September 20th, 2024
Virtual simulations allow real humans to interact with simulated environments and simulated systems. When those real humans are novice surgeons, using a simulated environment holds many benefits. When learning to cut and probe on humans, simulations must be considered. The skills required of these novice surgeons require practice to be mastered. A thoughtfully designed virtual simulation can save much pain and suffering in the training process.
In this article, I consider a skill trainer described by Rose and Pedowitz (2015) . Arthroscopy involves the use of a small camera and other instruments entering the body through incisions to inspect of joints. The surgeon must carefully manipulate the tools to avoid damaging surrounding tissue. Visual feedback is provided from a monitor displaying the view of the internal camera. This visual disconnection and spatial offset adds a complication to the task.
The authors described a simulation that used manipulators with force-feedback and a visualization that was presented via laptop screen to the participants. The authors cited Aggarwal et al. (2006) to relate that teaching 12 basic skills instead of the two most complex is more effective for learning. With that perspective, the simulation was designed to focus on three core tasks. These core tasks were distilled by deconstructing the more complex skills required. The authors’ aim was to provide a consistent and reproducible training aid for developing basic motor skills in novices. Their research demonstrated the simulator engaged the same motor skills used in surgery.
In considering the impact of virtual simulations on the general case of motor skill development, I looked at a detailed study on brain activity (Kamat et al., 2022). Their study followed activation of structures within the brain during training. Their use of advanced imaging techniques revealed details of specific brain activation.
If we consider the simulation described in Rose and Pedowitz (2015) in terms of the Modeling and Simulation Framework of Zeigler et al. (2019), the core components and their interactions can be identified. The experimental frame in this case encompasses the way in which an arthroscopic surgeon manipulates their tools during their work. The biological material and its differing textures and toughness were excluded from consideration. The stress of working on a living being were excluded from consideration. The experimental frame was ruthlessly scoped to the most basic aspects relevant to moving the tools during arthroscopic tasks. The movements were the focus. The set of behaviors was drawn from a task deconstruction of arthroscopy. This experimental frame was used for the model.
The model’s inputs and outputs were chosen to support three specific tasks in a training context. It received input from the controls and produced both force-feedback and updates to the visual display. The tools were represented within the display area as virtual figures. The model was designed to measure performance on three tasks: Steady and Telescope, Steady and Probe, and Track a Moving Target. The model also provided tolerances within which the participant would be considered on-target, but outside of that range indications were given to guide them back.
The simulation that gave life to this model provided the virtual environment in which participants interacted with their tasks. For the Steady and Telescope task, the simulation presented the user with a highlighted yellow target to hold in focus for five seconds. The authors indicated this task did not discriminate between experience levels - something they suggest may be corrected by increasing the difficulty. The Steady and Probe task required the user to keep the view centered using one hand while using the other hand to push small spheres into a target sphere. The Track a Moving Target task required the user to keep a target centered in the visual display while pushing a sphere along a track. Noteworthy in this last task is the feedback received as the sphere starts to deviate from the track. The sphere would change colors to indicate it was too far from the center. The experimental frame provided the tolerances, the tasks, and the tools which shaped the model. The simulation gave the model interactivity and collected information about the participants’ performances. A multitude of data points were collected, chief among them being the time to complete each task. In this simulated environment, the components were truly interconnected.
Using this virtual simulation exercised the same skills needed by its referent - arthroscopic surgery. The relationship between virtual trainer and motor skill development was investigated in Kamat et al. (2022). Their study examined brain activity and the biological structures involved during virtual simulation training. The authors described “internal models” built up through interactions with the environment. The feedback from body movement was crucial for the development of those internal models. Virtual trainers, such as the arthroscopic system described here, provide means for that feedback, aiding in development of motor skills. A subsequent study by Liao et al. (2023) validated the hypothesis by showing a statistically significant improvement in novice performance after a mere 70 minutes of training. Improvement in such a short time demonstrates the effectiveness of virtual simulation as a training mode.
When training motor skills for dangerous, expensive, or risky tasks, a virtual simulation should be considered. It has been shown that such simulators can lead to performance improvements with the targeted skills (Kamat et al., 2022; Liao et al., 2023). Moreover, Rose and Pedowitz (2015) demonstrated that a simulation based on the most essential elements of the required movement can be validated. Their research showed experience levels could be correlated to performance on two of the three tasks. Additionally, the level of ambidexterity on any of the three could be used to discriminate between experience levels. On the topic of ambidexterity among surgeons, Rose and Pedowitz (2015) demonstrated experimentally that more experienced surgeons displayed a common level of ambidexterity exceeding that of novices. Input devices that provide force-feedback of simulated tools can also be improved. Tsujita et al. (2022) described a force-feedback device intended for use in training neurosurgeons. Brain matter harvested from a pig was used to collect the resistance levels used in their model. Visualizations can be of greater and greater fidelity as the need dictates. Difficulty can be specifically tuned to the need. If one combines these facts with the reduced expense and increased safety of virtual simulators, it seems they should be considered in all cases.