A Pocket Laboratory for Functional Neuroimaging Research Using Mobile Visual Oddball, Multimodal
Peter Rokowski1, Meltem Izzetoglu1, Luis Gomero1
1Department of Electrical and Computer Engineering, Villanova University, 800 E. Lancaster Ave, Villanova, PA, 19085, United States, 1 610-519-4500.
Background:
The need to observe brain activity in more natural environments, that is, outside of laboratory settings, is critical for understanding cognition. Wearable low-cost neuroimaging modalities (electroencephalography [EEG] and functional near-infrared spectroscopy [fNIRS]) are portable, noninvasive, and robust to motion artifacts but lack similarly portable tools for use in ecologically valid studies. Smartphones are ubiquitous, programmable, wireless, and thus strong candidates for "pocket laboratories" companion platforms that travel with study subjects. Therefore, we developed the Wearable Cognitive Assessment and Augmentation Toolkit (WearCAAT), a cross-platform neuroimaging task platform that integrates external sensors via the lab streaming layer (LSL) and supports over 100 sensor types. We validated our implementation with healthy human participants under multimodal neuroimaging conditions prior to analysis of data collection in ongoing clinical settings.
Objective:
This study aimed to validate WearCAAT, as a platform for functional neuroimaging research, via analysis of human participant data collected during our ongoing National Institutes of Health-funded study.
Methods:
We analyzed data from healthy college-aged (ages 18-30 y) adult participants, who completed a battery of shortened neurocognitive tasks (each lasting 4 min) in WearCAAT, while outfitted with research-grade multimodal EEG and fNIRS sensors. We indicated validity via the presence of task-related behavioral responses and their neuroimaging correlates. As a representative example, we analyzed the visual oddball task due to its well-documented poststimulus features for EEG and fNIRS. We extracted behavioral responses, mean response accuracies, and response times for infrequent (target) and frequent (standard) stimuli classes. We examined, poststimulus, P300, positive amplitude deflection around 300 (ms) in EEG and increased average oxygenated hemoglobin (HbO) levels in fNIRS.
Results:
We enrolled a total of 57 (male individuals: n=27, 47%; female individuals: n=30, 53%; mean age 22, SD 3.4 y) participants for data collection. We excluded the first 4 (7%) participants from our analysis due to technical errors. Our analysis revealed increased mean response times for infrequent (target) stimuli (mean 718, SD 148 ms) compared to frequent (standard) stimuli (mean 542, SD 122 ms) with the Wilcoxon test (Z=6.33; P<.001; r=0.87); higher P300 amplitudes over midline regions (parietal and temporal) for EEG; and increased oxygenated hemoglobin over the prefrontal cortex for fNIRS. All participants completed the full battery and reported no usability concerns or app crashes. Similarly, we observed no data loss or corruption that would negatively impact analyses.
Conclusions:
WearCAAT-provided outcomes from our study, which analyzed multimodal neuroimaging data collected during a mobile app-based visual oddball task, matched expectations from the literature. While full validation is ongoing for other tasks, we demonstrated initial validity of our app for neurocognitive imaging use. Our app and approach represent the first attempt at dedicated neuroimaging mobile-pocket laboratory and contribute to greater studies in ecological validity.
