SU-E-T-685: Root Cause Analysis of in Vivo Dosimetry Controls for Dynamic Arc Treatments

Authors


Abstract

Purpose:

To produce an insight on the influence of selected parameters on the pass/fail rates of in-vivo dosimetry (IVD) controls for dynamic arc treatments.

Methods:

A group of 25 patient treatment plans optimized with the treatment planning system Eclipse™ (Varian) for dual-arc RapidArc treatment was analyzed. All plans were tracked with the IVD solution EPIgray (DOSIsoft S.A) over all sessions of the treatment course. Based on the acquired a-Si EPID images, the dose differences to the TPS at given reference points and the corresponding gamma indices were computed. In addition, the dose difference variations over the course were determined post-treatment. Five parameters — considered as possible fail or pass causes of IVD for dynamic arc plans — were chosen. Based on the treatment plan, two types of modulation indices qualifying the movement of the MLC were calculated. An “in-field” coefficient and a “weighted-in-field” coefficient were established to quantify the position of the dose reference points in the planned volume. Finally, the dose rate variations for each arc were also determined. The described parameters were confronted to the results of the IVD controls.

Results:

The analyzed plans have a mean relative dose difference of 0.2% (±7.2%). No correlation could be established between the modulation indices and the dose differences, the gamma indices or the dose difference variation. Slightly better passing rates for individual sessions and total treatment plans were observed for “in-field” coefficients superior to 50%. Similar passing rates were observed for arcs with below and above average dose rate deviations.

Conclusion:

Though all the analyzed parameters are cited as possible pass/fail criteria, only the position of the dose reference point can be linked to dose differences in IVD controls. Automatically generated dose measurement points could avoid false interpretations and an overflow of false negative alerts.

The author's PhD research is funded by DOSIsoft S.A.

Ancillary