A dose-volume histogram based decision-support system for dosimetric comparison of radiotherapy treatment plans

Abstract

Background: The choice of any radiotherapy treatment plan is usually made after the evaluation of a few preliminary isodose distributions obtained from different beam configurations. Despite considerable advances in planning techniques, such final decision remains a challenging task that would greatly benefit from efficient and reliable assessment tools.

Methods: For any dosimetric plan considered, data on dose-volume histograms supplied by treatment planning systems are used to provide estimates on planning target coverage as well as on sparing of organs at risk and the remaining healthy tissue. These partial metrics are then combined into a dose distribution index (DDI), which provides a unified, easy-to-read score for each competing radiotherapy plan. To assess the performance of the proposed scoring system, DDI figures for fifty brain cancer patients were retrospectively evaluated. Patients were divided in three groups depending on tumor location and malignancy. For each patient, three tentative plans were designed and recorded during planning, one of which was eventually selected for treatment. We thus were able to compare the plans with better DDI scores and those actually delivered.

Results: When planning target coverage and organs at risk sparing are considered as equally important, the tentative plan with the highest DDI score is shown to coincide with that actually delivered in 32 of the 50 patients considered. In 15 (respectively 3) of the remaining 18 cases, the plan with highest DDI value still coincides with that actually selected, provided that organs at risk sparing is given higher priority (respectively, lower priority) than target coverage.

Conclusions: DDI provides a straightforward and non-subjective tool for dosimetric comparison of tentative radiotherapy plans. In particular, DDI readily quantifies differences among competing plans with similar-looking dose-volume histograms and can be easily implemented for any tumor type and localization, irrespective of the planning system and irradiation technique considered. Moreover, DDI permits to estimate the dosimetry impact of different priorities being assigned to sparing of organs at risk or to better target coverage.

Fig. 1

Schematic representation of dose-volume histograms. a A DVH corresponding to a PTV and b an OAR or the RVR. The factors involved in the formulation of I _T, I _O and I _R are also indicated, see (1), (2) and (3) respectively

Fig. 2

Plots of the DDI scores for each set of three tentative plans corresponding to the fifty brain cancer patients considered. Applied treatment plan in red, and discarded tentative plans in green and purple. a 25 centrally located benign tumors. b 15 peripherally located benign and malignant tumors. c 10 centrally located malignant tumors. Mean DDI scores (black solid lines) and standard deviations (blue solid lines) are also represented

Fig. 3

Comparison of scoring results for the cases where ATP and DDIP did not coincide. The mean I _T and joint I _O and I _R defined as (I _O+I _R)/2 scores and standard deviations are provided for the ATPs (label AP) and remaining tentative plans (label RP)

Fig. 4

DVHs corresponding to the PTV (a), OARs (b) and RVR (c) for the three tentative plans of a patient diagnosed with a centrally located Meningioma. The D _p on the PTV was set equal to 50.4 Gy in 28 daily fractions (5 days per week) of 1.8 Gy