| dc.description.abstract | The identification and validation of reliable brain biomarkers to characterize neuropsychiatric and neurodegenerative disorders can be laborious and generally relies on time-consuming brain imaging methods. Brain activity is typically assessed using costly and non-portable MRI or PET scan technology which do not lend themselves to repeated use. Electroencephalography (EEG) consists of recording voltage fluctuations at the surface of the scalp and are informative of brain functions and cognition. A growing number of portable EEG devices are now being developed to enable the repeated sampling of brain activity remotely (i.e., outside of the laboratory) by participants themselves. These cutting-edge technologies make use of new types of electrodes that can be effortlessly placed onto the scalp without any kind of preparation or need for conductive gel. The possibility of conducting repeated EEG studies outside of the laboratory hold many promises regarding the identification of new disease biomarkers. It is expected that repeated measurements will allow the isolation of metrics more robust to lifestyle factors and fluctuating disease symptoms. Before deploying these new systems in clinical populations, it is necessary to study the feasibility and validity of the proposed method. The objective of this thesis was to characterise and explore this new way of collecting brain data by looking at the deployment of a neurocognitive platform indicative of this kind of approach outside the laboratory. Specifically, my analyses were centred around the Cumulus Neuroscience platform (www.cumulusneuro.com). This platform includes a wireless dry EEG headset coupled with an Android tablet for cognitive task presentation. At the beginning of a study, participants are invited to the laboratory to get familiarised with the technology and are then invited to take the hardware home to record sessions autonomously.
Empirical Chapter 1 was aimed at evaluating the usability and validity of the approach by conducting post-hoc analyses of two at-home datasets collected in the context of other research. The first dataset included 50 older adults (>55-years-old) and the second 30 young male adults (mean age=25.6 years-old). I quantified the usability of the platform by looking at adherence to the study protocol, participants? subjective feedback and percentage of successful sessions. Further, data quality was evaluated using an established metric of quality within the field and results were compared to a representative study using the traditional gold-standard technology. I found that on average, adherence was very high across the two groups and usability analyses did not reveal any differences between the younger and older cohorts. However, data collected in the older group was substantially noisier compared to the younger dataset and many more trials were discarded during signal preprocessing. When comparing my results to the gold standard, I found that between 2 to 3 at-home sessions needed to be aggregated to reach similar noise levels when comparing similar cohorts.
Empirical Chapter 2 was a use-case study of the platform when deployed within the context of a pharmacological study. My test case was a ketamine challenge study during which thirty young healthy males received in a double-blind cross-over design two infusions of ketamine and of saline solutions four weeks apart. The protocol included intensive bursts of data collection in the laboratory immediately before and after the infusions under research supervision to capture acute drug effects. Additionally, data was collected remotely from the home a week before and after each intervention to measure any persistent drug effect. The primary objective of this study was to evaluate if the neurocognitive dry EEG platform studied here could emulate results from the wet EEG literature. The second objective of this study was exploratory and evaluated if at-home unsupervised self-administered dry EEG could capture any persistent drug effects. I targeted the best-characterised EEG endpoints studied within the context of ketamine: resting state power spectrum modulations as well as the mismatch negativity event-related potential (ERP). Furthermore, I also targeted electrophysiological markers that have been associated with the anterior cingulate and prefrontal cortices activities (i.e., the Event-Related Negativity ? ERN, and the P300). The data collected with the dry EEG platform only partially emulated the wet EEG literature. Although, I observed the well-described gamma enhancement during the ketamine infusion, no other analyses yielded significant results. Other important factors unrelated to the technology itself, such as the timing of the sessions and the drug dosage may have contributed to the lack of significant results. Data collected in the home suggested that there was no persistent effect of ketamine on electrophysiological markers of cognitive functions in the week after the infusions.
Empirical Chapter 3 consisted in of an in-depth evaluation of a gamified Oddball task and its suitability to elicit a widely used and well characterized ERP in the EEG field: the P300. Gamified cognitive tasks are becoming popular within psychology research. However, very few studies have looked at their validity with respect to the elicitation of ERPs. In this chapter, I explored the effect of task gamification and repeated task administration on the elicitation the P300. I recruited 10 young adults from the local community and asked them to complete a gamified 2-stimulus Oddball task multiple times both in the laboratory and the home. As I focused on gamification in isolation from the rest of the platform, EEG data were collected using a standard wet EEG system under researcher supervision in a controlled laboratory environment. Participants completed six consecutive EEG sessions in the laboratory, were sent home to complete 7 behaviour-only sessions on separate days and came back in the laboratory to complete 6 consecutive additional EEG sessions. I evaluated ERP morphologies, timings, and topographies. I also looked the stability of the signals over time and asked participants for qualitative feedback on their strategies. Participants? questionnaires revealed that individuals? strategies were quite heterogenous and that sometimes participants could change strategy over time. I observed well-defined ERPs, with consistent morphologies and timings across all participants in accordance with what is reported in the literature. Topographical results were more mixed. The target P300 yielded a topography consistent with the literature whereas the Non-Target P300s were more anterior than expected. These results suggest that gamification features impacted the nature of cognitive functions at play during the task.
Overall, this thesis evaluated key features of remote dry EEG, showing that it is easily deployable in participants? home environment. Participants found the platform quite usable and repeatedly engaged with it while in the home. Data quality matched that observed in wet EEG. The use-case pharmacological trial demonstrated that it is possible to successfully deploy the technology in the context of a drug study that included intensive burst of recordings occurring alternatively in the home or the laboratory. My results only partially emulated the wet EEG literature, but it is likely that other factors unrelated to the technology itself prevented a complete emulation. The in-depth gamification analyses revealed that a gamified EEG task could elicit consistent ERPs within participants, although across participants different strategies could have resulted in the generation of different types of P300. To summarise, I have shown that remote, self-administered, gamified EEG is a valid approach for the study of neurocognition. | en |
