Realistic extension of partial-body pediatric CT for whole-body organ dose estimation in radiotherapy patients.

While modern radiotherapy treatments can deliver a localized radiation dose to the tumor, healthy tissues at distance are inevitably exposed to scatter radiation that has been linked to late health effects such as second cancers. Quantifying the radiation dose received by tissues beyond the target is critical for research on such late health effects. However, the typical radiotherapy planning CT only covers part of the body near the target and the organs of interest for late effects research are not always included. Therefore, the purpose of this study was to develop a method for extending a partial-body pediatric CT scan for estimating organ doses beyond the original CT scan range. Our method uses a library of CT images for 359 pediatric patients from which a candidate patient is selected for providing surrogate anatomy. The most appropriate surrogate patient images to use for the extension are determined based on patient demographic information pulled from the image metadata. Image registration is performed through comparison of the patients' skeletons. The images showing closest similarity are adapted by a transformation method and appended to the original partial-body CT and a new structure file containing organ contours is written; we refer to this extended CT scan with organ contours as the Anatomical Predictive Extension (APE). To test the APE method, three patients with nearly full-body anatomy were extracted from the library, and a continuous subset of the images was removed to simulate a partial-body CT. The APE method was then applied to the partial-body CT to create extended anatomies, with the original images serving as ground truth. Radiotherapy plans were simulated using the Monte Carlo code XVMC on both the original and APE anatomies, with the original serving as ground truth. Three pediatric radiotherapy cases were considered for performance testing: (1) head CT for a simulated brain tumor extended to chest; (2) superior chest CT for simulated Hodgkin's lymphoma extended to inferior chest; (3) pelvic CT for Wilms tumor extended to superior chest. Three geometric metrics (Dice similarity coefficient, overlap fraction, and volume similarity) were calculated to quantify the differences between the original patient and the extended anatomies. In all cases, calculated organ doses showed good agreement between the original and APE anatomies. The average absolute relative dose difference across all organs considered for the three cases was 11%, 12% and 15%, respectively. The APE method is useful for estimating radiation doses to peripheral organs in support of research on late effects following radiotherapy.