Title:
A Method for Online Dose Prediction for Image Guided Proton Therapy

Thumbnail Image
Author(s)
Lepain, William Andrew
Authors
Advisor(s)
Flampouri, Stella
Niu, Tianye
Wang, C.-K. Chris
Advisor(s)
Editor(s)
Associated Organization(s)
Series
Supplementary to
Abstract
This project was an attempt to design a deterministic algorithm for dose prediction using cone beam CTs (CBCTs) that could output a prediction on the time-scale of pa- tient setup during proton image guided radiotherapy (IGRT). We hypothesized that a com- parative analysis of a reference CBCT (RefCBCT) matching the anatomy of a treatment planning CT (TPCT) and a secondary CBCT (SecCBCT) taken as part of an IGRT setup sequence could be used to consistently and accurately predict the daily dose distribution using the original treatment plan as a guide. In the interest of simplicity and speed of pre- diction, only a small number of simple corrections were applied to the CBCTs to determine their effectiveness, and in an attempt to minimize the number of corrections necessary for accuracy. The corrections used included histogram matching both to the TPCT and the most recent quality assurance CT (QACT), the standard hounsfield unit (HU) to relative stopping power (RSP) curve used in the treatment planning system (TPS), as well as a simple gaussian smoothing to lessen the local effects of CBCT artifacts. The comparison itself involved interpolating each line dose along the beam path from the cumulative corrected hounsfield units (cHU) RefCBCT to the cumulative cHU Sec- CBCT. After all corrections were applied on a sample set of 21 oropharynx patients with 4-5 QACTs each making up 103 prediction sets, an average global gamma passrate of 96.75 % was achieved with an average prediction time of 21.89 s. Predictions were com- pared to dose distributions calculated by the TPS on deformed CTs matching the anatomy of the SecCBCT used in the prediction process. Gamma values were determined using parameters of 3 %/3 mm with a dose cutoff of 2 Gy, or one fraction.
Sponsor
Date Issued
2021-08-02
Extent
Resource Type
Text
Resource Subtype
Thesis
Rights Statement
Rights URI