2016 Section 5 Green Book

number of minimally invasive surgical techniques. 3,13–17 Despite advances in software, computer processing speeds, and registration algorithms, very few of these systems appear to be used during routine procedures. Poor adoption of computer-based assistance through advanced displays has been noted in many industries where technical preci- sion is required. 18,19 Although there is huge potential for this technology, functional deployment has been restricted by limited human factors research investigating interface design. 18 It was a significant undertaking to create a mock OR with a full set of skull base equipment and recruit experienced skull base surgeons for a 5-hour trial. We wished to make the setting, the procedure, and the sub- jects (end users) as close to reality as possible to obtain the feedback required to make the system changes nec- essary for integration into the OR. We gained valuable insight into how different surgeons used the system and noted many similar themes as well as individual differ- ences. Such observations are useful in adapting feedback to be as functional as possible while allowing customiza- tion where requested. The mental demand, effort, and frustration levels were shown to be significantly decreased when using the LIVE-IGS system. All surgeons also appreciated that the overall stress level was much lower operating on a cadaver than a live patient. Drilling adjacent to the carotid or toward the dura in a real operation would be significantly more stressful, and the workload differences may be even greater in a real situation. Subjects commented that it was less demanding to drill when they were some distance (e.g., around 5 mm) away from a critical structure and could quickly check their anatomical position and continue to ablate with confidence. They noted that if they were using traditional style intermittent IGS, or a Doppler probe, they would have checked less frequently, due to the time and interruption required to switch to a different probe, thereby continuing with ablation in a slower and more cautious fashion. We believe that the lower task workload scores reflect this level of reassurance offered by the technology. Incorporating testing of mental workload with tools such as the NASA-TLX has been recommended by other authors, given that traditional outcome parame- ters such as morbidity and mortality are practically impos- sible to quantify in this style of translational research. 20 Inferior task performance and a greater number of errors have been noted when mental workload is increased. 21 When providing a 3D virtual view of anatomical contours, a reference framework must be included to allow surgeons to conceptualize how they relate to the real-life anatomy they are viewing. A mesh surface ren- dering was presented to place the contours in context. We had mixed reviews on the virtual display, particu- larly in regard to the mesh. Some wanted less mesh, some wanted a more realistic surface rendering, and others did not use it or found it confusing. There was certainly no consensus on the best way to display 3D contours, and given this we believe allowing customized display settings incorporating various virtual views with adjustable opacity as well as fused augmented reality views may be appropriate.

The integration of live visual displays and auditory alerts involves additional stimuli that inherently demand some attention. Strategies to mitigate unneces- sary distraction are important in human–computer interface design. 22–24 Like some other groups, we placed all of the visual augmentation on the submonitor to avoid direct visual stimulation at critical points. 3,25 This design led to most surgeons referencing this data at their discretion rather than forcing their attention through on-screen visual cues. When abstract sound alerts were used as proximity alerts, we noted the sur- geon would often scan the submonitor to see why it was alerting. Auditory icons were developed and replaced abstract sounds for the proximity alerts after the first three subjects suggested that individual identification of structures would be preferable. Auditory icons are alarms that bear some relationship to their function and are easily learnt. 26 For the carotid artery, we created an alarm reminiscent of an arterial Doppler trace for the proximity zone and added a beep at the end when the instrument was tracked to be within, or extremely close to, the carotid volume. This was almost instantly recog- nizable and learnt by surgeons and provided informative feedback on the reason for alarm without disrupting dis- section. It was observed that the addition of this feature was associated with decreased scanning of the submoni- tor, and subjects reported being better informed and less distracted. Driving simulation studies have shown simi- lar results with visual distractions causing more erratic steering than auditory distractions. 27 Sonification, turn- ing data into sound (such as the change in tone with pulse oximetry), can inform without the need to scan for visual data. Investigation in anesthesia simulation showed sonification allowed greater time-sharing per- formance between a manual and monitoring tasks when compared to visual monitoring. 4 Although auditory icons and sonification show great promise in limiting distrac- tion, the plethora of other alarms and auditory stimuli in the operating environment must be taken into account. 26 This trial has provided our research group with a better understanding of the issues limiting clinical implementation of this technology as well as the poten- tial clinical uses. We feel that reporting the preclinical development of our system is important to promote effi- cient progress toward improved surgeon–computer inter- face design. Although registration accuracy and robustness will always be a concern, this does not appear to be a significant factor restricting adoption, especially within the relatively rigid framework of the skull base. The main barriers appear to relate to inter- face design and the lack of integration of the elements required to produce the features described. Streamlining the system into a single package that allows intuitive contouring, rapid registration and instrument calibra- tion, and the ability to customize visual display and alarm settings could make this technology only margin- ally more complex to set up than current IGS systems. We are unsure whether the cost and time required to provide this information would make it suitable for more routine cases such as endoscopic sinus surgery. Further

Laryngoscope 124: April 2014

Dixon et al.: Real-Time Navigation for Endoscopic Surgery

173