IMRT dose verification using the dose uncertainty prediction model

Abstract

Intensity modulated radiation therapy (IMRT) has been introduced to conform to a three-dimensional (3-D) shape of a tumor with a custom tailored radiation dose while protecting surrounding normal tissues. The complex fluence maps in IMRT, however, require a comprehensive patient-specific quality assurance (QA) with two-dimensional (2-D) detectors. In addition, the increased monitor units (MUs) and total number of beam segments potentially lead to escalated dose uncertainty. Assuming the Gaussian distribution, we have developed an uncertainty prediction model using dose level and dose gradient at points of interest. The total dose uncertainty at a point is assumed a square root of quadratic sum of the non-space-oriented and space-oriented uncertainties. To calculate the dose uncertainty, 1% relative dose uncertainty level at prescribed dose level (∼200 cGy) and 1 mm spatial displacement along each Cartesian orthogonal axis were employed as one standard deviation. In this study, the applicability of the dose uncertainty model to IMRT QA was investigated. Total of 10 IMRT fields from five patients (4 head and neck and 1 prostate cases) were generated by Philip Pinnacle treatment planning system. Each IMRT field consisted of 8 to 12 beam segments. Beam measurements were accomplished using MapCHECK (445 diode array 2-D detector) and compared with the planned data. By changing the tolerance levels for dose difference and distance-to-agreement (DTA), the failed points in IMRT dose verification were detected and compared with dose uncertainty of the points. Most failed points remarkably fell on the high dose uncertainty areas. The results clearly showed that the uncertainty model could provide a priori information about high-risk regions during IMRT dose verification. It is expected that the uncertainty model can bring a new paradigm of dose verification in which flexible tolerance values are assigned to different points considering predicted dose uncertainty at each point.