Machine Learning-Based Classification to Disentangle EEG Responses to TMS and Auditory Input

Multivariate empirical mode decomposition reveals markers of Alzheimer's Disease in the oscillatory response to transcranial magnetic stimulation

OBJECTIVE

To investigate EEG activity following transcranial magnetic stimulation (TMS) of the dorsolateral prefrontal cortex of Alzheimer's Disease (AD) patients and control subjects using a data-driven characterization of brain oscillatory activity without prescribed frequency bands.

METHODS

We employed multivariate empirical mode decomposition (MEMD) to analyze the TMS-EEG response of 38 AD patients and 21 control subjects. We used the distinct features of EEG oscillatory modes to train a classification algorithm, a support vector machine.

RESULTS

AD patients exhibited a weakened slow-frequency response. Faster oscillatory modes displayed a biphasic response pattern in controls, characterized by an early increase followed by a widespread suppression, which was reduced in AD patients. Classification achieved robust discrimination performance (85%/23% true/false positive rate).

CONCLUSIONS

AD causes an impairment in the oscillatory response to TMS that has distinct features in different frequency ranges. These features uncovered by MEMD could serve as an effective EEG diagnostic marker.

SIGNIFICANCE

Early detection of AD requires diagnostic tools that are both effective and accessible. Combining EEG with TMS shows great promise. Our results and method enhance TMS-EEG both as a practical diagnostic tool, and as a way to further our understanding of AD pathophysiology.

Association of TMS-EEG interhemispheric imbalance with upper limb motor impairment in chronic stroke patients: An exploratory study

OBJECTIVE

We aimed to investigate the involvement of interhemispheric cortical dynamics as measured by combined transcranial magnetic stimulation and electroencephalography (TMS-EEG) in recovery of upper limb (UL) motor functions in chronic stroke patients.

METHODS

Ten patients with a history of single ischemic chronic stroke were enrolled (4F, 63.8 ± 9.9 years). Each patient underwent TMS-EEG recordings to evaluate interhemispheric cortical dynamics as well as a reaching task recorded with inertial measurement units, and a series of clinical assessments. TMS-EEG neurophysiological data were analysed considering spatiotemporal, power response, and interhemispheric balance (IHB) dynamics.

RESULTS

We found that IHB index (IHBi) and low-frequency power (LFP) (4-13 Hz) in the affected hemisphere were associated with the degree of UL impairment.

CONCLUSION

Increased IHBi due to stroke is an unfavourable factor of UL' functions. Similarly, LFP of both hemispheres is strongly correlated with clinical and kinematic outcomes.

SIGNIFICANCE

TMS-EEG biomarkers of interhemispheric unbalance could be used to estimate functional recovery and drive tailored neuromodulation and neurorehabilitation approaches.

Causal evidence for increased theta and gamma phase consistency in a parieto-frontal network during the maintenance of visual attention

Endogenous visuo-spatial attention is under the control of a fronto-parietal network of brain regions. One key node in this network, the intra-parietal sulcus (IPS), plays a crucial role in maintaining endogenous attention, but little is known about its ongoing physiology and network dynamics during different attentional states. Here, we investigated the reactivity of the left IPS in response to brain stimulation under different states of selective attention. We recorded electroencephalography (EEG) in response to single pulses of transcranial magnetic stimulation (TMS) of the IPS, while participants (N = 44) viewed bilateral random-dot motion displays. Individual MRI-guided TMS pulses targeted the left IPS, while the left primary somatosensory cortex (S1) served as an active control site. In separate blocks of trials, participants were cued to attend covertly to the motion display in one hemifield (left or right) and to report brief coherent motion targets. The perceptual load of the task was manipulated by varying the degree of motion coherence of the targets. Excitability, variability and information content of the neural responses to TMS were assessed by analysing TMS-evoked potential (TEP) amplitude and inter-trial phase clustering (ITPC), and by performing multivariate decoding of attentional state. Results revealed that a left posterior region displayed reduced variability in the phase of theta and gamma oscillations following TMS of the IPS, but not of S1, when attention was directed contralaterally, rather than ipsilaterally to the stimulation site. A right frontal cluster also displayed reduced theta variability and increased amplitude of TEPs when attention was directed contralaterally rather than ipsilaterally, after TMS of the IPS but not S1. Reliable decoding of attentional state was achieved after TMS pulses of both S1 and IPS. Taken together, our findings suggest that endogenous control of visuo-spatial attention leads to changes in the intrinsic oscillatory properties of the IPS and its associated fronto-parietal network.

Breakdown of TMS evoked EEG signal propagation within the default mode network in Alzheimer's disease

BACKGROUND

The neural activity of the Default Mode Network (DMN) is disrupted in patients with In Alzheimer's disease (AD).

OBJECTIVES

We used a novel multimodal approach to track neural signal propagation within the DMN in AD patients.

METHODS

Twenty mild to moderate AD patients were recruited. We used transcranial magnetic stimulation (TMS) pulses to probe with a millisecond time resolution the propagation of evoked electroencephalography (EEG) signal following the neural activation of the Precuneus (PC), which is a key hub area of the DMN. Moreover, functional and structural magnetic resonance imaging (MRI) data were collected to reconstruct individual features of the DMN.

RESULTS

In AD patients a probe TMS pulse applied over the PC evokes an increased local activity unmasking underlying hyperexcitability. In contrast, the EEG evoked neural signal did not propagate efficiently within the DMN showing a remarkable breakdown of signal propagation. fMRI and structural tractography showed that impaired signal propagation was related to the same connectivity matrices derived from DMN BOLD signal and transferred by specific white matter bundles forming the cingulum. These features were not detectable stimulating other areas (left dorsolateral prefrontal cortex) or for different networks (fronto-parietal network). Finally, connectivity breakdown was associated with cognitive impairment, as measured with the Clinical Dementia Rating Scale sum of boxes (CDR-SB).

CONCLUSIONS

TMS-EEG in AD shows both local hyperexcitability and a lack of signal propagation within the DMN. These neurophysiological features also correlate with structural and cognitive attributes of the patients.

SIGNIFICANCE

Neuronavigated TMS-EEG may be used as a novel neurophysiological biomarker of DMN connectivity in AD patients.

Real-time TMS-EEG for brain state-controlled research and precision treatment: a narrative review and guide

Transcranial magnetic stimulation (TMS) modulates neuronal activity, but the efficacy of an open-loop approach is limited due to the brain state's dynamic nature. Real-time integration with electroencephalography (EEG) increases experimental reliability and offers personalized neuromodulation therapy by using immediate brain states as biomarkers. Here, we review brain state-controlled TMS-EEG studies since the first publication several years ago. A summary of experiments on the sensorimotor mu rhythm (8-13 Hz) shows increased cortical excitability due to TMS pulse at the trough and decreased excitability at the peak of the oscillation. Pre-TMS pulse mu power also affects excitability. Further, there is emerging evidence that the oscillation phase in theta and beta frequency bands modulates neural excitability. Here, we provide a guide for real-time TMS-EEG application and discuss experimental and technical considerations. We consider the effects of hardware choice, signal quality, spatial and temporal filtering, and neural characteristics of the targeted brain oscillation. Finally, we speculate on how closed-loop TMS-EEG potentially could improve the treatment of neurological and mental disorders such as depression, Alzheimer's, Parkinson's, schizophrenia, and stroke.

Reduced TMS-evoked EEG oscillatory activity in cortical motor regions in patients with post-COVID fatigue

OBJECTIVE

Persistent fatigue is a major symptom of the so-called 'long-COVID syndrome', but the pathophysiological processes that cause it remain unclear. We hypothesized that fatigue after COVID-19 would be associated with altered cortical activity in premotor and motor regions.

METHODS

We used transcranial magnetic stimulation combined with EEG (TMS-EEG) to explore the neural oscillatory activity of the left primary motor area (l-M1) and supplementary motor area (SMA) in a group of sixteen post-COVID patients complaining of lingering fatigue as compared to a sample of age-matched healthy controls. Perceived fatigue was assessed with the Fatigue Severity Scale (FSS) and Fatigue Rating Scale (FRS).

RESULTS

Post-COVID patients showed a remarkable reduction of beta frequency in both areas. Correlation analysis exploring linear relation between neurophysiological and clinical measures revealed a significant inverse correlation between the individual level of beta oscillations evoked by TMS of SMA with the individual scores in the FRS (r(15) = -0.596; p = 0.012).

CONCLUSIONS

Post-COVID fatigue is associated with a reduction of TMS-evoked beta oscillatory activity in SMA.

SIGNIFICANCE

TMS-EEG could be used to identify early alterations of cortical oscillatory activity that could be related to the COVID impact in central fatigue.

The Past, Current and Future Research in Cerebellar TMS Evoked Responses-A Narrative Review

Transcranial magnetic stimulation coupled with electroencephalography (TMS-EEG) is a novel technique to investigate cortical physiology in health and disease. The cerebellum has recently gained attention as a possible new hotspot in the field of TMS-EEG, with several reports published recently. However, EEG responses obtained by cerebellar stimulation vary considerably across the literature, possibly due to different experimental methods. Compared to conventional TMS-EEG, which involves stimulation of the cortex, cerebellar TMS-EEG presents some technical difficulties, including strong muscle twitches in the neck area and a loud TMS click when double-cone coils are used, resulting in contamination of responses by electromyographic activity and sensory potentials. Understanding technical difficulties and limitations is essential for the development of cerebellar TMS-EEG research. In this review, we summarize findings of cerebellar TMS-EEG studies, highlighting limitations in experimental design and potential issues that can result in discrepancies between experimental outcomes. Lastly, we propose a possible direction for academic and clinical research with cerebellar TMS-EEG.

Electroencephalography oscillations can predict the cortical response following theta burst stimulation

BACKGROUND

Continuous theta burst stimulation and intermittent theta burst stimulation are clinically popular models of repetitive transcranial magnetic stimulation. However, they are limited by high variability between individuals in cortical excitability changes following stimulation. Although electroencephalography oscillations have been reported to modulate the cortical response to transcranial magnetic stimulation, their association remains unclear. This study aims to explore whether machine learning models based on EEG oscillation features can predict the cortical response to transcranial magnetic stimulation.

METHOD

Twenty-three young, healthy adults attended two randomly assigned sessions for continuous and intermittent theta burst stimulation. In each session, ten minutes of resting-state electroencephalography were recorded before delivering brain stimulation. Participants were classified as responders or non-responders based on changes in resting motor thresholds. Support vector machines and multi-layer perceptrons were used to establish predictive models of individual responses to transcranial magnetic stimulation.

RESULT

Among the evaluated algorithms, support vector machines achieved the best performance in discriminating responders from non-responders for intermittent theta burst stimulation (accuracy: 91.30%) and continuous theta burst stimulation (accuracy: 95.66%). The global clustering coefficient and global characteristic path length in the beta band had the greatest impact on model output.

CONCLUSION

These findings suggest that EEG features can serve as markers of cortical response to transcranial magnetic stimulation. They offer insights into the association between neural oscillations and variability in individuals' responses to transcranial magnetic stimulation, aiding in the optimization of individualized protocols.

Using TMS-EEG to assess the effects of neuromodulation techniques: a narrative review

Over the past decades, among all the non-invasive brain stimulation (NIBS) techniques, those aiming for neuromodulatory protocols have gained special attention. The traditional neurophysiological outcome to estimate the neuromodulatory effect is the motor evoked potential (MEP), the impact of NIBS techniques is commonly estimated as the change in MEP amplitude. This approach has several limitations: first, the use of MEP limits the evaluation of stimulation to the motor cortex excluding all the other brain areas. Second, MEP is an indirect measure of brain activity and is influenced by several factors. To overcome these limitations several studies have used new outcomes to measure brain changes after neuromodulation techniques with the concurrent use of transcranial magnetic stimulation (TMS) and electroencephalogram (EEG). In the present review, we examine studies that use TMS-EEG before and after a single session of neuromodulatory TMS. Then, we focused our literature research on the description of the different metrics derived from TMS-EEG to measure the effect of neuromodulation.