Complexity Analysis of Resting-State MEG Activity in Early-Stage Parkinson’s Disease Patients

Budget-Based Classification of Parkinson's Disease From Resting State EEG

Early detection is vital for future neuroprotective treatments of Parkinson's disease (PD). Resting state electroencephalographic (EEG) recording has shown potential as a cost-effective means to aid in detection of neurological disorders such as PD. In this study, we investigated how the number and placement of electrodes affects classifying PD patients and healthy controls using machine learning based on EEG sample entropy. We used a custom budget-based search algorithm for selecting optimized sets of channels for classification, and iterated over variable channel budgets to investigate changes in classification performance. Our data consisted of 60-channel EEG collected at three different recording sites, each of which included observations collected both eyes open (total N = 178) and eyes closed (total N = 131). Our results with the data recorded eyes open demonstrated reasonable classification performance (ACC = .76; AUC = .76) with only 5 channels placed far away from each other, the selected regions including right-frontal, left-temporal and midline-occipital sites. Comparison to randomly selected subsets of channels indicated improved classifier performance only with relatively small channel-budgets. The results with the data recorded eyes closed demonstrated consistently worse classification performance (when compared to eyes open data), and classifier performance improved more steadily as a function of number of channels. In summary, our results suggest that a small subset of electrodes of an EEG recording can suffice for detecting PD with a classification performance on par with a full set of electrodes. Furthermore our results demonstrate that separately collected EEG data sets can be used for pooled machine learning based PD detection with reasonable classification performance.

Decreased aperiodic neural activity in Parkinson's disease and dementia with Lewy bodies

Neural oscillations and signal complexity have been widely studied in neurodegenerative diseases, whereas aperiodic activity has not been explored yet in those disorders. Here, we assessed whether the study of aperiodic activity brings new insights relating to disease as compared to the conventional spectral and complexity analyses. Eyes-closed resting-state electroencephalography (EEG) was recorded in 21 patients with dementia with Lewy bodies (DLB), 28 patients with Parkinson's disease (PD), 27 patients with mild cognitive impairment (MCI) and 22 age-matched healthy controls. Spectral power was differentiated into its oscillatory and aperiodic components using the Irregularly Resampled Auto-Spectral Analysis. Signal complexity was explored using the Lempel-Ziv algorithm (LZC). We found that DLB patients showed steeper slopes of the aperiodic power component with large effect sizes compared to the controls and MCI and with a moderate effect size compared to PD. PD patients showed steeper slopes with a moderate effect size compared to controls and MCI. Oscillatory power and LZC differentiated only between DLB and other study groups and were not sensitive enough to detect differences between PD, MCI, and controls. In conclusion, both DLB and PD are characterized by alterations in aperiodic dynamics, which are more sensitive in detecting disease-related neural changes than the traditional spectral and complexity analyses. Our findings suggest that steeper aperiodic slopes may serve as a marker of network dysfunction in DLB and PD features.

A convolutional-recurrent neural network approach to resting-state EEG classification in Parkinson's disease

BACKGROUND

Parkinson's disease (PD) is expected to become more common, particularly with an aging population. Diagnosis and monitoring of the disease typically rely on the laborious examination of physical symptoms by medical experts, which is necessarily limited and may not detect the prodromal stages of the disease.

NEW METHOD

We propose a lightweight (~20 K parameters) deep learning model to classify resting-state EEG recorded from people with PD and healthy controls (HC). The proposed CRNN model consists of convolutional neural networks (CNN) and a recurrent neural network (RNN) with gated recurrent units (GRUs). The 1D CNN layers are designed to extract spatiotemporal features across EEG channels, which are subsequently supplied to the GRUs to discover temporal features pertinent to the classification.

RESULTS

The CRNN model achieved 99.2% accuracy, 98.9% precision, and 99.4% recall in classifying PD from HC. Interrogating the model, we further demonstrate that the model is sensitive to dopaminergic medication effects and predominantly uses phase information in the EEG signals.

COMPARISON WITH EXISTING METHODS

The CRNN model achieves superior performance compared to baseline machine learning methods and other recently proposed deep learning model.

CONCLUSION

The approach proposed in this study adequately extracts spatial and temporal features in multi-channel EEG signals that enable accurate differentiation between PD and HC. The CRNN model has excellent potential for use as an oscillatory biomarker for assisting in the diagnosis and monitoring of people with PD. Future studies to further improve and validate the model's performance in clinical practice are warranted.

Using Lempel-Ziv complexity as effective classification tool of the sleep-related breathing disorders

BACKGROUND AND OBJECTIVE

People suffer from sleep disorders caused by work-related stress, irregular lifestyle or mental health problems. Therefore, development of effective tools to diagnose sleep disorders is important. Recently, to analyze biomedical signals Information Theory is exploited. We propose efficient classification method of sleep anomalies by applying entropy estimating algorithms to encoded ECGs signals coming from patients suffering from Sleep-Related Breathing Disorders (SRBD).

METHODS

First, ECGs were discretized using the encoding method which captures the biosignals variability. It takes into account oscillations of ECG measurements around signals averages. Next, to estimate entropy of encoded signals Lempel-Ziv complexity algorithm (LZ) which measures patterns generation rate was applied. Then, optimal encoding parameters, which allow distinguishing normal versus abnormal events during sleep with high sensitivity and specificity were determined numerically. Simultaneously, subjects' states were identified using acoustic signal of breathing recorded in the same period during sleep.

RESULTS

Random sequences show normalized LZ close to 1 while for more regular sequences it is closer to 0. Our calculations show that SRBDs have normalized LZ around 0.32 (on average), while control group has complexity around 0.85. The results obtained to public database are similar, i.e. LZ for SRBDs around 0.48 and for control group 0.7. These show that signals within the control group are more random whereas for the SRBD group ECGs are more deterministic. This finding remained valid for both signals acquired during the whole duration of experiment, and when shorter time intervals were considered. Proposed classifier provided sleep disorders diagnostics with a sensitivity of 93.75 and specificity of 73.00%. To validate our method we have considered also different variants as a training and as testing sets. In all cases, the optimal encoding parameter, sensitivity and specificity values were similar to our results above.

CONCLUSIONS

Our pilot study suggests that LZ based algorithm could be used as a clinical tool to classify sleep disorders since the LZ complexities for SRBD positives versus healthy individuals show a significant difference. Moreover, normalized LZ complexity changes are related to the snoring level. This study also indicates that LZ technique is able to detect sleep abnormalities in early disorders stage.

Quantitative Susceptibility Mapping and Resting State Network Analyses in Parkinsonian Phenotypes-A Systematic Review of the Literature

An imbalance of iron metabolism with consecutive aggregation of α-synuclein and axonal degeneration of neurons has been postulated as the main pathological feature in the development of Parkinson’s disease (PD). Quantitative susceptibility mapping (QSM) is a new imaging technique, which enables to measure structural changes caused by defective iron deposition in parkinsonian brains. Due to its novelty, its potential as a new imaging technique remains elusive for disease-specific characterization of motor and non-motor symptoms (characterizing the individual parkinsonian phenotype). Functional network changes associated with these symptoms are however frequently described for both magnetoencephalography (MEG) and resting state functional magnetic imaging (rs-fMRI). Here, we performed a systematic review of the current literature about QSM imaging, MEG and rs-fMRI in order to collect existing data about structural and functional changes caused by motor and non-motor symptoms in PD. Whereas all three techniques provide an effect in the motor domain, the understanding of network changes caused by non-motor symptoms is much more lacking for MEG and rs-fMRI, and does not yet really exist for QSM imaging. In order to better understand the influence of pathological iron distribution onto the functional outcome, whole-brain QSM analyses should be integrated in functional analyses (especially for the non-motor domain), to enable a proper pathophysiological interpretation of MEG and rs-fMRI network changes in PD. Herewith, a better understanding of the relationship between neuropathological changes, functional network changes and clinical phenotype might become possible.

A systematic review of MEG-based studies in Parkinson's disease: The motor system and beyond

Parkinson's disease (PD) is accompanied by functional changes throughout the brain, including changes in the electromagnetic activity recorded with magnetoencephalography (MEG). An integrated overview of these changes, its relationship with clinical symptoms, and the influence of treatment is currently missing. Therefore, we systematically reviewed the MEG studies that have examined oscillatory activity and functional connectivity in the PD‐affected brain. The available articles could be separated into motor network‐focused and whole‐brain focused studies. Motor network studies revealed PD‐related changes in beta band (13–30 Hz) neurophysiological activity within and between several of its components, although it remains elusive to what extent these changes underlie clinical motor symptoms. In whole‐brain studies PD‐related oscillatory slowing and decrease in functional connectivity correlated with cognitive decline and less strongly with other markers of disease progression. Both approaches offer a different perspective on PD‐specific disease mechanisms and could therefore complement each other. Combining the merits of both approaches will improve the setup and interpretation of future studies, which is essential for a better understanding of the disease process itself and the pathophysiological mechanisms underlying specific PD symptoms, as well as for the potential to use MEG in clinical care.

Algorithmic Complexity of EEG for Prognosis of Neurodegeneration in Idiopathic Rapid Eye Movement Behavior Disorder (RBD)

Idiopathic rapid eye movement sleep behavior disorder (RBD) is a serious risk factor for neurodegenerative processes such as Parkinson's disease (PD). We investigate the use of EEG algorithmic complexity derived metrics for its prognosis. We analyzed resting state EEG data collected from 114 idiopathic RBD patients and 83 healthy controls in a longitudinal study forming a cohort in which several RBD patients developed PD or dementia with Lewy bodies. Multichannel data from ~ 3 min recordings was converted to spectrograms and their algorithmic complexity estimated using Lempel-Ziv-Welch compression. Complexity measures and entropy rate displayed statistically significant differences between groups. Results are compared to those using the ratio of slow to fast frequency power, which they are seen to complement by displaying increased sensitivity even when using a few EEG channels. Poor prognosis in RBD appears to be associated with decreased complexity of EEG spectrograms stemming in part from frequency power imbalances and cross-frequency amplitude algorithmic coupling. Algorithmic complexity metrics provide a robust, powerful and complementary way to quantify the dynamics of EEG signals in RBD with links to emerging theories of brain function stemming from algorithmic information theory.

Comparison of oscillatory activity in subthalamic nucleus in Parkinson's disease and dystonia

OBJECTIVES

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) has been successfully used to treat both Parkinson's disease (PD) and dystonia. Local field potentials (LFPs) recorded from the STN of PD patients demonstrate prominent beta frequency band activity. It is unclear whether such activity occurs in the STN in dystonia, and, if not, whether dystonia has another distinctive neural population activity in the STN.

METHODS

Twelve patients with PD, and eight patients with dystonia underwent DBS electrode implantation targeting the STN. Seven dystonia patients were off medication and one was on aripiprazole and clonazepam. LFPs were recorded from the DBS electrodes in PD in the on/off medication states and in dystonia. Power spectra and temporal dynamics measured by the with Lempel-Ziv complexity of the LFPs were compared among these states.

RESULTS

Normalised power spectra and Lempel-Ziv complexity of subthalamic LFPs differed between dystonia off and PD on/off, and between PD off and on over the low frequency, beta and high gamma bands. Patients with dystonia and off medication had lower beta power but higher low frequency and high gamma power than PD. Spectral power in the low beta frequency (11-20Hz) range was attenuated in medicated PD.

CONCLUSION

The results suggest that dystonia and PD are characterized by different patterns of oscillatory activities even within the same nucleus, and exaggerated beta activity may relate to hypo-dopaminergic status.

Complexity of resting-state EEG activity in the patients with early-stage Parkinson's disease

To investigate the abnormal brain activities in the early stage of Parkinson's disease (PD), the electroencephalogram (EEG) signals were recorded with 20 channels from non-dementia PD patients (18 patients, 8 females) and age matched healthy controls (18 subjects, 8 females) during the resting state. Two methods based on the ordinal patterns of the recorded series, i.e., permutation entropy (PE) and order index (OI), were introduced to characterize the complexity of the cortical activities for two groups. It was observed that the resting-state EEG of PD patients showed lower PE and higher OI than healthy controls, which indicated that the early-stage PD caused the reduced complexity of EEG. We further applied two methods to determine the complexity of EEG rhythms in five sub-bands. The results showed that the gamma, beta and alpha rhythms of PD patients were characterized by lower PE and higher OI, i.e., reduced complexity, than healthy subjects. No significant differences were observed in theta or delta rhythms between two groups. The findings suggested that PE and OI were promising methods to detect the abnormal changes in the dynamics of EEG signals associated with early-stage PD. Further, such changes in EEG complexity may be the early markers of the cortical or subcortical dysfunction caused by PD.

Spectral changes in spontaneous MEG activity across the lifespan

OBJECTIVE

The aim of this study is to explore the spectral patterns of spontaneous magnetoencephalography (MEG) activity across the lifespan.

APPROACH

Relative power (RP) in six frequency bands (delta, theta, alpha, beta-1, beta-2 and gamma) was calculated in a sample of 220 healthy subjects with ages ranging from 7 to 84 years.

MAIN RESULTS

A significant RP decrease in low-frequency bands (i.e. delta and theta) and a significant increase in high bands (mainly beta-1 and beta-2) were found from childhood to adolescence. This trend was observed until the sixth decade of life, though only slight changes were found. Additionally, healthy aging was characterized by a power increase in low-frequency bands. Our results show that spectral changes across the lifespan may follow a quadratic relationship in delta, theta, alpha, beta-2 and gamma bands with peak ages being reached around the fifth or sixth decade of life.

SIGNIFICANCE

Our findings provide original insights into the definition of the 'normal' behavior of age-related MEG spectral patterns. Furthermore, our study can be useful for the forthcoming MEG research focused on the description of the abnormalities of different brain diseases in comparison to cognitive decline in normal aging.

The ontogeny of the human connectome: development and dynamic changes of brain connectivity across the life span

The human brain comprises distributed cortical regions that are structurally and functionally connected into a network that is known as the human connectome. Elaborate developmental processes starting in utero herald connectome genesis, with dynamic changes in its architecture continuing throughout life. Connectome changes during development, maturation, and aging may be governed by a set of biological rules or algorithms, forming and shaping the macroscopic architecture of the brain's wiring network. To explore the presence of developmental patterns indicative of such rules, this review considers insights from studies on the cellular and the systems level into macroscopic connectome genesis and dynamics across the life span. We observe that in parallel with synaptogenesis, macroscopic connectome formation and transformation is characterized by an initial overgrowth and subsequent elimination of cortico-cortical axonal projections. Furthermore, dynamic changes in connectome organization throughout the life span are suggested to follow an inverted U-shaped pattern, with an increasingly integrated topology during development, a plateau lasting for the majority of adulthood and an increasingly localized topology in late life. Elucidating developmental patterns in brain connectivity is crucial for our understanding of the human connectome and how it may give rise to brain function, including the occurrence of brain network disorders across the life span.

Caffeine-Induced Global Reductions in Resting-State BOLD Connectivity Reflect Widespread Decreases in MEG Connectivity

In resting-state functional magnetic resonance imaging (fMRI), the temporal correlation between spontaneous fluctuations of the blood oxygenation level dependent (BOLD) signal from different brain regions is used to assess functional connectivity. However, because the BOLD signal is an indirect measure of neuronal activity, its complex hemodynamic nature can complicate the interpretation of differences in connectivity that are observed across conditions or subjects. For example, prior studies have shown that caffeine leads to widespread reductions in BOLD connectivity but were not able to determine if neural or vascular factors were primarily responsible for the observed decrease. In this study, we used source-localized magnetoencephalography (MEG) in conjunction with fMRI to further examine the origins of the caffeine-induced changes in BOLD connectivity. We observed widespread and significant (p < 0.01) reductions in both MEG and fMRI connectivity measures, suggesting that decreases in the connectivity of resting-state neuro-electric power fluctuations were primarily responsible for the observed BOLD connectivity changes. The MEG connectivity decreases were most pronounced in the beta band. By demonstrating the similarity in MEG and fMRI based connectivity changes, these results provide evidence for the neural basis of resting-state fMRI networks and further support the potential of MEG as a tool to characterize resting-state connectivity.

Investigation of EEG abnormalities in the early stage of Parkinson's disease

The objective of the present study was to investigate brain activity abnormalities in the early stage of Parkinson's disease (PD). To achieve this goal, eyes-closed resting state electroencephalography (EEG) signals were recorded from 15 early-stage PD patients and 15 age-matched healthy controls. The AR Burg method and the wavelet packet entropy (WPE) method were used to characterize EEG signals in different frequency bands between the groups, respectively. In the case of the AR Burg method, an increase of relative powers in the δ- and θ-band, and a decrease of relative powers in the α- and β-band were observed for patients compared with controls. For the WPE method, EEG signals from patients showed significant higher entropy over the global frequency domain. Furthermore, WPE in the γ-band of patients was higher than that of controls, while WPE in the δ-, θ-, α- and β-band were all lower. All of these changes in EEG dynamics may represent early signs of cortical dysfunction, which have potential use as biomarkers of PD in the early stage. Our findings may be further used for early intervention and early diagnosis of PD.