Toward the Design of Personalized Continuum Surgical Robots

Abstract

Robot-assisted minimally invasive surgical systems enable procedures with reduced pain, recovery time, and scarring compared to traditional surgery. While these improvements benefit a large number of patients, safe access to diseased sites is not always possible for specialized patient groups, including pediatric patients, due to their anatomical differences. We propose a patient-specific design paradigm that leverages the surgeon's expertise to design and fabricate robots based on preoperative medical images. The components of the patient-specific robot design process are a virtual reality design interface enabling the surgeon to design patient-specific tools, 3-D printing of these tools with a biodegradable polyester, and an actuation and control system for deployment. The designed robot is a concentric tube robot, a type of continuum robot constructed from precurved, elastic, nesting tubes. We demonstrate the overall patient-specific design workflow, from preoperative images to physical implementation, for an example clinical scenario: nonlinear renal access to a pediatric kidney. We also measure the system's behavior as it is deployed through real and artificial tissue. System integration and successful benchtop experiments in ex vivo liver and in a phantom patient model demonstrate the feasibility of using a patient-specific design workflow to plan, fabricate, and deploy personalized, flexible continuum robots.

Keywords: 3D printing; Continuum robots; Minimally invasive procedures; Personalized surgical robots.

Fig. 1.

The proposed patient- and procedure-specific design workflow includes a 3D reconstruction of the anatomy based on medical images, followed by the design and fabrication of the personalized tools, and finally deployment of the robot in a procedure.

Fig. 2.

The surgeon design interface includes (a) virtual reality headset and six degree-of-freedom haptic device. The headset is used to render (b) a 3D model of the relevant patient anatomy, (c) Foot pedals are used to switch between the different interaction modes of the interface, (d) Concentric tube robot parameters shown in blue are preselected and include the number of tubes (n), outer and inner diameters (OD_i, ID_i), straight length (L_si), and translation actuator distance (β_i). Concentric tube robot parameters shown in green can be modified with the design interface and include individual tube curvature (κ_i,), curved length (L_ci), rotation actuator angle (α_i), and Young’s Modulus (E_i). (e) Iterative workflow for the surgeon design interface. Inputs to the interface include the reconstructed model of the patient anatomy and surgeon-defined tube parameters, and the output is a set of patient-specific concentric tube parameters.

Fig. 3.

Initialization of a concentric tube robot design includes (a) using the interface to place via points through which the robot will ideally pass, (b) fitting a piecewise, constant-curvature spline to the via points, (c) computing the forward kinematics of the robot and aligning the configuration with the via points, and (d) displaying in the interface the initial design to the surgeon. Interactions with the virtual concentric tube robot can change the (e) curvature, (f) curved length, and (g) actuator angle of any of the tubes.

Fig. 4.

Sample tubes printed with polycaprolactone (a) during the 3-D printing process using a MakerBot Replicator 2X and shown as (b) individual tubes and (c) assembled into a three-tube concentric tube robot. Modular actuation and control system shown (d) deploying a three-tube concentric tube robot and (e) separated into individual actuation modules.

Fig. 5.

Experimental setup for measuring the forces resulting from driving a three-tube concentric tube robot through various tissues. Inset image shows tip of the concentric tube robot with attached sharp needle tip added to enable easy tissue penetration.

Fig. 6.

(a) Two potential concentric tube robot paths, entering under the 12th rib and snaking up into the kidney to the diseased site (either a stone or tumor), (b) Phantom patient model based on CT scans from a 9-year-old patient. Model was constructed from 3-D printed organs and a gelatin kidney, (c) Teleoperation scheme developed to enable movement that best approximated follow-the-leader deployment, (d) Actuation system mounted on a passive positioning arm for easy maneuverability.

Fig. 7.

Results of Study 1. Magnitude of forces during insertion of a three-tube concentric tube robot through (a) gelatin and (b) cow liver. The approximate follow-the-leader deployment sequence is shown at the top of (a) as all three tubes were extended simultaneously to 10 mm, the inner two extended to 20 mm, and the innermost tube extended to 30 mm. (c) Peak insertion forces reached over 9 N during insertion of a concentric tube robot into the kidney, our target organ.

Fig. 8.

Results of Study 2. (a) Screenshots with artificially highlighted (yellow) concentric tube robot and (blue) 12th rib. The first column shows Set 1 which was designed as a “c-shape”, and the second column shows Set 2 which was designed as an “s-shape”. Two different views are given, and these views are slightly different for each set since the images are acquired from the surgeon’s view during the design process, (b) Example of time spent by a pediatric urologic surgeon in various phases throughout the design process. Phases include initialization of the design, controlling the camera, moving the concentric tube robot with respect to the anatomy, and time spent doing design, computation, and observation, (c) Three viewpoints of a surgeon-designed concentric tube robot (artificially highlighted in blue) curving below the 12th rib, into the kidney, and up to the tumor. The two outermost tubes (Tube 2 and Tube 1) are labeled in each view, and the innermost tube (Tube 0) is inside the kidney and therefore not visible, (d) Closeup view of the concentric tube robot inside the kidney with the ablation probe successfully in contact with the target tumor, (e) Insertion values (β_iactual) during actual teleoperation to a target compared to planned (or desired) values β_ides).