SU‐E‐U‐13: Repeatability of Robotic Placement of Ultrasound Probes for An Integrated US‐CT Approach to Image‐Guided Radiotherapy

Purpose: Ultrasound (US) has potential to provide real‐time monitoring of radiotherapy, but the tissue deformations and CT artifacts caused by US probes limit clinical utility. This work investigates the repeatability of robot‐assisted placement of geometically‐identical imaging and model probes. The CT‐compatible model is proposed for use during treatment planning. Methods: An ex vivo bovine liver was fixed in gelatin. Six 2mm‐diameter spherical metallic fiducials were embedded in the liver at depths 1– 2cm and 4–5cm. Volumetric data were acquired with a Philips CT scanner and Elekta Clarity® ultrasound system. Probe placement was controlled with a robot consisting of a passive arm, three translation stages, and a sensor to monitor probe‐tissue contact forces. Fiducial positions were measured in US images acquired after placement of the imaging probe, as well as in CT images acquired before and after placement of both probes. Results: The mean absolute difference between fiducial displacements after six repeated probe placements was 0.4±0.4 mm in US images and 0.3±0.2 mm in CT images acquired with the model probe. Maximum displacements generally occurred in the anterior‐posterior (AP) direction (i.e. normal to the probe), where the imaging and model probes displaced the three shallow fiducials by 5.0±1.7mm and 3.7±1.6mm, respectively. The AP displacement of the three deeper fiducials was 1.0±0.7mm with the model probe. The mean contact forces of the imaging and model probes were 1.3±0.13N and 0.3±0.01N, respectively, for the shallow fiducials, and 1.1±0.2N and 0.1±0.02N, respectively, for the deeper fiducials. Conclusions: Robot position, fiducial displacements, and contact forces were sufficiently repeatable with either probe. Displacement and force discrepancies between probes are likely due to weight and alignment differences. Solutions to reduce the discrepancies are under investigation. Yet, with <2mm mean difference between probes, this work is promising as it demonstrates the feasibility of a potentially revolutionary approach to IGRT. This work is supported in part by NCI CA R01‐161613.