Simulation-Based Training in Robotic Surgery: Contemporary and Future Methods

Introduction

The acceptance of, role for, and familiarity with robotic surgery have grown exponentially over the last decade. Originally introduced in 2000, surgical robots remained at the fringe of many operating suites in their first decade of use, gaining only devoted adoption by surgeons performing a limited number of pelvic operations (i.e., prostatectomy and hysterectomy). A second wave of robotic surgery, however, has brought wider adoption in thoracic surgery, transoral robotic surgery (TORS), complex hepatobiliary operations, nearly all colorectal cases, and a wide array of both complex and routine hernia repairs. Along with this broader adoption, there is a growing need for effective and standardized training in robotics for new surgeons, as well as for surgeons new to robotics but with experience in open and laparoscopic surgery. Because of its video-based format, simulation has played a central role in robotic training paradigms. In this review, we discuss existing simulation modalities for training in robotic surgery and explore areas for future development of simulation.

The Need for Simulation in Robotic Training

While simulation plays a role in all surgical training, there exists a unique need for simulation in robotics. The following conditions, in particular, argue for a greater need for simulation in robotics than elsewhere in surgical training. First, the operating experience in robotics is fully immersive. By interfacing complete visual, auditory, and sensory exchange with the robotic console, the learner must master new neural and coordination tasks that are not otherwise transferrable from previously learned tasks. Second, the adoption of robotic techniques is a major paradigm shift from other modes of surgery with which the learner may have experience. It has been shown that minimal robotic skill is derived from open and laparoscopic surgery skill sets. ¹ Third, the operating experience in robotics is not as amenable to hands-on teaching as in open and even laparoscopic surgery. Robotic instrument control presently only permits single-master/single-slave pairing. Although a dual console facilitates instrument switching and on-screen illustration, the instructor cannot guide or demonstrate a maneuver in real time without fully resuming control of the robot. Fourth, because of their inherent complexity, the learning curve for some robotic operations (e.g., pancreaticoduodenectomy) is sufficiently high to be both impractical and prohibitive to adopting the technique only in clinical practice. Nonetheless, it has been demonstrated that structured simulation along with dedicated mentorship can reduce the number of procedures required to master a new robotic operation.^2,3 Fifth, while many changes to open and laparoscopic surgery are incremental (e.g., a new technique or a new instrument), the technological changes in robotic surgery often represent quantum leaps forward in experience for the surgeon. For example, with the introduction of the da Vinci single port (SP) robotic system (Intuitive Surgical, Sunnyvale, CA), the operating room setup, console controls, camera function, and instrument design are substantially different from the conventional multiport robot. ⁴ Therefore, an entire generation of robotic surgeons will need to learn, through simulation, this novel robotic platform. This pattern of big leaps forward in robotic technology will likely be duplicated many times in the near future with other robotic platforms entering the market. Finally, while historical training models in surgery had fully separated training from certification and credentialing, simulated environments can provide a platform for standardized training along with simultaneous skill assessment and determination of qualification.

The Environments and Phases of Simulation in Robotic Surgery

Robotic simulation can take place across multiple environments. At its most basic, dry laboratory simulation uses fashioned models for practice in cutting, dissection, and suturing practice using actual robotic instruments within a “trainer box.” A great deal of familiarity and skill can be achieved in the dry laboratory environment, but this comes at the not-insignificant cost of the finite number of allowable uses of robotic instruments and occupies time on a robot that would otherwise be used in operations. The degree to which dry laboratory materials simulate the look and feel of human tissues (i.e., face validity) is arguably the greatest limitation of this environment. In the wet laboratory environment, cadaveric and animal models provide an excellent platform for learning the responsiveness of tissue to robotic instruments, mastering gentle tissue handling while coping with the absence of haptic feedback, and learning techniques such as hemostasis and cautery that cannot be simulated in a dry laboratory. The training models in a wet laboratory, however, are often prohibitively expensive and may be only available as a single experience for most trainees. Perhaps the most widely developed environment for simulation is virtual reality (VR). The VR environment allows for an infinite variety of training scenarios and can achieve incredible realism while remaining fully reproducible and amenable to programmed assessment and feedback. Because of the three-dimensional visual outputs of robotic consoles, robotics provides the optimal substrate for VR-based simulation environments. ¹

Phases of simulation range from fully deconstructed tasks that build fine motor skills to entirely immersive operations that are simulated start to finish integrating dissection, assessment of pathology, decision-making, and management of complications. Simulated tasks include activities such as moving rings from peg to peg and arranging shapes. These tasks build instrument control, spatial awareness, arm clutching, and camera management. For beginners, simulated tasks hold perhaps the greatest yield for learning to be comfortable and relatively safe at the robot console and therefore will occupy a preponderance of initial training time. Simulated operative skills include individual surgical activities such as needle handling and application of electrocautery. They begin to approximate the motor memory required to appropriately undertake basic operating. Mastering simulated operative skills will help avoid blunders such as applying cautery to the wrong instrument arm. Training in this phase can spare precious moments in the simple but repetitive task of transferring a needle from one hand to the other and appropriately grasping it along the mid shaft of the needle. Simulated operative steps facilitate mastery of a complete surgical activity such as sewing a bowel anastomosis. While refining the previously acquired operative skills, this phase demands the learner to manage suture length and position, orientation and retraction of tissues, and efficiency in camera use and coordination with the third arm. Simulated operations bring together all prior components of skill while training learners to master coordination of steps and remaining mindful of timeliness. Finally, future modalities of robotic simulation yet to be developed might include simulated operating rooms, including platforms for bedside assistants, surgical technologists, and circulating nurses. Examples of simulated tasks, simulated operative skills, simulated operative steps, simulated operations, and simulated operating room orchestrations are listed in Table 1, which demonstrate examples of these phases of simulation.

Establishing Validity in Robotic Simulation

Psychometric analysis describes the utility of a learning tool in terms of its face, content, and construct validity. Face validity represents the degree to which the simulated environment represents reality (e.g., does the simulated tissue look/feel real?). Content validity is a measure of the usefulness of a training tool from the perspective of an expert user (e.g., is the simulated task one that surgical trainees need to master?). ⁵ Finally construct validity represents the ability of a learning tool to differentiate degree of skill acquisition (e.g., can one differentiate between the performance of an intern and a fellow?). For example, when untrained subjects took part in intensive skill training on simulated tasks they later demonstrated higher scores in knot tying than subjects who did not receive skill training. ⁶ An ideal robotic simulation platform will establish face, construct, and content validity. ¹ In one study, eye movement and pupil diameter tracking along with electroencephalogram (EEG) measurements were used to sample cognitive engagement, gaze entropy, and task workload assessment among robotic surgery trainees. Researchers found that changes in cognitive and behavioral states predicted training outcomes with 72.5% accuracy. ⁷ These data demonstrate that when trainees are well engaged with the VR tasks, they tend to more reliably acquire the trained skill.

A higher standard for simulation is predictive validity, which measures the extent to which performance in simulation reflects future performance in actual robotic surgery. ¹ Predictive validity can be difficult to establish because of the wide variety of anatomic and pathological variables across the spectrum of surgical procedures. ⁸ Culligan et al. demonstrated predictive validity by association of performance on VR robotic simulators with performance in the actual robotic operating room (OR), including improved blinded skill assessment, lower volumes of blood loss, and shorter operative times. ⁹

Robotic Simulation Platforms

Several simulation platforms for robotic surgery training and skill assessment are commercially available.^1,10 The SEP robot training system (SimSurgery, Norway) provides a low-cost VR-based trainer using a computer monitor interface. The Robotic Surgical Simulator (RoSS, Simulated Surgical Systems, Buffalo, NY) is a fully integrated VR-based trainer. The Robotix Mentor™ package (3D Systems, Israel) includes several very life-like whole operation modules (e.g., radical prostatectomy). ¹¹ The dV-Trainer^® (Mimic Technologies, Inc., Seattle, WA) and da Vinci^® Skills Simulator (dVSS; Intuitive Surgical) are most closely aligned with currently available surgical robots. As a result, the dVSS is now widely used and is commonly preferred over other simulators in terms of both ergonomics and usability. ¹⁰ The dVSS does, however, occupy a fully functional console that could otherwise be used for operating. While many published studies have assessed these simulators and face and content validity has been established for each, ¹² there remains a need for a trial of virtual simulation for acquisition of competency and safe techniques in robotic surgery.^13,14

Optimal Features of a Simulated Robotic Training Curriculum

Along with the variety of simulation platforms, there is a multitude of training curricula for robotic training. Rather than prescribing a single curriculum for robotic simulation, one can glean best practices through these thoroughly published training programs. Ideal curricula in robotics include proficiency-based VR simulation, inanimate biotissue dry laboratory modules, video library training, intraoperative evaluation, and skill maintenance with ongoing assessment.^15–17 While open surgical skills do not tend to directly translate to robotic skills, learners with extensive open skills can expect to acquire skills more rapidly through simulation than those less experienced in any form of surgery. ¹⁸

Some have argued that VR-based robotic trainers can provide self-directed and mentor-free learning. However, evidence from randomized controlled trials demonstrates that skilled mentorship on VR trainers can dramatically improve skills over controls. ¹⁹ Nonetheless, curricula should not limit the number of attempts made by learners at any given task, especially with beginner skills. In one study, an average of 74 attempts was needed before a plateau of skills. ¹ In addition to oversight by skilled mentors, learners can gain meaningful feedback through use of automated performance metrics such as efficiency in use of working space and theoretical force applied to tissue. ⁸ In a trial of trainees performing a simulated task with and without feedback from a social networking webpage, those with feedback scored higher on repeat assessment and achieved faster times. ²⁰ Regardless of the specific content of the curriculum, there is agreement that robotic training curricula should integrate simulation into existing surgical training programs. ²¹ Lendvay et al. have demonstrated that VR-based robotic warm-ups before actual operations can reduce both procedure time and intraoperative errors. ²²

Nontechnical Skills

To date, there is a paucity of training in nontechnical skills as they pertain to robotic surgery. ⁸ Nontechnical skills include both cognitive and interpersonal skills such as decision-making, leadership, teamwork, and situational awareness. As most surgeons trained in an era where the entire OR team stood at the bedside, the paradigm shift to a remote surgeon console demands awareness of the OR from a new perspective. Surgeons may benefit from coaching in communication that relies less on visual cues and more on closed-loop communication, designated roles, and careful use of perioperative checklists. Simulation for events such as safe emergent conversion from robotic to an open operation requires excellent teamwork, communication, planning, and other nontechnical skills. ²³

Simulation of the Operating Room Environment

With the addition of the surgeon console, robotic patient-side cart, and vision cart to the OR table, anesthesia machine, overhead lights, monitors, surgical instruments, and back table, the setup and flow of a crowded robotic operating room require advanced planning. Coordination of the flow of personnel and equipment can benefit from simulation. While VR-based robotic simulation focuses on the activities within the field of view at the surgeon's console, a great deal of activity outside of this field of view plays a role in the safe conduct of surgery. Commonly, surgeons master the technical gestures of instrument control on-screen yet struggle with robotic arm clashes, difficulty in accessing critical structures, and mastery of OR working space. In addition to technical console training, robotic surgeons must also train to avoid pitfalls of torque on the body wall, gas insufflation of the subcutaneous tissues, inadvertent injury to the patient from external arm trauma, pressure and strain injury from positioning, and injury from instruments inserted off-screen. All of these tasks differ from those in the open and laparoscopic environment, and since many of these tasks are shared, proper instruction of bedside assistants and OR staff is essential.

Bedside (i.e., patient side) training is important for learning the steps of the operation, bedside skills, and understanding the function of the robot. ¹¹ The Xperience™ Team Trainer (Mimic Technologies) facilitates the training of bedside assistants and emphasizes the importance of teamwork. Face and content validity has been measured and reported in observed training of both inexperienced and experienced bedside assistants. ²⁴ This trainer utilizes team-based task completion (match board, ring walk, etc.) and trains assistants on handoffs, retraction, and clip application.

Learning to Acquire Synesthesia and Intuition

Synesthesia is a process by which the brain learns to see tension in the tissue. In the absence of haptic feedback, the nervous system transforms the visually acquired cues of tissue tension into a perceived tactile sensation of tension. ²⁵ This acquisition of synesthesia is essential to being able to effectively and safely take advantage of the fine motor movements and magnification afforded by the robot. It is believed that routine practice leads to the acquisition of synesthesia. Robotic surgeons and bedside assistants must also acquire intuition to perceive when something is wrong about the configuration of the arms, an arm is jammed, a port is malpositioned and not transmitting an instrument, there is an external clash between arms, or any collision with the patient. Because of the immense force generated by the robot, all these conditions must be recognized and corrected even when the signs of a mechanical problem are subtle. Team-based training both with simulators and in-room equipment can build these intuitive skills.

Simulation of Complex Operations

With a rising number of advanced operations now considered amenable to a robotic approach, there is a need for simulation modules to match their complexity. There are now VR-based modules to simulate subspecialized operations such as TORS. ²⁶ Comprehensive programs for training in robotic pancreaticoduodenectomy use time-based proficiency goals for allowing trainees to progress to more complex steps of the operation.^2,16,17 Despite the abundance of VR-based training, tissue-based models still retain some relevance for training to perform advanced procedures. In a randomized controlled trial, wet laboratory simulation of robotic dissection of the internal mammary artery and mitral valve annuloplasty yielded better improvement in scores compared with VR simulation, dry laboratory simulation, and controls. ²⁷

Standards for Certification Through Simulation

Like the previously described Fundamentals of Laparoscopic Surgery and Fundamentals of Endoscopic Surgery, the Fundamentals of Robotic Surgery (FRS) has been developed as a curriculum and testing standard using a simulated, dry-laboratory exercise, and animal model. ²⁸ FRS was designed to be device- and specialty independent and is effective in improving technical skills in addition to serving as a standard for competency. ²⁸ An internationally used metric for assessment of skills is the Global Evaluative Assessment of Robotic Skills (GEARS). ²⁹ This evaluation applies observer-scored values for depth perception, bimanual dexterity, efficiency, force sensitivity, autonomy, and robotic control. Regardless of the standard, it can be expected that standardized skill certification will become more ubiquitous.

Future Directions in Robotic Simulation

The future of robotic simulation will include advanced VR modules, personalized feedback from expert mentors and sophisticated analytic software, and certainly granular assessment of surgeon qualification. As new robotic modalities such as computed tomography (CT) guided navigation, heads up imaging overlay, and automation emerge, it will be up to both the designers and users to determine how simulation of these modalities will be acquired. Surgeons should anticipate ongoing paradigm shifts that will require retraining and therefore facilitate continued applications of simulation (e.g., natural orifice-based surgery, SP modalities, robotic mastectomy, and so on). ¹⁵ While competitor robots to the da Vinci are in development and have even been approved, the need for adaptive training systems will only increase as these robots become more widely available.