Semester of Graduation

Spring

Degree

Master of Science (MS)

Department

Physics and Astronomy

Document Type

Thesis

Abstract

Magnetic resonance-guided adaptive radiation therapy (MRgART) is a promising approach for improving the precision of prostate cancer treatment. A key step in MRgART is generating a synthetic CT (synCT) from MR images to enable accurate dose calculations. However, uncertainties introduced by the use of synCTs must be quantified to ensure safe and effective treatment. This study investigates dosimetric differences introduced by two synCT generation methods, bulk density overrides and artificial intelligence (AI)-generated pseudo-CTs (pCTs), in the context of prostate MRgART. Dose differences were evaluated for both single fractions and a five-fraction treatment course to assess cumulative impact.

Specific Aim 1 assessed baseline dose differences introduced by BDOs. A conventional CT scan was compared to the same scan modified with a BDO, where mean electron densities were homogeneously assigned to each anatomical structure. Identical treatment parameters were used for both plans, and dose volume histograms (DVHs) were analyzed for key organs included the prostate, bladder, rectum, penile bulb, and femoral heads.

Specific Aim 2 focused on evaluating the dosimetric differences between treatment plans calculated on BDOs from the MR image set and AI-derived pseudo-CT (pCT). For this analysis, pseudo-CTs were generated using TheraPanacea’s AI algorithm, and dose calculations were compared for the same anatomical structures across both image sets. This comparison assessed how accurately pCTs replicate the dosimetric characteristics of density override plans to identify potential sources of uncertainty in single fraction treatments.

Specific Aim 3 extended the dosimetric analysis from Aim 2 by evaluating cumulative dose differences over a five-fraction treatment course. Dose accumulation was performed for the BDO and pCT image sets, allowing for comparison of the total calculated dose to each anatomical structure across the entire treatment. This analysis aims to assess whether dose deviations between BDO and pCT plans increase over the course of an adaptive radiation therapy regimen, thereby highlighting potential long-term uncertainties in dose delivery.

Together, these aims provide a comprehensive evaluation of the dosimetric accuracy of synCTs in prostate MRgART. The results establish a baseline for understanding BDO-induced dose differences and assess the reliability of AI-generated pCTs for both single and multi-fraction treatments.

Date

4-4-2025

Committee Chair

Schneider, Christopher

Available for download on Thursday, March 25, 2032

Included in

Physics Commons

Share

COinS