PP05 Presentation Time: 4:06 PM

Purpose Breast conserving surgery plus accelerated partial breast irradiation is increasingly popular for early-stage breast cancer patients with promising clinical outcomes. However, the identification of the seroma cavity on CT can be challenging, leading to inaccuracy and inconsistency in seroma contouring. For brachytherapy techniques this presents additional challenges. Three-dimensional ultrasound (3DUS) technology will significantly improve target visualization and implant accuracy. In this work, we introduce a robotic 3DUS system with a spatial tracking arm that has great potential for improved visual guidance in breast brachytherapy, and describe the calibration and commissioning process for this system. Materials and Methods The robotic 3DUS system consists of a 3D-printed probe holder, customizable for any commercial ultrasound transducer (BKMedical 8811 for our system), a motor that drives the probe simultaneously both linearly and rotationally to acquire hybrid 3DUS images, an ultrasound transparent TPX plate that supports the probe motion, a spatially-tracked counterbalanced arm, and software for visualization and target segmentation (Figure 1a). The tracking arm consists of five encoded joints, shown in Figure 1a, and allows arm movement to provide sufficient tracking space in clinical use. The linear accuracy of the 3DUS scan was evaluated using a phantom composed of strings separated by 10.00±0.05 mm in three dimensions in a 7.25% isopropyl alcohol solution (22°C). Volume segmentation accuracy was evaluated by imaging and segmenting a sphere (1.2cc) embedded in an ultrasound phantom. Lavenberg-Marquardt based nonlinear regression was used to determine the optimal calibration parameters for each joint, which corrects the rotational deviations of each joint and translational offsets of the end-effector (the transducer). An optical tracking device (Polaris Spectra, Northern Digital Inc., Canada) was used for arm calibration. A total of 202 data points distributed in a 360×650×95mm3 working space were collected, among which, 80% (162 points) were used to calibrate the encoder parameters. The rest were used to evaluate spatial tracking accuracy. Results The 3DUS scanning volume is 5cm × 10cm × scan depth (from 5.2cm to 8.0cm). For a commonly used scan depth (6.0cm), the image resolution is 0.075mm, 0.082mm, and 0.33mm in lateral, axial, and elevational direction, respectively. The mean linear scan accuracy was within 2% in all directions, and the volume segmentation accuracy was 3%, all below TG128 action levels. Mean deviations between the mechanical and optical tracking are 0±0.5mm, 0±0.7mm, and0±0.8mm in three dimensions (plotted in Figure1b), with a mean Euclidean distance of 1.0±0.5mm. Conclusions We have developed, calibrated and commissioned a robotic 3DUS system for breast brachytherapy. The imaging and tracking accuracies were both excellent for clinical use. The spatial tracking arm has exciting potential to achieve rigid registration to other imaging modalities or the implant templates that are used in interstitial partial breast brachytherapy. Acknowledgements The authors would like to thank the support of NSERC and the OICR Imaging Program. Breast conserving surgery plus accelerated partial breast irradiation is increasingly popular for early-stage breast cancer patients with promising clinical outcomes. However, the identification of the seroma cavity on CT can be challenging, leading to inaccuracy and inconsistency in seroma contouring. For brachytherapy techniques this presents additional challenges. Three-dimensional ultrasound (3DUS) technology will significantly improve target visualization and implant accuracy. In this work, we introduce a robotic 3DUS system with a spatial tracking arm that has great potential for improved visual guidance in breast brachytherapy, and describe the calibration and commissioning process for this system. The robotic 3DUS system consists of a 3D-printed probe holder, customizable for any commercial ultrasound transducer (BKMedical 8811 for our system), a motor that drives the probe simultaneously both linearly and rotationally to acquire hybrid 3DUS images, an ultrasound transparent TPX plate that supports the probe motion, a spatially-tracked counterbalanced arm, and software for visualization and target segmentation (Figure 1a). The tracking arm consists of five encoded joints, shown in Figure 1a, and allows arm movement to provide sufficient tracking space in clinical use. The linear accuracy of the 3DUS scan was evaluated using a phantom composed of strings separated by 10.00±0.05 mm in three dimensions in a 7.25% isopropyl alcohol solution (22°C). Volume segmentation accuracy was evaluated by imaging and segmenting a sphere (1.2cc) embedded in an ultrasound phantom. Lavenberg-Marquardt based nonlinear regression was used to determine the optimal calibration parameters for each joint, which corrects the rotational deviations of each joint and translational offsets of the end-effector (the transducer). An optical tracking device (Polaris Spectra, Northern Digital Inc., Canada) was used for arm calibration. A total of 202 data points distributed in a 360×650×95mm3 working space were collected, among which, 80% (162 points) were used to calibrate the encoder parameters. The rest were used to evaluate spatial tracking accuracy. The 3DUS scanning volume is 5cm × 10cm × scan depth (from 5.2cm to 8.0cm). For a commonly used scan depth (6.0cm), the image resolution is 0.075mm, 0.082mm, and 0.33mm in lateral, axial, and elevational direction, respectively. The mean linear scan accuracy was within 2% in all directions, and the volume segmentation accuracy was 3%, all below TG128 action levels. Mean deviations between the mechanical and optical tracking are 0±0.5mm, 0±0.7mm, and0±0.8mm in three dimensions (plotted in Figure1b), with a mean Euclidean distance of 1.0±0.5mm. We have developed, calibrated and commissioned a robotic 3DUS system for breast brachytherapy. The imaging and tracking accuracies were both excellent for clinical use. The spatial tracking arm has exciting potential to achieve rigid registration to other imaging modalities or the implant templates that are used in interstitial partial breast brachytherapy.