Absolute and relative dose distribution comparisons for convolution/superposition and Monte Carlo based treatment planning

Historically, the algorithms used for dose computation in radiotherapy treatment planning (RTP) have been based on measured data in water. The so-called model based algorithms (convolution, Monte Carlo) are now emerging as the dose engines of choice for 3D RTP as they can predict more accurately the dose distribution inside the patient based on the CT anatomy with minimum measured data input. In this work, we studied the effect of the dimensionality of a convolution/superposition dose algorithm on the absolute dose and relative dose distribution computed in several clinical cases and compared the outcome to Monte Carlo calculations. The convolution algorithm, calculates the dose at a point by summing together the total energy released per unit mass (TERMA) at all primary interaction sites as modified by the convolution kernel; the latter, reflects the percent of the energy released that is absorbed at the dose deposition site. Patient tissue inhomogeneity can be (i) ignored, (ii) included in the TERMA calculation only and (iii) included in both the TERMA and the convolution kernel. The resulting isodose distribution and monitor units correspond then to a homogeneous, 2.5D and 3D calculation type respectively. We used four clinical cases to study the dimensionality of the dose engine and compare to MC. We found remarkable differences between the three convolution calculation modes, but not much difference against the MC computations. The dosimetric and clinical implications in the choice of the algorithm are presented as applied to the clinical sites that were investigated