Advancement Of A 3D Computational Phantom And Its Age Scaling Methodologies For Retrospective Dose Reconstruction Studies

Author

Aashish Gupta, The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical SciencesFollow

Author ORCID Identifier

https://orcid.org/0000-0002-0178-0921

Date of Graduation

8-2021

Document Type

Thesis (MS)

Program Affiliation

Medical Physics

Degree Name

Masters of Science (MS)

Advisor/Committee Chair

Rebecca M. Howell, Ph.D.

Committee Member

Stephen F. Kry, Ph.D.

Committee Member

Choonsik Lee, Ph.D.

Committee Member

Peter A. Balter, Ph.D.

Committee Member

David S. Followill, Ph.D.

Committee Member

James P. Long, Ph.D.

Abstract

We have used a 3D age-scalable computational phantom for over two decades for retrospective dose reconstruction studies of childhood cancer survivors (CCS) treated with 2D historic radiotherapy (RT). However, our phantom and its age scaling functions (ASF) must be updated so that it can be used in studies that include survivors treated with contemporary RT. We aimed to implement our phantom and its age scaling functions in DICOM format and determine the feasibility of applying our ASFs to accurately scale the whole-body CT-based anatomies.

In the implementation study, we developed Python scripts that model the phantom and ASFs in a treatment planning system (TPS). We validated the implementation by comparing several geometric and anthropometric parameters with reference datasets. We then conducted a dosimetric analysis to determine the accuracy of dose calculation using our phantom. In the feasibility study, we downscaled various computed tomography (CT)-based phantoms from the University of Florida/National Cancer Institute (UF/NCI) phantom library to arbitrary ages. We quantified the geometric accuracy of scaling by comparing several overlaps, distance, and anthropometric parameters of the scaled phantom with reference datasets. We also assessed the dosimetric impact of ASFs by quantifying the difference in dose from standard Wilms’ tumor RT plan simulated on exact age-scaled and nearest age-matched phantom while using the same field size and anatomical landmark dependent field size in two different scenarios.

This study showed that phantoms were implemented in DICOM format within 3% of points/volume of our original phantoms. The heights and dosimetric accuracy were within 7% of ground-truth values. In the feasibility study, overlap metrics showed “good” agreement for most cases except pancreas and kidneys. The maximum displacement of 4.1cm was obtained in the scaled liver. In both implementation and feasibility studies, organ masses were smaller than reference masses in general. A difference of 6% and 1.3Gy was obtained for percent volume ≥ 15Gy (V15) and mean dose, respectively, across two phantom categories when the same field size was used. Both metrics were significantly different (p<0.05) for partially in-beam organs when field size varied. Overall, our results show that phantom and ASFs can be accurately used in TPS for modern RT studies, and our ASFs can accurately scale whole-body CT-based anatomy.

Keywords

Computational Phantoms, Late Effects, Pediatric Phantoms, Dose Reconstruction, Phantom Scaling, Scaling Factors