Session Item

For correct dose delivery in cervix brachytherapy (BT) the dwell positions and times need to precisely correspond to the treatment plan. Correspondence of the actual applicator and needle positions with the MRI/CT-based treatment plan is therefore essential. Positional deviations however might occur due to applicator and/or needle shifts, transfer tube swaps, anatomical changes or reconstruction uncertainties. Electromagnetic tracking (EMT) has been proposed for detection of such errors and uncertainties. The feasibility of EMT using a prototype of a hybrid EMT/BT system, i.e. a Flexitron afterloader with an EMT sensor integrated in the check cable (Elekta, Veenendaal) has previously been demonstrated by phantom experiments in typical clinical BT environments. An accuracy of less than a millimeter was found in 6F interstitial plastic needles. The purpose of this study is to prospectively analyze the feasibility of this system to detect treatment errors and reconstruction uncertainties in cervix HDR BT patients.

Twenty cervix patients, treated with three or four HDR BT fractions using an intracavitary/interstitial applicator as part of their treatment, were included in a prospective study, simulating a workflow with EMT-based treatment verification. EMT measurements, scheduled before dose delivery in one to three BT fractions per patient, were performed using the hybrid EMT/BT afterloader by automatically moving the EMT sensor through all needles in the implant according to a predefined MRI- or CT-based treatment plan. Dose delivery was performed afterwards using our clinical afterloader, ensuring that the clinical treatment was not affected by the EMT experiments. The intracavitary channels were discarded in this study as our phantom study showed unacceptable deviations. The residual errors, i.e., the Euclidean distances between the registered EMT measured and planned dwell positions, were calculated.

Data sets from 37 fractions were available for analysis. In three data sets (3, 23 and 28) a discrepancy between the registered EMT and planned dwell positions, e.g. a channel swap (fig. 1a), was detected. The increased difficulty in needle reconstruction based on MRI scans compared to CT scans was reflected by a significant difference in mean residual error (p<0.001), i.e. 0.90 mm and 0.68 mm for MRI-based and CT-based reconstruction, respectively. Moreover, outliers were detected in 5 data sets (20, 22, 32, 33 and 34), i.e., a single needle with a high residual error. These datasets were all planned using MRI and contained ≥ 6 needles, again suggesting reduced reconstruction accuracy for MRI (fig. 1b). The results are summarized in figure 2.

This study demonstrates that the hybrid EMT/BT system can detect treatment errors and needle reconstruction inaccuracies in HDR BT cervix patients in a clinical setting.