Dosimetry of small beams is challenging given their small size compared to the detectors, high dose gradient and the lack of lateral electronic equilibrium. The Ph.D. thesis aims to improve the accuracy of the dose delivered to the patient in stereotactic radiotherapy.
On the one hand, dosimetric data used to calibrate the treatment planning system (TPS) were determined using numerical simulations. To achieve this, two CyberKnife radiotherapy facilities were modelled using the PENELOPE Monte Carlo code. Output ratios measurements were performed with several active detectors and with two passive dosimeters (radiochromic film and micro-LiF) and compared with output factors calculated by simulation. Six detectors were modeled in order to study the detectors response in small beams. Among the detectors studied, only the radiochromic films were in agreement with the simulations, they can be used without correction factors. The disturbance of the detectors response in small beams was explained either by the volume effect induced by the active volume, which is too high compared to the beam size, or by the mass density effect induced by the detector body materials which are too far from water mass density. The correction factors, required to correct the disturbance caused by the non-water-equivalence and/or the low spatial resolution of each detector, were calculated for the two CyberKnife systems.
On the other hand, a 2D dose measurement protocol using radiographic films and a MatLab program were developed. Stereotactic treatment plans were then performed for a phantom in order to assess the calculation algorithms implemented in the MultiPlan TPS (associated with the CyberKnife system). The analysis of the 2D dose distributions related to the stereotactic treatment plans has shown that the “Pencil Beam” based algorithm implemented in MultiPlan is suitable for dose calculation in homogeneous water-equivalent media but not in low electronic density media such as the lung. Indeed, the dose is overestimated (up to 40%) inside the field and may lead to reduce the tumor treatment efficiency while it is underestimated outside the field which can underestimate the dose to critical organs within proximity of the tumor. Regarding the Monte Carlo algorithm implemented in MultiPlan, calculated and measured dose distributions are consistent and, as a consequence, it is the most suitable algorithm available in MultiPlan to estimate the dose received by a patient when low density media are involved.