Linear accelerator output variability.

Abstract

The Wellington Cancer Centre is equipped with two matched linear accelerators (Varian 2100CD) linear accelerators (identified as SN1027 and SN42). Each morning, before treatments commence, a radiation therapist carries out an output constancy check of the radiation output and every fortnight a physicist measures, in a phantom, the delivered radiation dose to check on the machine calibration. The daily output checks have been recorded into a database (Argus QA for Radiation Oncology) since August 1997 and in June 1998 the fortnightly calibration measurements were added. The information in the database, up to April 2003, has been analysed to consider the quality of the daily constancy checks as compared with the fortnightly calibration measurements and whether the data contains useful information on machine performance. After allowance for the effects of machine recalibration the fortnightly calibration measurements had an average standard deviation of 0.4% and the daily constancy checks 0.8%. The daily constancy checks had a greater number of large deviations than would be expected assuming a normal distribution and were not a good predictor of the need for a recalibration. The fortnightly calibration measurements with a much lower spread give a reliable indication of the need for a recalibration allowing the adoption of a +/- 1% tolerance. Over the period analysed one accelerator (SN42) was relatively stable with the output generally drifting between +/- 1% while the other (SN1027) had a consistent increase in the average output of about 2.5% per year.