Session Item

Standard gamma-analysis for individual plan QA can be difficult to interpret and its clinical relevance is doubtful. Deviations from clinical objectives are readily understandable by DVH-based portal transit dosimetry. We aim to demonstrate the feasibility of our in-house developed portal dosimetry solution for 3D dosimetric verification of extreme hypofractionated VMAT treatments, using DVH metrics.

Energy fluency measured with an aS1200 Varian EPID is used by our EPICoreMedPhys (ECMP) software to perform 3D dose verification without (pre-treatment) and with a phantom (in phantom) or patient in the beam (in vivo). ECMP reconstructs 3D dose using a Monte Carlo code (XVMC) and compares it to the 3D TPS dose (AAA, Eclipse v15.6). The in phantom tests were performed using the ArcCheck cylindrical phantom. 30 hypofractionated (D_fx>4Gy) VMAT plans (13 Prostate, 5 Bone, 4 Brain, 4 Lymph Nodes, 2 Gyneacologic, 1 Pancreas and 1 Lung) with 6 and 10 FFF beams were included. To study the influence of plan complexity, the mean distance between opposing MLC leaves (mdMLC) was calculated for each plan and correlated with median dose differences (ΔD50) between planned and reconstructed dose distributions for the VOI defined by the planned 50% isodose volume (VID50). For in vivo measurements, we selected 12 prostate plans (5×9Gy) treated with 10 FFF beams, yielding 32 fractions in vivo. Average DVH differences (ΔD2, ΔD50 and ΔD98) for the VID50, targets and OARs were determined.

Table 1.A shows our pre-treatment and in phantom results. Substantial dose differences were found for mdMLC <1cm with highly modulated fields, yielding an average ΔD50 for VID50 of -4.7 ± 2.8% and -2.4 ± 1.7%, respectively, which can go (partly) unnoticed when using gamma passing rates. Geometric and dosimetric calibration of these very small leaf distances is complicated in many aspects (LINAC, dosimeters, TPS, ECMP, etc.), so for further analysis we concentrated on 12 hypo-fractionated prostate treatments with mdMLC of 1-2cm. Table 1.B shows smaller systematic dose differences (<1% underdose) for in phantom prostate measurements, still explainable by minimal errors in leaf settings like the dosimetric leaf gap (DLG). For in vivo measurements, mean underdosage increases up to 5%, except for the urethra wall that is enclosed by the PTV (Fig.1). Apart from tissue inhomogeneities, anatomical changes and set-up errors that take effect during patient’s treatments, we are currently investigating whether the extra underdose might stem from residual uncertainties in the transit EPID dose conversion itself and from differences in 3D dose calculation algorithms (XVMC vs AAA).

Our portal dosimetry 3D dose deviations increase with decreasing leaf distances used in complex RT plans. The very small segments can cause true underdosages due to calibration inaccuracies but might also decrease the accuracy of the portal dosimetry model that needs adjustment.