Background and purpose: The feasibility of a real-time in vivo dosimeter to detect errors has previously been demonstrated. The purpose of this study was to: (1) quantify the sensitivity of the dosimeter to detect imposed treatment errors under well controlled and clinically relevant experimental conditions, and (2) test a new statistical error decision concept based on full uncertainty analysis.
Materials and methods: Phantom studies of two gynecological cancer PDR and one prostate cancer HDR patient treatment plans were performed using tandem ring applicators or interstitial needles. Imposed treatment errors, including interchanged pairs of afterloader guide tubes and 2-20mm source displacements, were monitored using a real-time fiber-coupled carbon doped aluminum oxide (Al(2)O(3):C) crystal dosimeter that was positioned in the reconstructed tumor region. The error detection capacity was evaluated at three dose levels: dwell position, source channel, and fraction. The error criterion incorporated the correlated source position uncertainties and other sources of uncertainty, and it was applied both for the specific phantom patient plans and for a general case (source-detector distance 5-90 mm and position uncertainty 1-4mm).
Results: Out of 20 interchanged guide tube errors, time-resolved analysis identified 17 while fraction level analysis identified two. Channel and fraction level comparisons could leave 10mm dosimeter displacement errors unidentified. Dwell position dose rate comparisons correctly identified displacements ≥ 5mm.
Conclusion: This phantom study demonstrates that Al(2)O(3):C real-time dosimetry can identify applicator displacements ≥ 5mm and interchanged guide tube errors during PDR and HDR brachytherapy. The study demonstrates the shortcoming of a constant error criterion and the advantage of a statistical error criterion.
Copyright © 2011 Elsevier Ireland Ltd. All rights reserved.