Development of a robotic trans-oesophageal ultrasound system and its application in automatic acquisition

• Main Author: Wang, Shuangyi
• Other Authors: Rhode, Kawaldeep Singh ; Althoefer, Kaspar Alexander
• Published: King's College London (University of London) 2017
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.727026

Trans-oesophageal echocardiography (TOE) has been widely utilized for cardiac disease diagnosis and interventional procedure guidance. However, the TOE operator is required to manually manipulate the probe, often for long periods of time and sometimes in an X-ray environment where there is exposure to ionising radiation. Considering these challenges, a robotic TOE system with intelligent functions that can be operated remotely would be a revolutionary design. Though many research works have been done for ultrasound robotics, no work for robotizing the TOE probe and exploring its application has been reported in the literature. In this thesis, a new design of a robotic TOE system is reported after the review of related knowledge on echocardiography and previous works on ultrasound robotics. The system has four degrees of freedom to remotely operate a commercial TOE probe. The preliminary tests of the robot demonstrate good repeatability performance of the system. Additionally, the kinematic modelling of the system relating the motor parameters to the ultrasound images is developed, verified, and reported. The model has been included into a view-planning platform designed for planning TOE views and robot paths before a surgical procedure using pre-scanned MR images. With the proposed robot and the view-planning method, an automatic acquisition workflow is reported. This includes an auto-patient adaption method and a tracking method for locating the probe position when applying feedback. Phantom experiments were performed and the results were analysed. Motivated by a reliability problem identified in the workflow, an improved tracking method of the probe is reported. The new experiments indicate reliable performance of the workflow and the mean error in the near field of ultrasound, where most structures of clinical interest are located, is at the same magnitude as heart motions (1 cm). The study is encouraging for the eventual clinical application of robotic-based TOE acquisition.