Fiducial registration error as a statistical process control metric in image-guidance radiotherapy with fiducial markers.

Portal imaging of implanted fiducial markers has been in use for image-guided radiotherapy (IGRT) of prostate cancer, with ample attention to localization accuracy and organ motion. The geometric uncertainties in point-based rigid-body matching algorithms during localization of prostate fiducial markers can be quantified in terms of a fiducial registration error (FRE). In this study, the aim is to demonstrate how statistical process control (SPC) can be used to intercept potential problems with rigid-body matching algorithms in a retrospective study of FRE for a pilot cohort of 34 patients with fiducial markers. A procedure for estimating control parameters of a SPC control chart (x-chart) from a small number of initial observations (N) of FRE was implemented. The sensitivity analysis of N on the number of 'in-control' and 'out-of-control' x-charts was also performed. Uncorrected rotational offsets of an individual patient were examined to elucidate possible correlations with the behaviours of an x-chart. Four specific types of qualitative x-chart behaviour have been observed. The number of out-of-control processes was insensitive to the choice of N, provided N ≥ 5. Residual errors of rigid-body registration were contributed from uncorrected rotational offsets in 5 out of 15 'out-of-control' x-charts. Out-of-control x-charts were also shown to be correlated with potential changes in the IGRT processes, which may compromise the quality of the radiation treatment delivery. The SPC methodology, implemented in the form of individually customized x-charts, has been shown to be a useful tool for monitoring process reliability during fiducial-based IGRT for prostate cancer.