Lee W. White

Raven-II: An Open Platform for Surgical Robotics Research

The Raven-II is a platform for collaborative research on advances in surgical robotics. Seven universities have begun research using this platform. The Raven-II system has two 3-DOF spherical positioning mechanisms capable of attaching interchangeable four DOF instruments. The Raven-II software is based on open standards such as Linux and ROS to maximally facilitate software development. The mechanism is robust enough for repeated experiments and animal surgery experiments, but is not engineered to sufficient safety standards for human use. Mechanisms in place for interaction among the user community and dissemination of results include an electronic forum, an online software SVN repository, and meetings and workshops at major robotics conferences.

Content and Construct Validation of a Robotic Surgery Curriculum Using an Electromagnetic Instrument Tracker

PURPOSE Rapid adoption of robot-assisted surgery has outpaced our ability to train novice roboticists. Objective metrics are required to adequately assess robotic surgical skills and yet surrogates for proficiency, such as economy of motion and tool path metrics, are not readily accessible directly from the da Vinci® robot system. The trakSTAR™ Tool Tip Tracker is a widely available, cost-effective electromagnetic position sensing mechanism by which objective proficiency metrics can be quantified. We validated a robotic surgery curriculum using the trakSTAR device to objectively capture robotic task proficiency metrics. MATERIALS AND METHODS Through an institutional review board approved study 10 subjects were recruited from 2 surgical experience groups (novice and experienced). All subjects completed 3 technical skills modules, including block transfer, intracorporeal suturing/knot tying (fundamentals of laparoscopic surgery) and ring tower transfer, using the da Vinci robot with the trakSTAR device affixed to the robotic instruments. Recorded objective metrics included task time and path length, which were used to calculate economy of motion. Student t test statistics were performed using STATA®. RESULTS The novice and experienced groups consisted of 5 subjects each. The experienced group outperformed the novice group in all 3 tasks. Experienced surgeons described the simulator platform as useful for training and agreed with incorporating it into a residency curriculum. CONCLUSIONS Robotic surgery curricula can be validated by an off-the-shelf instrument tracking system. This platform allows surgical educators to objectively assess trainees and may provide credentialing offices with a means of objectively assessing any surgical staff member seeking robotic surgery privileges at an institution.

Crowdsourcing to assess surgical skill

What Is the Innovation? Surgical skills impact patient outcomes.1 Our profession needs methods that accurately and objectively evaluate surgical skills.2 These methods must provide timely and meaningful feedback, minimize review time burden, and scale for widespread use. This evaluation hurdle may be overcome by leveraging crowdsourcing to help triage outliers and focus improvement efforts. Crowdsourcing is the process of using large groups of decentralized, independent people providing aggregated feedback.3 Crowdsourcing to assess surgical skill is a method by which surgical technique can be assessed by crowds of reviewers, some of whom are nonmedically trained. Crowdsourcing has enjoyed broad success in health care—discovering protein-folding patterns, assisting disabled patients, locating automatic defibrillators within cities, and annotating electronic medical records.3

Crowd-sourced assessment of technical skills: an adjunct to urology resident surgical simulation training.

Abstract Background: Crowdsourcing is the practice of obtaining services from a large group of people, typically an online community. Validated methods of evaluating surgical video are time-intensive, expensive, and involve participation of multiple expert surgeons. We sought to obtain valid performance scores of urologic trainees and faculty on a dry-laboratory robotic surgery task module by using crowdsourcing through a web-based grading tool called Crowd Sourced Assessment of Technical Skill (CSATS). Methods: IRB approval was granted to test the technical skills grading accuracy of Amazon.com Mechanical Turk™ crowd-workers compared to three expert faculty surgeon graders. The two groups assessed dry-laboratory robotic surgical suturing performances of three urology residents (PGY-2, -4, -5) and two faculty using three performance domains from the validated Global Evaluative Assessment of Robotic Skills assessment tool. Results: After an average of 2 hours 50 minutes, each of the five videos received 50...

Comparison of Two Simulation Systems to Support Robotic-Assisted Surgical Training: A Pilot Study (Swine Model)

OBJECTIVE To compare the efficacy of simulation-based training between the Mimic dV- Trainer and traditional dry lab da Vinci robot training. DESIGN A prospective randomized study analyzing the performance of 20 robotics-naive participants. Participants were enrolled in an online da Vinci Intuitive Surgical didactic training module, followed by training in use of the da Vinci standard surgical robot. Spatial ability tests were performed as well. Participants were randomly assigned to 1 of 2 training conditions: performance of 3 Fundamentals of Laparoscopic Surgery dry lab tasks using the da Vinci or performance of 4 dV-Trainer tasks. Participants in both groups performed all tasks to empirically establish proficiency criterion. Participants then performed the transfer task, a cystotomy closure using the daVinci robot on a live animal (swine) model. The performance of robotic tasks was blindly assessed by a panel of experienced surgeons using objective tracking data and using the validated Global Evaluative Assessment of Robotic Surgery (GEARS), a structured assessment tool. RESULTS No statistically significant difference in surgeon performance was found between the 2 training conditions, dV-Trainer and da Vinci robot. Analysis of a 95% confidence interval for the difference in means (-0.803 to 0.543) indicated that the 2 methods are unlikely to differ to an extent that would be clinically meaningful. CONCLUSION Based on the results of this study, a curriculum on the dV- Trainer was shown to be comparable to traditional da Vinci robot training. Therefore, we have identified that training on a virtual reality system may be an alternative to live animal training for future robotic surgeons.

Finding a feature on a 3D object through single-digit haptic exploration

Humans use tactile and kinesthetic cues to easily identify the location of an object feature without vision. This preliminary work quantitatively examined the behaviors of 15 individuals locating a target on an object with a single digit (index or thumb). Search methods could be categorized into one of three search strategies, termed “scanning,” “landmark,” and “direct.” The scanning strategy consisted of (1)back and forth scanning and (2)confirmation of the target, and was always utilized in trials in which the target location was unknown. The landmark strategy consisted of (1)contour following, (2)ballistic movement from an edge toward the target, and (3)error correction and confirmation of the target. The direct strategy consisted of (1)ballistic movement from the start position directly to the target, and (2)error correction and confirmation of the target. Follow-up relocalization trials were executed with the landmark strategy 42% and the direct strategy 58% of the time. The landmark strategy was utilized significantly more often by the index finger than by the thumb. Future work to understand how strategy selection depends on the strength of an individuals internal model of the 3D object and the uncertainty of finger location with respect to the object will inform the development of a computational model of human object exploration.

Preliminary Articulable Probe Designs With RAVEN and Challenges: Image-Guided Robotic Surgery Multitool System

The primary focus of the vision systems in current minimally invasive surgery (MIS) surgical systems has been on the improvement of immersive experience through a static approach. One of the current limitations in an MIS robotic surgery is the limited field of view and restricted perspective due to the use of a sole rigid 3D endoscope. We seek to integrate a modular articulable imaging device and the teleoperated surgical robot, RAVEN. Another additional flexible imager can be helpful in viewing occluded surgical targets, giving increased visualization options. Two probe designs are proposed and tested to evaluate a robotized steering mechanism within the MIS robot framework. Both designs, a separate flexible imager and a fixed camera on a tool tip, did not show much improvement in reducing task completion time. The new system may have some potential in improved precise manipulation of surgical tools, which may offer safety benefits once the surgeon is trained. We have demonstrated feasibility of a novel MIS instrument imaging device to aid in viewing potentially occluded surgical targets. A new concept, a modular axis-shared articulable imaging probe located at the vicinity of a tool tip, is proposed for future evaluation. Full integration of the new flexible imaging device into the grasper of the RAVEN surgical robot is under study coordinated with clinicians.

Exploratory Visualization of Surgical Training Databases for Improving Skill Acquisition

A new visualization system analyzes multidimensional surgical performance databases of information collected via emerging surgical robot and simulator technologies. In particular, it has visualized force, position, rotation, and synchronized video data from 300 bimanual laparoscopic surgery tasks performed by more than 50 surgeons. To explore data, the system uses a multiple-coordinated-views framework. It provides techniques to select and filter multivariate time series data, visualize animated force plots in conjunction with contextual videos, encode multivariate bimanual tool trace data in 3D visualizations, and link visualizations to a database management system via a new generalizable data model. Insights and feedback from an interdisciplinary iterative design process and use case studies support the utility of visualization in this emerging area of data-driven surgical training.

Raven surgical robot training in preparation for da vinci.

The rapid adoption of robotic assisted surgery challenges the pace at which adequate robotic training can occur due to access limitations to the da Vinci robot. Thirty medical students completed a randomized controlled trial evaluating whether the Raven robot could be used as an alternative training tool for the Fundamentals of Laparoscopic Surgery (FLS) block transfer task on the da Vinci robot. Two groups, one trained on the da Vinci and one trained on the Raven, were tested on a criterion FLS block transfer task on the da Vinci. After robotic FLS block transfer proficiency training there was no statistically significant difference between path length (p=0.39) and economy of motion scores (p=0.06) between the two groups, but those trained on the da Vinci did have faster task times (p=0.01). These results provide evidence for the value of using the Raven robot for training prior to using the da Vinci surgical system for similar tasks.

SurgTrak — A Universal Platform for Quantitative Surgical Data Capture

The climbing costs of healthcare coupled with the alarming rate of malpractice suits have highlighted the need for an efficient and effective method of surgical training. In 2009, approximately 3.4 billion dollars in malpractice payments were awarded. A quarter of these claims stemmed from adverse events in the surgical setting [1]. Surgical errors also increase hospitalization time and adversely affect the health of patients, often resulting in death or major injury [1–3]. Improving surgical training methods has become a priority because the majority of mistakes in the operating room have been attributed to lack of skill and experience [4–6]. While the need for an improved surgical training system is clear, quantifying surgical competence has proved more elusive [7]. By capturing the metrics of surgical training, a surgeon’s skill can be objectively evaluated. Expensive, procedure-specific simulators are frequently utilized as a means to gather data. But this method is limited by the simulator’s capabilities. Surgical systems such as the Da Vinci surgical robot have also been used to collect data in the operating room [2]. This approach, however, is restricted to measuring the limited number of procedures performed with Da Vinci tools. Financial and legal obstacles also hinder system customization and preclude widespread use. Fabrication of a surgical metrics system is a cost effective alternative. Emphasis must be placed on how tool positioning is tracked and recorded as a wide variety of solutions are possible and prove to be effective. A variety of tool tracking techniques prove to be effective [2,8]. However, each new experiment requires a new tracking software implementation, which can be both costly and time consuming. To address these issues we present SurgTrak, an open, configurable software platform to enable quantitative data collection in surgical environments. SurgTrak combines a multitude of inputs from cameras, USB devices, or network devices and writes the values to a computer file. The biggest advantage of SurgTrak is the system flexibility. Many different sensors are available. SurgTrak has been designed for the surgical setting as a development platform to enable quantitative data collection and thus create objective metrics for surgical skill [9]. 2 Methods