Teleoperation of Steerable Needles

Abstract

Needles are commonly used in medical practice as a minimally invasive means to reach subsurface targets for diagnosis or therapy delivery. Recent results indicate that steerable needles may enhance targeting accuracy and allow needles to avoid obstacles along the path to the target. This work considers teleoperation of needles made of a superelastic alloy that steer through tissue using forces generated by the standard asymmetric bevel tip. The needle may be modeled as a nonholonomic system, with inputs of insertion along and spin about the needle axis. A teleoperation system consisting of a commercial master haptic device, a custom needle-steering robot slave, and visual feedback to the operator was assembled. Human subjects experiments were performed to evaluate targeting accuracy in phantom tissue for three needle control methods: teleoperation of both insertion and spin, teleoperation of insertion with open-loop-controlled spin, and open-loop control of both insertion and spin. Targeting accuracy improved with increasing degrees of freedom of human (teleoperation) control, primarily because tissue deformation and modeling limitations result in open-loop control errors. Subjects typically performed multiple spins of the needle during insertion in order to fine tune the needle path. In addition, position, rate, and a nonlinear hybrid control were compared during teleoperation of the insertion degree of freedom. The hybrid method resulted in significantly better targeting accuracy