Session Item

To investigate inter-pulse and -dwell dose variation during pulsed dose rate (PDR) brachytherapy (BT) for cervical cancer via in vivo dosimetry (IVD) with a scintillation-based detector and utilize time-resolved data to increase IVD reliability.

At our department IVD with a scintillator-based detector measuring the dose rate every 50ms is routinely performed during PDR BT for cervical cancer. In this study the patients received two fractions of 17.5 Gy for tandem ring (TR) treatments and 15 Gy for multi-channel vaginal cylinder (MVC) treatments each divided into 20 hourly pulses. Treatment planning was based on MRI. Before treatment, the scintillator was placed in a separate channel. The dosimeter position and corresponding expected dose recording were determined for each plan.

Retrospectively, measured and expected dose were compared for each fraction. Furthermore, the pulse-to-pulse variations of the dose deviations were investigated to study the implant stability. A methodology was applied for correcting for uncertainties in detector position and evaluated in 15 treatments with MVC. The methodology involved estimation of positional shifts of the dosimeter along the MVC’s longitudinal axis based on a least square fit of the TG43-calculated dose rates to the measured ones for the MVC's central channel. Finally, a comparison of the measured and expected dose rates were performed for each dwell position in a single treatment. The comparison was done both before and after applying the positional correction of the detector.

Data from 79 PDR BT fractions was analysed; 33 MVC and 46 TR. The mean±1SD of the deviation between the total measured and expected dose were -1.1±1.7 Gy (-9±13%) for TR and 0.0±0.5 Gy (0±4%) for MVC. For the first pulse the dose deviation was -0.4±3% for TR and -0.1±0.6% for MVC, with an increase in variation (1SD) of up to ±18% and ±12% respectively for subsequent pulses, fig. 1.

15 MVC treatments showed an average positional dosimeter correction of 0.5±9 mm. Five of these plans exhibited deviations of more than 10% between measured and expected dose in one or more pulses. The mean deviation between measured and expected dose rate pr. dwell-position in the non-central MVC channels dropped from 22±95% to 3.4±7.9% after applying the correction for detector position, fig. 2.

Real-time dosimetry during PDR BT treatments has been performed in a large cohort, showing good agreement with the expected dose for the MVC. The larger deviation seen with TR is expected to originate from positional offsets of the detector. The measured dose rates enabled a more thorough investigation of the treatment, including determining positional offsets in the detector position. Correcting for detector position based on the irradiation of the first source channel increased the reliability of the subsequent dose rate comparison. The next step is to test this method in the full cohort including TR+needle treatments.