文章摘要

文章信息/Info

Title:: A study on ROS-based robotic arm-assisted ultrasound scanning system

作者:: 张娟1; 李锐1; 程威1; 秦浩1; 吴诗尧1; 刘立1; 2; 3; 倪东1; 2; 3; 陈思平1; 2; 3; 钱建庭1; 2; 3△; 1.深圳大学医学部生物医学工程学院，深圳 518060;2.医学超声关键技术国家地方联合工程实验室，深圳 518060; 3.广东省生物医学信息检测与超声成像重点实验室，深圳 518060

Author(s):: ZHANG Juan1; LI Rui1; CHENG Wei1; QIN Hao1; WU Shiyao1; LIU Li1; 2; 3; NI Dong1; 2; 3CHEN Siping1; 2; 3; Chin Chien Ting1; 2; 3; 1.School of Biomedical Engineering，Shenzhen University，Shenzhen 518060，China;2.National-Regional Key Technology Engineering Laboratory for Medical Ultrasound，Shenzhen 518060;3.Guangdong Key Laboratory for Biomedical Measurement and Ultrasound Imaging，Shenzhen 518060

Keywords:: Robot operating system; Robotic arm; Ultrasound scanning; KinectV2; Point cloud

摘要:: 本研究提出了一种基于机器人操作系统（robot operating system，ROS）的机械臂辅助超声扫描系统，旨在减轻超声医生的劳动强度，降低对操作者的依赖，标准化扫查流程，提高扫描操作的准确度和效率。系统由ROS、Kinect V2视觉传感器、六轴机械臂和超声采集装置等硬件平台组成。本系统首先利用 Kinect V2实时获取场景的RGB彩色和深度图像，经内外参标定、配准后形成RGB-点云数据用以充分描述三维空间信息。通过对点云进行去噪、分割、表面法向量分析和运动轨迹提取等系列处理，再经由坐标系统转换，本系统即可进行机械臂坐标系统下的路径规划，进而引导机械臂持超声探头到指定区域的表面实施扫描动作，完成超声影像的实时采集。结果表明，基于ROS搭建系统平台，结合Kinect V2视觉导航，本系统可引导机械臂夹持超声探头较快速、较准确、较稳定地到达待扫查区域，实现自主的超声图像采集。

Abstract:: To present a ROS-based robot-assisted ultrasound scanning system. The system was designed to relieve the labor intensity of sonographers, make the scanning operator-independent, standardize the workflow and therefore improve the accuracy and efficiency of scanning. Hardware platforms in the proposed system consisted of a central control system (ROS), a Kinect V2 visual sensor, a six-axis manipulator and an ultrasound machine. Based on the RGB and depth images acquired by Kinect V2 in real time and a followed internal and external parameter calibration and registration, the system firstly generated an aligned RGB-Point cloud data to describe the appearance and structure information of the scene. Then, this RGB-Point cloud data was sequentially processed by denoising, segmentation, surface normal calculation and trajectory extraction. With a further transformation in coordinate system, the proposed system could conduct the path planning under the robotic arm coordinate system. The generated path could finally guide the robot arm holding an ultrasound probe to implement scanning on the surface of the target area and acquire real-time ultrasound images. Our results show that, by leveraging the ROS to build system platform and the Kinect V2 for visual analysis, the proposed system can guide the robotic arm with an ultrasound probe to reach the target area with decent efficiency, accuracy and stability, and finally achieve autonomous ultrasound image acquisition.