Toward Clinically Intuitive Quality Assurance

Abstract

The need for clinically intuitive quality assurance procedures has been well-documented; current QA methods such as 2D gamma analysis have been shown (Nelms, Zhen et al. 2011) to be inadequate in predicting clinically relevant errors. This thesis investigates the accuracy of a novel "transform method" (Oldham, Thomas et al. 2012) which claims to create "measured" patient dose-volume histograms (DVHs) through the use of 3D dosimetry techniques; a measured 3D phantom dose distribution is "transformed" back onto the patient geometry, enabling a clinically relevant analysis through the DVHs. The transform method was tested by inducing a series of known mechanical and delivery errors onto simulated measurements of six different head-and-neck treatment plans; the accuracy of this method was then examined through the comparison of the transformed patient dose distributions and the known actual patient dose distributions through dose-volume histogram metrics and normalized dose difference analysis (Jiang, Sharp et al. 2006). Through these metrics, the transform method was found to be highly accurate in predicting measured patient dose distributions for these types of errors. Further work is needed to investigate other types of errors, such as beam model errors and treatment sites of great inhomogeneity, such as the lung.