Session Item

Catheter reconstruction is the most time-consuming step of treatment planning for interstitial catheter implants in breast brachytherapy. The manual reconstruction time is typically 3-5 minutes per catheter, adding up to an hour or more for the whole implant. Besides that, it is prone to systematic human errors. The objective of this work was to evaluate the capabilities of an electromagnetic tracking (EMT) system integrated into an afterloader for automated online catheter reconstruction.

An afterloader prototype (Flexitron, Elekta Brachytherapy, Veenendaal, The Netherlands) equipped with an EMT sensor was used to assess the geometry of the catheter implant of 20 patients immediately after the planning CT with the patient still on the CT couch. The acquired EMT data were processed and registered to the clinically used, manual catheter reconstruction which served as a ground truth. In the future the registration is to be done based on the fixation buttons as mutual landmarks. For this reason, an approach to detect the buttons using EMT and tiny embedded magnets is also currently investigated. The geometric reconstruction accuracy was determined as the minimum orthogonal distance of an EMT reconstruction point to the clinical catheter path (Fig. 1, a). For dosimetric evaluation the EMT reconstructions were imported into the research version of the treatment planning system Oncentra Brachy (Elekta Brachytherapy, Veenendaal, The Netherlands), along with the clinically approved structure set and treatment plan including dwell positions and dwell times. Standard figures of merit (CI, DNR, and COIN) were calculated in order to appraise the quality of the implant.

The acquisition time of the EMT system is 22 s per catheter and typically 6-11 min for the entire implant, the time for data processing is < 1 min. The median distance between EMT reconstruction points and clinical reconstruction is 1.1 mm with an interquartile range of 1.1 mm. The median CI decreased from 94.7 % for the clinical reconstruction to 94.1 % for the EMT based reconstruction while both the median DNR (0.24) and the median COIN (0.69) did not change. However, in two individual patients a drop of the CI as large as 7.5 % could be observed (Fig. 1, b). In three cases the CI was < 90 %. None of the implants failed the requirements for maximum skin dose.

The EMT based implant reconstruction was found to be geometrically accurate within the intrinsic limits of the EMT system and did not deteriorate the dosimetric implant quality in 17 of 20 patients. In the three cases in which the implant quality decreased, only small manual adjustments of some of the reconstruction points were sufficient to meet the clinical dose requirements again. Therefore strategies to further refine the EMT based reconstruction are currently being explored.