Residual dose deviation differential histogram analysis using evolutionary-optimized transform parameters for dose distribution warping in patient-specific quality assurance in external beam radiation therapy

Abstract

Patient-specific quality assurance in external beam radiation therapy is performed to ensure the safe and effective administration of ionizing radiation to patients. This individualized quality assurance process often involves the measurement of a planar radiation dose distribution obtained during irradiation of a surrogate phantom, in order to sample the actual delivered doses to the patient; this measured dose distribution is then compared with that calculated by the radiation therapy treatment planning software. The current work, a continuation of the authors' previous efforts, uses a covariance matrix adaption-evolutionary strategy approach to optimize transform parameters of an affine transformation which includes normalization, translation, dilation (or erosion), roll, pitch, and yaw, such that the application of the transformation warps the measured dose distribution to the planned dose distribution. The deviations, from unity, of the transformation parameters associated with translations and with scaling, and the deviations, from zero, of those associated with rotations, are a means of assessing possible errors in experimental setup, as well as possible errors in the dose calculation algorithm used in the treatment planning system. The optimization selects a Pareto-optimal set of eight transformation parameters that minimizes the difference, in the sense of the sum of least squares, between the doses at distinct points in the calculated and measured dose distributions. The method was tested using four distinct treatment plans, and the results indicate that it may prove useful in the clinic.