An AI-enabled robotics system is developed to automate handheld medical imaging and improve efficiency and consistency. Currently, clinicians operate Ultrasound or FibroScan probes manually. While effective, this approach varies widely between operators. Differences in hand pressure, variabilities in patient anatomies, probe angles, and scanning patterns can affect imaging quality as well as diagnosis results. Automated or semi-automated handheld medical imaging systems that employ robot arms facilitate diagnosis results that are more consistent, repeatable, and less affected by human fatigue caused by repetitive work or long working hours. These systems also help reduce the cognitive load on clinicians, allowing them to focus more on interpretation and patient care rather than the physical demands of scanning.

To begin, the automated system projects a predefined structured grid onto the torso of a patient. A robot arm holding the imaging probe then moves through the points on the predefined grid in a smooth, continuous path, just as expected from an experienced radiologist. The system avoids abrupt stops and maintains a stable probe orientation, which is important for ensuring consistent contact with the torso surface. This structured approach ensures that every region of interest is covered uniformly, minimizing the risk of missed areas and improving the reliability of follow up comparisons.

To test the system experimentally, the team used a Universal Robots UR series collaborative robot arm equipped with a dummy probe. As the robot moved through the grid on a mannequin torso, it continuously recorded the position of the probe’s centre point. These recorded trajectories (paths and timestamps) were later compared with the planned trajectories. The results were promising. Across all the scan points, the robot recorded a maximum positional deviation of less than 0.010 mm, a level of accuracy that is unlikely to be achieved by manual human operation.

Beyond accuracy, the system also demonstrated strong responsiveness. The robot completed the full scan in less than a minute, moving fluidly from point to point without pauses or jerky transitions. The data showed that the robot’s controller generated a continuous path rather than stopping at each intermediate point, resulting in a smooth sweep across the mannequin torso. This kind of uninterrupted motion is essential for real world imaging, where sudden stops can disrupt probe contact or distort readings. Smooth motion also reduces the likelihood of patient discomfort and helps maintain consistent pressure throughout the scan.

While the current stage of research focused on robot motion behaviour rather than actual imaging, the findings suggest that the proposed system has the potential to help standardize external torso imaging in clinical settings. By ensuring uniform coverage and reducing operator dependent variability, such a system may support more reliable data collection. It may also help streamline training requirements, as new clinicians would rely less on mastering complex scanning techniques and more on interpreting high quality, robot acquired data.

As hospitals continue to seek ways to improve efficiency and consistency, robotic systems like this may soon become valuable partners in handheld medical imaging rooms. Instead of having to operate an ultrasound probe manually, radiologists may work in a control room to monitor and verify the operation, data collection, and analysis from robotic assisted handheld medical imaging systems. This shift could mark a significant step toward more automated, scalable, and patient friendly diagnostic workflows.

This work was developed by Sara Hany Tawfik Hussien (PhD candidate), Assoc Prof Ir Dr Riady S. Jo and Dr Richard Wong Teck Ken.

Associate Professor Ir Dr Riady S. Jo & Sara Hany Tawfik Hussien (PhD Candidate)
School of Engineering
Faculty of Engineering and Technology
Email: @email