PP02 Presentation Time: 4:09 PM

Abstract

Purpose

To enable a real-time applicator navigation and guidance for brachytherapy, we used a novel prototype infra-red tracking camera system directly integrated into a mobile cone-beam computed tomography (CBCT) scanner for the first time. We provide the first characterization of the combined device and a corresponding performance evaluation, assessed the respective tracking accuracy, and implemented a prediction of needle courses and tips on CBCT scans based on the infra-red tracking.

Materials and Methods

We conducted a comprehensive assessments of the system's camera calibration and camera-CBCT registration by means of dedicated phantoms. For this, we first evaluated the effects of intrinsic parameters such as temperature or source/detector gantry positions on the positions of the tracked markers. Afterwards, calibrations with various setting (sample number, field of view coverage, calibration directions, calibration distances, and light conditions) were conducted to determine the setting-requirements for achieving maximum tracking quality based on an in-house phantom. The corresponding effects on camera-CBCT registration were also determined by comparing markers tracked and measured via CBCT. Long-term stability was assessed by comparing tracking and a ground-truth weekly for six weeks. Furthermore, a marker-tool rigidly attachable to needles was 3D printed. The respective tool tracking precision and accuracy was evaluated for stationary settings as well as dynamic movements. Moreover, we evaluated Euclidean deviations between tracked markers and marker positions determined via CBCT. For implementing needle tracking, ground-truth models of the tool attached to needles of 200mm and 160mm length were first created and then matched onto the tracked positions, aiming to project the needle courses into CBCT scans. Deviations between projected and actual needle tips were measured. Finally, we put our results into perspective regarding simulations of the devices's tracking uncertainties.

Results

Robust and reliable tracking was feasible after a 100 min warm-up period using the tracking system ‘as is’, with positional variations of only 0.11±0.10 mm. Gantry rotations impacted the tracking procedure and led to inaccuracies of up to 0.70 mm. We identified 4000 samples and full coverage to be required for ensuring an optimum marker tracking and camera-CT registration with geometric variations of only 0.18±0.03 mm and 0.42±0.07 mm, respectively. Nevertheless, long-term stability analyses revealed geometric variations larger than two standard deviations compared to the initial calibration occurring after only 3 weeks. For the stationary settings and dynamic movements, we achieved a tool tracking precision and accuracy of 0.04±0.06 mm and 0.16±0.18 mm, respectively. Tracked marker positions deviated by 0.57±0.18mm from the marker positions identified via CBCT. Moreover, predicted needle tips in air deviated by only 1.54±0.68 mm (200mm needle) and 1.40±0.62 mm (160mm needle) from the actual tip locations. The simulated tracking uncertainties resulted in tip variations of 1.75±0.91 mm and 1.43±0.69 mm for the 200mm and 160mm needle, respectively.

Conclusions

We implemented for the first time a standalone combined camera-CT system for applicator and needle tracking in brachytherapy. With the innovative system, achieving a high tracking and prediction accuracy of marker-tools and needles was feasible. The system showed excellent pre-conditions and high potential for corresponding tracking workflows.