Reliable evaluation of functional connectivity and graph theory measures in source-level EEG: How many electrodes are enough?

OBJECTIVE: Using EEG to characterise functional brain networks through graph theory has gained significant interest in clinical and basic research. However, the minimal requirements for reliable measures remain largely unaddressed. Here, we examined functional connectivity estimates and graph theory metrics obtained from EEG with varying electrode densities. METHODS: EEG was recorded with 128 electrodes in 33 participants. The high-density EEG data were subsequently subsampled into three sparser montages (64, 32, and 19 electrodes). Four inverse solutions, four measures of functional connectivity, and five graph theory metrics were tested. RESULTS: The correlation between the results obtained with 128-electrode and the subsampled montages decreased as a function of the number of electrodes. As a result of decreased electrode density, the network metrics became skewed: mean network strength and clustering coefficient were overestimated, while characteristic path length was underestimated. CONCLUSIONS: Several graph theory metrics were altered when electrode density was reduced. Our results suggest that, for optimal balance between resource demand and result precision, a minimum of 64 electrodes should be utilised when graph theory metrics are used to characterise functional brain networks in source-reconstructed EEG data. SIGNIFICANCE: Characterisation of functional brain networks derived from low-density EEG warrants careful consideration.

Intelligent digital tools for screening of brain connectivity and dementia risk estimation in people affected by mild cognitive impairment: the AI-Mind clinical study protocol.

More than 10 million Europeans show signs of mild cognitive impairment (MCI), a transitional stage between normal brain aging and dementia stage memory disorder. The path MCI takes can be divergent; while some maintain stability or even revert to cognitive norms, alarmingly, up to half of the cases progress to dementia within 5 years. Current diagnostic practice lacks the necessary screening tools to identify those at risk of progression. The European patient experience often involves a long journey from the initial signs of MCI to the eventual diagnosis of dementia. The trajectory is far from ideal. Here, we introduce the AI-Mind project, a pioneering initiative with an innovative approach to early risk assessment through the implementation of advanced artificial intelligence (AI) on multimodal data. The cutting-edge AI-based tools developed in the project aim not only to accelerate the diagnostic process but also to deliver highly accurate predictions regarding an individual's risk of developing dementia when prevention and intervention may still be possible. AI-Mind is a European Research and Innovation Action (RIA H2020-SC1-BHC-06-2020, No. 964220) financed between 2021 and 2026. First, the AI-Mind Connector identifies dysfunctional brain networks based on high-density magneto- and electroencephalography (M/EEG) recordings. Second, the AI-Mind Predictor predicts dementia risk using data from the Connector, enriched with computerized cognitive tests, genetic and protein biomarkers, as well as sociodemographic and clinical variables. AI-Mind is integrated within a network of major European initiatives, including The Virtual Brain, The Virtual Epileptic Patient, and EBRAINS AISBL service for sensitive data, HealthDataCloud, where big patient data are generated for advancing digital and virtual twin technology development. AI-Mind's innovation lies not only in its early prediction of dementia risk, but it also enables a virtual laboratory scenario for hypothesis-driven personalized intervention research. This article introduces the background of the AI-Mind project and its clinical study protocol, setting the stage for future scientific contributions.

Introducing Region Based Pooling for handling a varied number of EEG channels for deep learning models.

Introduction: A challenge when applying an artificial intelligence (AI) deep learning (DL) approach to novel electroencephalography (EEG) data, is the DL architecture's lack of adaptability to changing numbers of EEG channels. That is, the number of channels cannot vary neither in the training data, nor upon deployment. Such highly specific hardware constraints put major limitations on the clinical usability and scalability of the DL models. Methods: In this work, we propose a technique for handling such varied numbers of EEG channels by splitting the EEG montages into distinct regions and merge the channels within the same region to a region representation. The solution is termed Region Based Pooling (RBP). The procedure of splitting the montage into regions is performed repeatedly with different region configurations, to minimize potential loss of information. As RBP maps a varied number of EEG channels to a fixed number of region representations, both current and future DL architectures may apply RBP with ease. To demonstrate and evaluate the adequacy of RBP to handle a varied number of EEG channels, sex classification based solely on EEG was used as a test example. The DL models were trained on 129 channels, and tested on 32, 65, and 129-channels versions of the data using the same channel positions scheme. The baselines for comparison were zero-filling the missing channels and applying spherical spline interpolation. The performances were estimated using 5-fold cross validation. Results: For the 32-channel system version, the mean AUC values across the folds were: RBP (93.34%), spherical spline interpolation (93.36%), and zero-filling (76.82%). Similarly, on the 65-channel system version, the performances were: RBP (93.66%), spherical spline interpolation (93.50%), and zero-filling (85.58%). Finally, the 129-channel system version produced the following results: RBP (94.68%), spherical spline interpolation (93.86%), and zero-filling (91.92%). Conclusion: In conclusion, RBP obtained similar results to spherical spline interpolation, and superior results to zero-filling. We encourage further research and development of DL models in the cross-dataset setting, including the use of methods such as RBP and spherical spline interpolation to handle a varied number of EEG channels.

Long term potentiation-like neural plasticity and performance-based memory function.

OBJECTIVE: Experience-dependent modulation of the visual evoked potential (VEP) has emerged as a promising non-invasive proxy for assaying long term potentiation (LTP)-like plasticity in the cerebral cortex. LTP is considered the principal candidate mechanism underlying learning and memory. There is, however, a paucity of evidence exploring associations between LTP-like plasticity and performance-based learning and memory. The present study aimed to explore the relationship between VEP-plasticity and higher-order learning and memory in healthy adults. METHOD: Visual and verbal learning and memory was assessed using the Aggie Figures Learning Test (AFLT) and the Rey Auditory Verbal Learning Test (RAVLT). The study included 111 healthy adults (61.1% females; mean age 37.6 years, range 17-71) who underwent a VEP paradigm employing visual high-frequency stimulation to induce a change in visual evoked responses recorded by scalp EEG. In addition, a more comprehensive neuropsychological assessment was administered. RESULTS: Several significant moderate age-corrected positive correlations were found between modulation of the later VEP components (N1 and P1-N1 peak-to-peak) and both visual and verbal learning and memory performance. Further, there were significant differences in learning and memory performance between participants showing a higher degree of modulation (>1 SD above mean) compared to participants showing a lower degree of modulation. No significant associations were found between VEP-plasticity and other neurocognitive domains. CONCLUSIONS: The current results suggest that LTP-like plasticity indexed by VEP modulation reflect processes specific to learning and memory. Future research is needed to further delineate the complex relationship between neural plasticity and learning and memory, specifically concerning possible clinical implications in populations with deficits in learning and memory function.

BIDS-structured resting-state electroencephalography (EEG) data extracted from an experimental paradigm.

Electroencephalography (EEG) offers a unique window into the dynamics of the neuronal symphony that powers our brains. Here, we describe a publicly available dataset of EEG recorded from 111 healthy subjects. The data were recorded with 64 electrodes in a resting-state condition, an approach that offers broad-spectred analysis options, including functional connectivity and graph theory. In a subset of the subjects (n = 42), a second EEG recording was performed, 2-3 months after the initial recording, allowing measurement stability to be assessed. Furthermore, in connection with the EEG acquisition, a range of neuropsychological test scores were obtained for each subject. The dataset is comprehensive and organised according to the Brain Imaging Data Structure (BIDS) specification, providing a valuable starting point for both aspiring and experienced researchers in a range of fields, including cognitive neuroscience, data science, machine learning, and clinical neurophysiology.

Long-Term Potentiation-Like Visual Synaptic Plasticity Is Negatively Associated With Self-Reported Symptoms of Depression and Stress in Healthy Adults.

Long-term potentiation (LTP) is one of the most extensively studied forms of neuroplasticity and is considered the strongest candidate mechanism for memory and learning. The use of event-related potentials and sensory stimulation paradigms has allowed for the translation from animal studies to non-invasive studies of LTP-like synaptic plasticity in humans. Accumulating evidence suggests that synaptic plasticity as measured by stimulus-specific response modulation is reduced in neuropsychiatric disorders such as major depressive disorder (MDD), bipolar disorders and schizophrenia, suggesting that impaired synaptic plasticity plays a part in the underlying pathophysiology of these disorders. This is in line with the neuroplasticity hypothesis of depression, which postulate that deficits in neuroplasticity might be a common pathway underlying depressive disorders. The current study aims to replicate and confirm earlier reports that visual stimulus-specific response modulation is a viable probe into LTP-like synaptic plasticity in a large sample of healthy adults (n = 111). Further, this study explores whether impairments in LTP-like synaptic plasticity is associated with self-reported subclinical depressive symptoms and stress in a healthy population. Consistent with prior research, the current study replicated and confirmed reports demonstrating significant modulation of visual evoked potentials (VEP) following visual high-frequency stimulation. Current results further indicate that reduced LTP-like synaptic plasticity is associated with higher levels of self-reported symptoms of depression and perceived stress. This indicate that LTP-like plasticity is sensitive to sub-clinical levels of psychological distress, and might represent a vulnerability marker for the development of depressive symptoms.

Sensory-Induced Human LTP-Like Synaptic Plasticity - Using Visual Evoked Potentials to Explore the Relation Between LTP-Like Synaptic Plasticity and Visual Perceptual Learning.

OBJECTIVE: Stimulus-selective response modulation (SRM) of sensory evoked potentials represents a well-established non-invasive index of long-term potentiation-like (LTP-like) synaptic plasticity in the human sensory cortices. Although our understanding of the mechanisms underlying stimulus-SRM has increased over the past two decades, it remains unclear how this form of LTP-like synaptic plasticity is related to other basic learning mechanisms, such as perceptual learning. The aim of the current study was twofold; firstly, we aimed to corroborate former stimulus-SRM studies, demonstrating modulation of visual evoked potential (VEP) components following high-frequency visual stimulation. Secondly, we aimed to investigate the association between the magnitudes of LTP-like plasticity and visual perceptual learning (VPL). METHODS: 42 healthy adults participated in the study. EEG data was recorded during a standard high-frequency stimulus-SRM paradigm. Amplitude values were measured from the peaks of visual components C1, P1, and N1. Embedded in the same experimental session, the VPL task required the participants to discriminate between a masked checkerboard pattern and a visual "noise" stimulus before, during and after the stimulus-SRM probes. RESULTS: We demonstrated significant amplitude modulations of VEPs components C1 and N1 from baseline to both post-stimulation probes. In the VPL task, we observed a significant change in the average threshold levels from the first to the second round. No significant association between the magnitudes of LTP-like plasticity and performance on the VPL task was evident. CONCLUSION: To the extent of our knowledge, this study is the first to examine the relationship between the visual stimulus-RM phenomenon and VPL in humans. In accordance with previous studies, we demonstrated robust amplitude modulations of the C1 and N1 components of the VEP waveform. However, we did not observe any significant correlations between modulation magnitude of VEP components and VPL task performance, suggesting that these phenomena rely on separate learning mechanisms implemented by different neural mechanisms.

The organization of functional neurocognitive networks in focal epilepsy correlates with domain-specific cognitive performance.

Understanding and diagnosing cognitive impairment in epilepsy remains a prominent challenge. New etiological models suggest that cognitive difficulties might not be directly linked to seizure activity, but are rather a manifestation of a broader brain pathology. Consequently, treating seizures is not sufficient to alleviate cognitive symptoms, highlighting the need for novel diagnostic tools. Here, we investigated whether the organization of three intrinsic, resting-state functional connectivity networks was correlated with domain-specific cognitive test performance. Using individualized EEG source reconstruction and graph theory, we examined the association between network small worldness and cognitive test performance in 23 patients with focal epilepsy and 17 healthy controls, who underwent a series of standardized pencil-and-paper and digital cognitive tests. We observed that the specific networks robustly correlated with test performance in distinct cognitive domains. Specifically, correlations were evident between the default mode network and memory in patients, the central-executive network and executive functioning in controls, and the salience network and social cognition in both groups. Interestingly, the correlations were evident in both groups, but in different domains, suggesting an alteration in these functional neurocognitive networks in focal epilepsy. The present findings highlight the potential clinical relevance of functional brain network dysfunction in cognitive impairment.

Evidence for Reduced Long-Term Potentiation-Like Visual Cortical Plasticity in Schizophrenia and Bipolar Disorder.

Several lines of research suggest that impairments in long-term potentiation (LTP)-like synaptic plasticity might be a key pathophysiological mechanism in schizophrenia (SZ) and bipolar disorder type I (BDI) and II (BDII). Using modulations of visually evoked potentials (VEP) of the electroencephalogram, impaired LTP-like visual cortical plasticity has been implicated in patients with BDII, while there has been conflicting evidence in SZ, a lack of research in BDI, and mixed results regarding associations with symptom severity, mood states, and medication. We measured the VEP of patients with SZ spectrum disorders (n = 31), BDI (n = 34), BDII (n = 33), and other BD spectrum disorders (n = 2), and age-matched healthy control (HC) participants (n = 200) before and after prolonged visual stimulation. Compared to HCs, modulation of VEP component N1b, but not C1 or P1, was impaired both in patients within the SZ spectrum (χ 2 = 35.1, P = 3.1 × 10-9) and BD spectrum (χ 2 = 7.0, P = 8.2 × 10-3), including BDI (χ 2 = 6.4, P = .012), but not BDII (χ 2 = 2.2, P = .14). N1b modulation was also more severely impaired in SZ spectrum than BD spectrum patients (χ 2 = 14.2, P = 1.7 × 10-4). N1b modulation was not significantly associated with Positive and Negative Syndrome Scale (PANSS) negative or positive symptoms scores, number of psychotic episodes, Montgomery and Åsberg Depression Rating Scale (MADRS) scores, or Young Mania Rating Scale (YMRS) scores after multiple comparison correction, although a nominal association was observed between N1b modulation and PANSS negative symptoms scores among SZ spectrum patients. These results suggest that LTP-like plasticity is impaired in SZ and BD. Adding to previous genetic, pharmacological, and electrophysiological evidence, these results implicate aberrant synaptic plasticity as a mechanism underlying SZ and BD.

Source-level EEG and graph theory reveal widespread functional network alterations in focal epilepsy.

OBJECTIVE: The hypersynchronous neuronal activity associated with epilepsy causes widespread functional network disruptions extending beyond the epileptogenic zone. This altered network topology is considered a mediator for non-seizure symptoms, such as cognitive impairment. The aim of this study was to investigate functional network alterations in focal epilepsy patients with good seizure control and high quality of life. METHODS: We compared twenty-two focal epilepsy patients and sixteen healthy controls on graph metrics derived from functional connectivity of source-level resting-state EEG. Graph metrics were calculated over a range of network densities in five frequency bands. RESULTS: We observed a significantly increased small world index in patients relative to controls. On the local level, two left-hemisphere regions displayed a shift towards greater alpha band "hubness". The findings were not mediated by age, sex or education, nor by age of epilepsy onset, duration or focus lateralisation. CONCLUSIONS: Widespread functional network alterations are evident in focal epilepsy, even in a cohort characterised by successful anti-seizure medication therapy and high quality of life. These findings might support the position that functional network analysis could hold clinical relevance for epilepsy. SIGNIFICANCE: Focal epilepsy is accompanied by global and local functional network aberrancies which might be implied in the sustenance of non-seizure symptoms.

Do visual and auditory stimulus-specific response modulation reflect different mechanisms of neocortical plasticity?

Stimulus response modulation (SRM) of sensory evoked potentials represents a promising method as a non-invasive index of long-term potentiation (LTP)-like synaptic plasticity in the human sensory cortices. As of today, however, no consensus exists regarding which experimental parameters elicit the most robust SRM response. The aim of the current study was twofold; firstly, we aimed to replicate former studies demonstrating visual SRM in healthy adults. Second, we integrated visual and auditory stimuli within the same SRM recording session to assay potential cross-modal associations. Such an association between modalities would strengthen the assumption that the SRM effect reflects common mechanisms underlying synaptic plasticity rather than reflecting modality-specific phenomena. A replication of previous findings showing robust potentiation of the visual evoked potential was evident, supporting the majority of previous work using similar paradigms, lending further support to the notion that high-frequent visual stimulation is a viable probe into LTP-like synaptic plasticity in the human visual cortex. The auditory evoked potentials (AEPs) did not, however, fully replicate previous work, though a significant increase of temporally later AEP components was found. In contrast to our hypothesis, there were no significant within-subject cross-modality correlations between the visual and auditory SRM. This lack of significant association might suggest that auditory and visual SRM depend on different mechanisms, and that further SRM studies on non-invasive LTP-like synaptic plasticity should focus on optimizing paradigms within the visual modality.

Experience-dependent modulation of the visual evoked potential: Testing effect sizes, retention over time, and associations with age in 415 healthy individuals.

Experience-dependent modulation of the visual evoked potential (VEP) is a promising proxy measure of synaptic plasticity in the cerebral cortex. However, existing studies are limited by small to moderate sample sizes as well as by considerable variability in how VEP modulation is quantified. In the present study, we used a large sample (n = 415) of healthy volunteers to compare different quantifications of VEP modulation with regards to effect sizes and retention of the modulation effect over time. We observed significant modulation for VEP components C1 (Cohen's d = 0.53), P1 (d = 0.66), N1 (d=-0.27), N1b (d=-0.66), but not P2 (d = 0.08), and in three clusters of total power modulation, 2-4 min after 2 Hz prolonged visual stimulation. For components N1 (d=-0.21) and N1b (d=-0.38), as well for the total power clusters, this effect was retained after 54-56 min, by which time also the P2 component had gained modulation (d = 0.54). Moderate to high correlations (0.39≤ρ≤0.69) between modulation at different postintervention blocks revealed a relatively high temporal stability in the modulation effect for each VEP component. However, different VEP components also showed markedly different temporal retention patterns. Finally, participant age correlated negatively with C1 (χ2=30.4), and positively with P1 modulation (χ2=13.4), whereas P2 modulation was larger for female participants (χ2=15.4). There were no effects of either age or sex on N1 and N1b potentiation. These results provide strong support for VEP modulation, and especially N1b modulation, as a robust measure of synaptic plasticity, but underscore the need to differentiate between components, and to control for demographic confounders.