Presentation
Using Functional Near Infrared Spectroscopy to Quantify Differences in Surgeon Cognitive Load
SessionPoster Session 2
DescriptionIncreases in health care professional (HCP) cognitive load, in a surgical environment, can lead to adverse consequences resulting from mental fatigue and cognitive fog. These consequences can lead to use error or use difficulty with a severity that can put patient safety at risk. Robotic-assisted solutions for orthopaedic procedures, inclusive of a navigation system to increase visibility and robotic-assistance to aide trajectory and guidance of instrumentation, can help increase surgical accuracy and efficiency. However, the novel interaction between surgical workflow and HCP has introduced new components to the overall experience. Therefore, the mental work of the HCP can potentially be different from the standard operating procedures they are accustomed to.
Previous research has shown that cognitive load is associated with increased blood flow and increased oxyhaemoglobin (HbO) in the blood of the prefrontal cortex [Li,2025]. Recently, functional near infrared spectroscopy (fNIRS) has been utilised to quantify changes in blood flow patterns. Function NIRS sensors are non-invasive; they employ near-infrared light propagating diffusely through the scalp and brain, for functional monitoring and imaging of human brain hemodynamics. Changes in hemodynamic have been shown to be linked to cognitive function and effort, specifically fNIRS time series will increase if levels of oxygenated blood increase and decrease if levels of oxygenated blood decreases. Therefore, by analysing the level of oxyhaemoglobin (HbO) and deoxyhemoglobin (HbR), conclusions can be drawn about how much a task is affecting the cognitive process of the HCP. Changes in blood flow and oxyhemoglobin concentration (∆C) both indicate more activity in the prefrontal cortex. As robotic-assisted workflows may involve interaction with a digital interface, it is crucial to understand if the changed interaction between HCP and surgical field is producing alteration in HbO and what the consequent effect is.
Feasibility of fNIRS data collection during robotic-assisted workflows is yet to be addressed and is limited in the wider surgical field. Quantitative user performance measurements recorded with fNIRS, while performing surgical workflows providing different user-medical device interactions, pertain directly to the overall cognitive workload of the system and action being executed. This data can be augmented with user perception measurements captured from standardized surveys executed after each action is completed. Survey results can be used to quantify user perception and assess the usability and overall user experience of the system used to execute the actions. System Usability Survey (SUS) data can be used to assess user perception of usability, satisfaction, benefit, effort and value. Questions in the SUS have undergone psychometric qualification, and assessed for reliability, validity, and sensitivity [Sauro,2012]. Surgery Task Load Index (Surg-TLX) data can be used to assess workload in six dimensions: mental, physical, and temporal demands, as well as task complexity, situational stress, and distractions. The Surg-TLX is a validated measure of surgeon workload [McNair,2023; Wilson,2011] which has been used previously to assess cardiac surgery [Kennedy-Metz,2020].
In the present study, use of fNIRS sensors were implemented to investigate the feasibility of collecting data in a simulated surgical environment, to quantify cognitive load and mental fatigue, by means of quantifying changes in hemodynamic. Comparisons were made between the mean ∆C for active robotic-assisted guidance versus fluoroscopic guidance alone, and the mean ∆C for freehand navigation versus fluoroscopic guidance alone. Fluoroscopic guidance uses continuous X-ray imaging to provide real-time, moving X-ray images during medical procedures, allowing doctors to guide instruments precisely into a specific area of the body. Freehand navigation uses a pre-operative CT scan to plan instrument trajectory, and a camera system with instrument and patient arrays to track alignment, movement, and progression of the device relative to the pre-operative plan and patient anatomy. Active robotic-assisted guidance also uses a camera system with instrument and patient arrays, however in this workflow, the pre-operative CT scan is used to plan and define the path of a robotic arm which holds the instrument in the precise location and orientation relative to the patient’s anatomy.
Six (n=6) HCPs (fellowship-trained spine surgeons) took part in the study, each implanting a total of 24 pedicle screws percutaneously in a porcine thoracolumbar specimen, with eight screws placed in four vertebral bodies for three workflows: fluoroscopic guidance alone (baseline measurements), freehand navigation and active robotic-assisted guidance (impact measurements). The order of the workflows for each surgeon was prescribed according to a non-randomized block design.
Prior to the data collection, an off-the shelf fNIRS device, Artinis Brite Lite (Artinis Medical Systems BV) cap with pre-made holes to place the fNIRS sensors in desired locations, was placed on the participant and any hair moved away from the skin-electrode interface to allow for clear data recording. Participants were then asked to perform the assigned workflows.
At the end of each workflow completion, the Surg-TLX and SUS were conducted. Surg-TLX was completed with a series of anchored, visual analogue scale (VAS) questions, to record participant perception of workload. Scores were calculated for each of the six dimensions, as well as overall score. Raw ratings for each dimension were based on a 0-100 scale, from low demand to high demand. Weighted ratings accounted for a pairwise comparison of each dimension, whereas raw/Unweighted scores ranged from 0-100. Anything <9 was classified as low demand, 10-29 a medium demand, 30-49 a somewhat high demand, 50-79 a high demand, 80-100 a very high demand.
The SUS metrics measured included ratings from the PSSUQ (Post-Study System Usability Questionnaire) [Lewis,2002] for Overall, System Usefulness, Information Quality, Interface Quality, and ratings from SUPR-Q (Standardized User Experience Percentile Rank Questionnaire) [Sauro,2015] for Usability, Credibility, Appearance, and Loyalty. This data also enables calculation of a systems Customer Satisfaction (CSAT), Net Promoter Score (NPS) and a Perceived Value Score (PVS). Perceived value defined as the perceived benefit to the user relative to the perceived demand of the user (i.e., time, energy, effort). All metrics were compared to current standard practice.
Analysis of the user performance data (fNIRS) was performed to quantify metrics based on concentration changes of oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) to draw conclusion on prefrontal cortex activity and therefore cognitive load. Four metrics were calculated to understand the changes in concentration: (i) peak oxygenation (OXY), defined as the sum of five maximum OXY peaks (OXY=HbO-HbR), (ii) peak total hemoglobin (HbT), defined as the sum of five maximum HbT peaks (HbT=HbO+HbR, (iii) duration-dependent oxygenation, integrated oxygenation throughout each screw placement, (iv) average oxygenation per second, integrated oxygenation ÷ duration in seconds.
Results from this study showed that robotic-assisted guidance and freehand navigation workflows significantly reduce cognitive load required for percutaneous pedicle screw placement compared to fluoroscopic guidance alone. Specifically, there was significant reduction in peak oxygenation (p<0.001), duration-dependant oxygenation (p<0.002), average oxygenation per second (p<0.01), and peak total haemoglobin (p<0.001) with active robotic-assisted guidance compared to operating with fluoroscopic guidance alone. In addition, user perception results showed robotic-assisted guidance and freehand navigation workflows reduced overall workload and lessened mental, physical, and temporal demands of the surgeon. Specifically, there was significant reduction in overall demand (p<0.001), mental demand (p<0.01), physical demand (p<0.04), temporal demand (p<0.03), and situational stress (p<0.03). Both workflows were also seen to provide more benefit to the user while requiring less time and effort, Perceived Value Score of 71 (-100 to 100 scale). Results from this study suggest that using robotic assistance for percutaneous pedicle screw placement, may reduce cognitive load for spinal surgeons, leading to an improved surgeon functional work experience, and consequently a reduced risks of errors and enhanced patient safety.
Previous research has shown that cognitive load is associated with increased blood flow and increased oxyhaemoglobin (HbO) in the blood of the prefrontal cortex [Li,2025]. Recently, functional near infrared spectroscopy (fNIRS) has been utilised to quantify changes in blood flow patterns. Function NIRS sensors are non-invasive; they employ near-infrared light propagating diffusely through the scalp and brain, for functional monitoring and imaging of human brain hemodynamics. Changes in hemodynamic have been shown to be linked to cognitive function and effort, specifically fNIRS time series will increase if levels of oxygenated blood increase and decrease if levels of oxygenated blood decreases. Therefore, by analysing the level of oxyhaemoglobin (HbO) and deoxyhemoglobin (HbR), conclusions can be drawn about how much a task is affecting the cognitive process of the HCP. Changes in blood flow and oxyhemoglobin concentration (∆C) both indicate more activity in the prefrontal cortex. As robotic-assisted workflows may involve interaction with a digital interface, it is crucial to understand if the changed interaction between HCP and surgical field is producing alteration in HbO and what the consequent effect is.
Feasibility of fNIRS data collection during robotic-assisted workflows is yet to be addressed and is limited in the wider surgical field. Quantitative user performance measurements recorded with fNIRS, while performing surgical workflows providing different user-medical device interactions, pertain directly to the overall cognitive workload of the system and action being executed. This data can be augmented with user perception measurements captured from standardized surveys executed after each action is completed. Survey results can be used to quantify user perception and assess the usability and overall user experience of the system used to execute the actions. System Usability Survey (SUS) data can be used to assess user perception of usability, satisfaction, benefit, effort and value. Questions in the SUS have undergone psychometric qualification, and assessed for reliability, validity, and sensitivity [Sauro,2012]. Surgery Task Load Index (Surg-TLX) data can be used to assess workload in six dimensions: mental, physical, and temporal demands, as well as task complexity, situational stress, and distractions. The Surg-TLX is a validated measure of surgeon workload [McNair,2023; Wilson,2011] which has been used previously to assess cardiac surgery [Kennedy-Metz,2020].
In the present study, use of fNIRS sensors were implemented to investigate the feasibility of collecting data in a simulated surgical environment, to quantify cognitive load and mental fatigue, by means of quantifying changes in hemodynamic. Comparisons were made between the mean ∆C for active robotic-assisted guidance versus fluoroscopic guidance alone, and the mean ∆C for freehand navigation versus fluoroscopic guidance alone. Fluoroscopic guidance uses continuous X-ray imaging to provide real-time, moving X-ray images during medical procedures, allowing doctors to guide instruments precisely into a specific area of the body. Freehand navigation uses a pre-operative CT scan to plan instrument trajectory, and a camera system with instrument and patient arrays to track alignment, movement, and progression of the device relative to the pre-operative plan and patient anatomy. Active robotic-assisted guidance also uses a camera system with instrument and patient arrays, however in this workflow, the pre-operative CT scan is used to plan and define the path of a robotic arm which holds the instrument in the precise location and orientation relative to the patient’s anatomy.
Six (n=6) HCPs (fellowship-trained spine surgeons) took part in the study, each implanting a total of 24 pedicle screws percutaneously in a porcine thoracolumbar specimen, with eight screws placed in four vertebral bodies for three workflows: fluoroscopic guidance alone (baseline measurements), freehand navigation and active robotic-assisted guidance (impact measurements). The order of the workflows for each surgeon was prescribed according to a non-randomized block design.
Prior to the data collection, an off-the shelf fNIRS device, Artinis Brite Lite (Artinis Medical Systems BV) cap with pre-made holes to place the fNIRS sensors in desired locations, was placed on the participant and any hair moved away from the skin-electrode interface to allow for clear data recording. Participants were then asked to perform the assigned workflows.
At the end of each workflow completion, the Surg-TLX and SUS were conducted. Surg-TLX was completed with a series of anchored, visual analogue scale (VAS) questions, to record participant perception of workload. Scores were calculated for each of the six dimensions, as well as overall score. Raw ratings for each dimension were based on a 0-100 scale, from low demand to high demand. Weighted ratings accounted for a pairwise comparison of each dimension, whereas raw/Unweighted scores ranged from 0-100. Anything <9 was classified as low demand, 10-29 a medium demand, 30-49 a somewhat high demand, 50-79 a high demand, 80-100 a very high demand.
The SUS metrics measured included ratings from the PSSUQ (Post-Study System Usability Questionnaire) [Lewis,2002] for Overall, System Usefulness, Information Quality, Interface Quality, and ratings from SUPR-Q (Standardized User Experience Percentile Rank Questionnaire) [Sauro,2015] for Usability, Credibility, Appearance, and Loyalty. This data also enables calculation of a systems Customer Satisfaction (CSAT), Net Promoter Score (NPS) and a Perceived Value Score (PVS). Perceived value defined as the perceived benefit to the user relative to the perceived demand of the user (i.e., time, energy, effort). All metrics were compared to current standard practice.
Analysis of the user performance data (fNIRS) was performed to quantify metrics based on concentration changes of oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) to draw conclusion on prefrontal cortex activity and therefore cognitive load. Four metrics were calculated to understand the changes in concentration: (i) peak oxygenation (OXY), defined as the sum of five maximum OXY peaks (OXY=HbO-HbR), (ii) peak total hemoglobin (HbT), defined as the sum of five maximum HbT peaks (HbT=HbO+HbR, (iii) duration-dependent oxygenation, integrated oxygenation throughout each screw placement, (iv) average oxygenation per second, integrated oxygenation ÷ duration in seconds.
Results from this study showed that robotic-assisted guidance and freehand navigation workflows significantly reduce cognitive load required for percutaneous pedicle screw placement compared to fluoroscopic guidance alone. Specifically, there was significant reduction in peak oxygenation (p<0.001), duration-dependant oxygenation (p<0.002), average oxygenation per second (p<0.01), and peak total haemoglobin (p<0.001) with active robotic-assisted guidance compared to operating with fluoroscopic guidance alone. In addition, user perception results showed robotic-assisted guidance and freehand navigation workflows reduced overall workload and lessened mental, physical, and temporal demands of the surgeon. Specifically, there was significant reduction in overall demand (p<0.001), mental demand (p<0.01), physical demand (p<0.04), temporal demand (p<0.03), and situational stress (p<0.03). Both workflows were also seen to provide more benefit to the user while requiring less time and effort, Perceived Value Score of 71 (-100 to 100 scale). Results from this study suggest that using robotic assistance for percutaneous pedicle screw placement, may reduce cognitive load for spinal surgeons, leading to an improved surgeon functional work experience, and consequently a reduced risks of errors and enhanced patient safety.
Event Type
Poster Presentation
TimeTuesday, March 244:45pm - 6:15pm EDT
LocationRhinelander Gallery
Medical and Drug Delivery Devices