Guiding Medical Needle Insertions Using Robotic Tissue Manipulation
Meysam Torabi, Kris Hauser, Ron Alterovitz, Vincent Duindam and Ken Goldberg
This project investigates the use of robotic tissue manipulation devices to help guide medical needle insertions, in order to help improve targeting accuracy and also shift sensitive tissues away from the needle's path. We use a control system that displaces points deep inside the tissue using both feedforward and feedback control of the manipulator. A PI controller uses feedback from a medical imaging device, such as ultrasound or fluoroscopy, and feedforward displacements are computed using a linearized, quasistatic model of the tissue.
We also use an automated planning procedure that uses a stochastic optimization to choose needle insertion points, manipulation points, and controller parameters. The objective function trades off between robustness to disturbances and tissue stress.
Our system has been evaluated in 2D mass-spring tissue simulations, including simulations of a prostate brachytherapy procedure. It reliably achieves millimeter-level targeting accuracy and avoids piercing sensitive tissues. In future work, we will integrate more realistic 3D tissue and needle models, combine tissue manipulation with various needle steering technologies, and investigate applications to other medical procedures.
Figure 1: A needle insertion with tissue manipulation. The tissue is manipulated to move an obstacle (blue region) downward and the target (red dot) upward.