The accuracy of CT-based inhomogeneity corrections and in vivo dosimetry for the treatment of lung cancer

Abstract

Purpose: To determine the accuracy of dose calculations based on CT-densities for lung cancer patients irradiated with an anterioposterior parallel-opposed treatment technique and to evaluate, for this technique, the use of diodes and an Electronic Portal Imaging Device (EPID) for absolute exit dose and relative transmission dose verification, respectively. Materials and methods: Dose calculations were performed using a 3-dimensional treatment planning system, using CT-densities or assuming the patient to be water-equivalent. A simple inhomogeneity correction model was used to take CT-densities into account. For 22 patients, entrance and exit dose calculations at the central beam axis and at several off-axis positions were compared with diode measurements. For 12 patients, diode exit dose measurements and exit dose calculations were compared with EPID transmission dose values. Results: Using water-equivalent calculations, the actual exit dose value under lung was, on average, underestimated by 30%, with an overall spread of 10% (1 SD) in the ratio of measurement and calculation. Using inhomogeneity corrections, the exit dose was, on average, overestimated by 4%, with an overall spread of 6% (1 SD). Only 2% of the average deviation was due to the inhomogeneity correction model. The other 2% resulted from a small inaccuracy in beam fit parameters and the fact that lack of backscatter is not taken into account by the calculation model. Organ motion, resulting from the ventilatory or cardiac cycle, caused an estimated uncertainty in calculated exit dose of 2.5% (1 SD). The most important reason for the large overall spread was, however, the inaccuracy involved in point measurements, of about 4% (1 SD), which resulted from the systematic and random deviation in patient set-up and therefore in the diode position with respect to patient anatomy. Transmission and exit dose values agreed with an average difference of 1.1%. Transmission dose profiles also showed good agreement with calculated exit dose profiles. Conclusions: The study shows that, for this treatment technique, the dose in the thorax region is quite accurately predicted using CT-based dose calculations and a simple heterogeneity correction model. Point detectors such as diodes are not suitable for exit dose verification in regions with inhomogeneities. The EPID has the advantage that the dose can be measured in the entire irradiation field, thus allowing an accurate verification of the dose delivered to regions with large: dose gradients.