Comparison between force and ultrasound image-based controller for autonomous ultrasound acquisition during varying tissue stiffnesses

The MURAB project combines pre-operative high-quality MR images, which provide excellent soft tissue imaging, with robotic-guided intraoperative ultrasound images to perform biopsies of breast tissue with possible cancer lesions. The precision and quality of the biopsy are dependent on the fusion between ultrasound and MRI. However, after the MR volume is acquired, the patient is moved, which may result in an error between the shape of the breast during the MR imaging and the robotic-guided ultrasound imaging.

The goal of the research is to optimize ultrasound acquisition during the scanning of a target area. An optimal ultrasound image should satisfy two conditions: first, the images should be acquired using a low contact force between the ultrasound probe and the tissue to minimize tissue deformation; and second, the contact surface of the ultrasound probe should be large enough to scan the target area in a reasonable amount of time. To satisfy these conditions, a force feedback controller is being developed, which will ensure that the force is minimized while the contact surface area is sufficient, using force and torque readings obtained from a force sensor.

The force sensor is placed in the end effector of a KUKA robotic arm, and the ultrasound probe is attached to the sensor plate. The force sensor will detect the forces applied at the transducer and use them in the robotic control loop to minimize the contact forces while satisfying the conditions mentioned above. In addition, a comparison between the force feedback controller and the currently used confidence-driven controller will be carried out.

To conduct this comparison, phantom models of varying compliances, which simulate different tissues, will be used, and ultrasound scanning will be performed using both controllers. The obtained results will be analysed to evaluate the performance of each controller in different tissues and to identify which controller is more suitable for each tissue.