Purpose: To investigate the accuracy by which the beam configuration settings are transferred to the logfiles of the linear accelerator.
Methods: Ten patients with head & neck cancers were used to create two sets of dosimetric data. The first one consists of the doses to the different targets and organs at risk (OARs) of the physician approved treatment plans. The second set was derived from dose calculation using the linac logfiles to generate a DICOM plan. For the latter method, the Standard Imaging LinacView system was used. All the plans were optimized and calculated using the RayStation v9.2 treatment planning system. The dose volume histograms of the targets and OARs were used to calculate the corresponding mean, maximum and minimum doses (Dmean, Dmax and Dmin) as well as the tumor control and normal tissue complications probabilities (TCP and NTCP) for the respective datasets.
Results: The results of the first three patients analyzed indicate that the dose distributions derived from the logfiles deliver less mean dose to the HR, IR and SR targets (by 0.7±0.3Gy, 0.9±0.2 Gy, 0.4±0.2 Gy, respectively). This difference is also reflected to the corresponding TCP values (difference of 1.6±2.7%, 1.5±2.7%, 0.3±0.9%). Regarding the OARs, left brachial plexus and right parotid gland showed the largest differences (0.9±1.6Gy, 0.3±0.9Gy in Dmax and 0.9±1.5%, 0.6±1.1% in NTCP, respectively). Much smaller differences were observed in the rest of the OARs.
Conclusion: The dosimetric and radiobiological analysis of the planned and logfiles-derived dose distributions showed that there are uncertainties in the transfer of the beam configurations from the treatment planning system to the linear accelerator that lead to observable differences in the dose delivery.