Design Considerations for Robotic Needle Steering * Robert J. Webster III, Jasenka Memisevic, and Allison M. Okamura Department of Mechanical Engineering Engineering Research Center for Computer-Integrated Surgical Systems and Technology The Johns Hopkins University Baltimore, MD, 21218 USA robert.webster@jhu.edu, memisej@slu.edu, aokamura@jhu.edu Abstract— Many medical procedures involve the use of needles, but targeting accuracy can be limited due to obstacles in the needle’s path, shifts in target position caused by tissue deformation, and undesired bending of the needle after insertion. In order to address these limitations, we have developed robotic systems that actively steer a needle in soft tissue. A bevel (asymmetric) tip causes the needle to bend during insertion, and steering is enhanced when the needle is very flexible. An experimental needle steering robot was designed that includes force/torque sensing, horizontal needle insertion, stereo image data acquisition, and controlled actuation of needle rotation and translation. Experiments were performed with a phantom tissue to determine the effects of insertion velocity and bevel tip angle on the needle path, as well as the forces acting on the needle during insertion. Results indicate that needle steering inside tissue does not depend on insertion velocity, but does depend on bevel tip angle. In addition, the forces acting on the needle are directly related to the insertion velocity. Index Terms— Keywords: medical robotics, needle steering, nonholonomic systems. I. I NTRODUCTION Needle insertion is an important aspect of many medical diagnoses and treatments, particularly percutaneous proce- dures requiring therapy delivery to or sample removal from a specific location. However, errors in needle targeting can mitigate the effectiveness of diagnosis or therapy. Biopsies, for example, cannot completely rule out malignancy due to inaccuracy in positioning the needle tip. Also, radioactive seeds in procedures such as prostate brachytherapy are often placed at locations substantially different than those pre-planned for optimal dosage. Needle steering has the potential to correct targeting errors and steer around obstacles to reach previously inaccessible locations. Control and planning based on a steering model can compensate for targeting disturbances due to needle bending, error in insertion angle, and tissue deformation. In this paper, we focus on the design of two different devices for steerable needle insertion, as well as experiments to determine the effect of needle insertion velocity and bevel tip angle on needle path. ∗ This work is partially supported by NIH Grant #R21-EB003452 to A. M. Okamura, an NDSEG fellowship to R. J. Webster III, and NSF Grant #EEC-9731478 for an REU fellowship to J. Memisevic. In our systems, the steering effect is caused by the asymmetry of a bevel tip on a flexible needle [9], [11]. Lateral motion and tissue deformation can also cause steering, although our systems do not explicitly employ those techniques. Clinically, needles are manually steered through a combination of lateral, twisting, and inserting motions under visual feedback from imaging systems such as ultrasound [13]. However, these techniques can yield inconsistent results and are difficult to learn. Physicians also sometimes continually spin bevel tip needles during insertion to prevent them from bending. A. Previous Work The effect of needle bending for the purpose of steering has recently been explored by several groups. Examples of canula-based steering methods include the use of a pre- bent stylus inside a straight canula [6] and a telescoping double canula where the internal canula is pre-bent [4]. DiMaio and Salcudean [5] formulate a needle Jacobian that describes tip motion due to needle base motion and a deformable finite element tissue model. However, their work does not explore the effect of tip asymmetry. In addition, Glozman and Shoham [7] analyzed needle paths for steering and obstacle avoidance, but did not model the bevel tip. Our approach differs significantly from previous needle steering work because of the use of tip asymmetry. A pri- mary advantage of our technique is that it does not require significant deformation of potentially sensitive tissues, thereby minimizing tissue damage. Tissue damage could re- sult in changing material properties, making needle steering plans based on estimates of tissue properties inaccurate. We presented a nonholonomic model of needle steering and our preliminary experiments in [8]. Our group is also exploring reachability, stochastic modeling and probabilistic planning [12], as well as finite element modeling for planning paths around obstacles in deformable tissue [1]. In this paper, we report our efforts in the design of needle steering robots and the results of experiments to determine the effects of insertion velocity and bevel angle on steering. Proceedings of the 2005 IEEE International Conference on Robotics and Automation Barcelona, Spain, April 2005 0-7803-8914-X/05/$20.00 ©2005 IEEE. 3599