Development, Safety, and Patient Trust of Autonomous Robotic Medical Ultrasound Diagnostics

Medical ultrasound imaging is used as an inexpensive point of care diagnostic tool to visualize anatomy below the surface of the skin. Still, access to ultrasound-trained clinicians is limited, especially in remote medicine and for diagnostic scans, like breast ultrasound. Autonomous robotic ultrasound scans could increase access to this critical diagnostic tool by decreasing the need for expert clinicians to be present to perform scans and increase uniformity in scans collection.

For autonomous robotic ultrasound system to be safe and effective it must perform consistently within safe force thresholds and be trusted by patients to do so. First, this thesis addresses the development of a method to develop and perform trajectories consistently on phantom (physically simulated) human arm tissue. This work showed the efficacy of the system to collect, identify, and track vasculature scans precisely within safe force limits. Second, the same procedure is used for a human study with better force sensors, a novel arm restraint device, and a user interface. Participants secured themselves, confirm the procedure to be undertaken, and underwent a effective and safe forearm scan along the arm. Images were determined to be adequate for vasculature identification, which could be used for procedure planning. Participants significantly increased their perception of safety and trust in the robotic procedure from only one interaction with the system.

Next a novel, mechanically accurate phantom was developed to tackle a particularly important and difficult surface of the body to traverse, human breast tissue. Finally, the novel custom phantom developed was compared with a commercially available breast phantom to assess the ability to track force effectively and complete the trajectory using the PID controller and trajectory generation of the thesis work. This showed the necessity for mechanically accurate phantoms as the system was more effective on the stiffer commercially available phantom versus the more realistic custom phantom.

Overall, this work demonstrated the safety and efficacy of an autonomous robotic arm to perform ultrasound peripheral vascular scans in vivo. Additionally, this work devised custom phantoms to promote development of effective ultrasound scanning methods for geometrically varied and less stiff surfaces of the body. Finally, the human study survey data showed, for the first time, that one experience with an autonomous ultrasound scan could increase trust and perceived safety of the procedure.

Further, diagnostic scans performed by autonomous robotic arms with localization of target tissue supports future work in autonomous clinical procedure development.