â±9.3 mm). The right lung and brain showed a higher average error of Â±7.12 mm and Â±4.80 mm. Organs showing maximum and minimum percent difference in volume between our 3D phantoms and NCI phantoms were left lung and liver of 1 year old, 84.04% and 25.04%, respectively.
Conclusion: We developed a 3D model of our computational phantom and validated it by comparison with previous 2D model. This new phantom can be broadly used and allows 3D visualization of phantom/organs