Position-Based Teleoperation of Steerable Needles

Steerable needles can be used to improve the effectiveness of needle-based clinical procedures by allowing needles to reach targets and avoid critical structures (e.g. nerves, vessels, bones) though curved paths. These needles may be difficult to control manually due to underactuation and kinematic constraints. We propose a robotic-assisted teleoperation paradigm to aid the placement of steerable needles with a haptic interface.

Previously, our group has teleoperated tip-asymmetric steerable needles using "joint space" control, where the user controls the input degrees of freedom to the needle steering robot (i.e. insertion and rotation). Here, the user has the burden of predicting the nonholonomic behavior of the needle resulting from their inputs. In our current work, the needle is teleoperated using "spatial" control, where the user controls the 3D needle tip position. Preliminary human subjects experiments indicate improved performance for reaching targets with steerable needles when using the position-based teleoperation algorithm.

Reconstructing Curved Needle Shapes using 3D Doppler Ultrasound Imaging

In the future, our group aims to use 3D ultrasound imaging to provide control feedback to robotic systems manipulating needles in tissue. Before this is possible, we must first develop methods for automatic segmentation of curved needles from 3D ultrasound data, which is a challenging problem. Needles have an irregular appearance in ultrasound images. In some configurations they are barely visible, while in others they generate substantial artifacts. Overall, it is difficult to implement computer algorithms that can detect and localize needles in ultrasound that are fast enough for real-time robot control. Our approach is to use the robot to generate controlled motion of the needle that can be recognized using ultrasound Doppler imaging. Ultrasound Doppler imaging reveals frequency shifts that result from moving tissue or fluids. Our current focus is developing models of needle-tissue motion, and using these models to create new filters that highlight moving needles in ultrasound Doppler data. While other groups have described using Doppler to identifying moving interventional instruments, our work is the first to reconstruct the entire curved shape of a needle in tissue using this approach.

People

• Troy Adebar
• Joey Greer

Support

• NIH R01: Steering Flexible Needles in Soft Tissue (NIH R01 EB006435)

Click here to return to the main research page.