The effect of repetitive manual fiducial localization on target localization in image space.

摘要

OBJECTIVE: In this clinical study, we quantify intra- and interobserver variability of manual fiducial localization in image space, as the effect of repetitive manual fiducial localization is still unclear, especially on a target position. METHODS: After uploading eight imaging datasets with a total of 56 skin adhesive fiducial markers in a commercially available image-guidance system, the centroids of the fiducial markers were tagged. This task was executed repeatedly at three separate moments by six different observers. The fiducial localization variability and its target shift effect in image space were determined out of 1008 tagged fiducial markers. RESULTS: The maximal intraobserver target shift effect measured 0.72 +/- 0.14 mm in computed tomographic image space and 0.95 +/- 0.21 mm in magnetic resonance image space. CONCLUSION: If a fiducial tagging task is well understood, repetitive manual detection of fiducial markers can be done with a low intraobserver fiducial localization variability, resulting in a submillimetric effect on a target position, either in computed tomographic or magnetic resonance image space. Therefore, we think it is justified to determine the centroids of a skin adhesive fiducial marker in the image space by hand.