Fully automated robotic ultrasound gallbladder imaging with subcostal scanning.

To overcome the operator-dependence of ultrasound (US) imaging, robotic US systems have been widely developed for fully automated scanning. In this study, toward fully automated screening of gallbladder in abdominal US examinations, we propose a scan path planning method and a probe posture control strategy using our robotic US system. The path planning method uses an RGB-D camera to capture changes in the patient’s chest height position due to breathing. This allows the boundary line of the subcostal, which is an important landmark for the gallbladder scan, to be identified, and the scanning path for the robot is automatically generated. Regarding the probe posture control strategy, for proper contact between the probe and the body surface, we combine probe posture control with body surface normal estimation using the RGB-D camera before scanning and deep learning-based visual servoing to avoid shadows during scanning. These were implemented in our robotic US system and feasibility experiments were conducted on six healthy subjects. When the path planning method and the posture control strategy proposed in this study were applied, the detection rate of the gallbladder was 78.3%. These results demonstrate the possibility of fully automated gallbladder screening using the proposed system.