Purpose: To determine the uncertainties in dose volume histogram (DVH) analysis used in modern brachytherapy treatment planning systems (TPSs).
Materials and methods: A phantom with three different volumes was scanned with CT and MRI. An inter-observer analysis was based on contouring performed by 5 persons. The volume of a standard contour set was calculated using seven different TPSs. For five systems a typical brachytherapy dose distribution was used to compare DVH determination.
Results: The inter-observer variability (1SD) was 13% for a small cylindrical volume, 5% for a large cylinder and 3% for a conical shape. A standardized volume for a 4mm CT scan contoured on seven different TPS varied by 7%, 2%, and 5% (1SD). Use of smaller slice thickness reduced the variations. A treatment plan with the sources between the large cylindrical shape and the cone showed variations for D(2cc) of 1% and 5% (1SD), respectively. Deviations larger than 10% were observed for a smaller source to cylinder surface distance of 5mm.
Conclusions: Modern TPSs minimize the volumetric and dosimetric calculation uncertainties. These are comparable to inter-observer contouring variations. However, differences in volume result from the methods of calculation in the first and last slice of a contoured structure. For this situation and in case of high dose gradients inside analyzed volumes, high uncertainties were observed. The use of DVH parameters in clinical practice should take into account the method of calculation and the possible uncertainties.