Abstract (eng)
Background:
Cone Beam Computed Tomography (CBCT), integrated to the medical linac, is currently utilized for correction of patient positioning uncertainties. Additional possible applications of these image data sets are conceivable for Adaptive Radiation Therapy (ART). By repetitive imaging, one can react on anatomical changes in the patient (e.g. by the adaptation of the irradiation plan) and hence a more sparing therapy can be achieved. In the actual study should be investigated in early steps to what extent CBCT images are feasible for treatment planning purposes in ART.
Materials and Methods
All evaluations were done with the CBCT device integrated to the Elekta Synergy® linac at the Department of Radiotherapy at the Medical University of Vienna/AKH Vienna. The images, which were used for dose comparison studies, were acquired with a conventional multi slice CT scanner of Siemens (Siemens VolumeZoom). For two calibration phantoms (CATPhan® and Gammex® RMI), the average Hounsfield Units (HU) were detected in various material inserts in CT and in CBCT images. Furthermore, the dependency of HU values on image acquisition parameters (tube voltage, mAs, filtering and collimation) was evaluated and HU/ED (Electron Density) correlation curves of various clinical protocols were established. In the course of an extensive Quality Assurance (QA) process of the CBCT device, representative Image Quality (IQ) parameters were observed and the calibration curves were tested concerning their long-term stability. Subsequently, dose variations, compared to the planning CT, were calculated for an Anterior – Posterior (AP) treatment plan with the Treatment Planning System (TPS) iPlan® of BrainLAB. For commissioning of iPlan®, basic beam data sets had been acquired and implemented. For the determination of dose calculation accuracy, a gamma evaluation was applied. The gamma evaluation was performed on data sets of an inhomogeneous multipurpose phantom with polystyrene as well as cork inserts.
Results
Generally, IQ parameters in CBCT images were demonstrated to be accurate enough for a possible implementation for ART. In comparison to the CT, the experimentally determined CBCT HU/ED correlation curves of both calibration phantoms showed average deviations of 27% in ED. In contrast to the Gammex® RMI based correlation curves, the CATPhan® curves exhibited parallel behavior among each other and obvious similarities to the desired CT correlation curves. Thus, they were applied for dose calculation based on CBCT images. The requirement of temporal stability could be satisfied for one image protocol (Image Quality protocol). Over the investigation period of 10 month, maximum deviations of 30 HU around the actual mean values of the inserts in the CATPhan® phantom could be determined. A comparison of depth dose curves between CBCT- and CT based dose calculation, relying on the Pencil Beam Algorithm (PBA), showed mean Gamma Indexes up to 0.51 and maximum values of 0.94. For dose profile comparison, 0.24 and 0.874 was found for mean and maximum Gamma Indexes, respectively. An underestimation of dose could be detected in case of CBCT based dose calculation due to a significant reduction of the outer contour of the phantom in CBCT images. Concerning dose calculation, carried out by utilizing the Monte Carlo Algorithm (MCA), the representative Gamma Indexes were higher and more pronounced.
Discussion
In the frame of this study, the feasibility of CBCT based dose calculation for ART under certain conditions could be confirmed. However, one has to bear in mind that image artifacts highly influence the accuracy of dose calculation relying on CBCT images. A possible opportunity for handling this problem might be advanced image reconstruction algorithms in order to trace artifacts already in frequency space and hence minimize image errors. Additionally, an autosegmentation tool may be used in order to autocontour artifact affected volumes and assign well defined HUs in order to bypass their influence on dose calculation. Consequently, CBCT images can be used for positioning correction on the treatment table as well as for contemporaneous adaptive re-planning. Further studies that show the dosimetric advantage of ART are planned for CRT and IMRT for H&N- and pelvis cases.