CC BY
18
Нелинейная
динамика и нейронаука
Известия высших учебных заведений. Прикладная нелинейная динамика. 2021. Т. 29, № 3 Izvestiya Vysshikh Uchebnykh Zavedeniy. Applied Nonlinear Dynamics. 2021;29(3)
Article
DOI: 10.18500/0869-6632-2021-29-3-386-397
Recurrence quantification analysis provides the link between age-related decline in motor brain response and complexity of the baseline EEG
Abstract. The goal of the present study is to investigate the effect of healthy aging on the neuronal mechanisms supporting human brain activity during motor task performance. Such biomarkers of the age-related changes can be detected using mathematical methods of time-series analysis and complexity analysis. Methods. In the present paper, recurrence quantification analysis (RQA) measures are employed to explore the complexity of the pre-movement EEG in young and elderly adult groups. To evaluate the neural brain response during motor execution, we applied the traditional methods of time-frequency analysis. Results. The proposed approach demonstrated that (i) the RQA measures show a significant increase of complexity in elderly adults; (ii) increased pre-movement EEG complexity comes with the reduced motor-related brain response in the a/^-band (p < 0.01), estimated via the traditional methods of time-frequency analysis. It allows to conclude that the increased pre-movement EEG complexity indicates the weak neuronal plasticity degenerated under the factor of age. Conclusion. The complexity of the pre-movement a/^-band neuronal oscillations could be considered as a relevant measure for the detection of age-related cognitive or motor impairments. Besides, applied RQA method demonstrated a good ability to assess the complexity features of pre-stimulus EEG and provided a clear interpretation of age-related changes in electrical activity of the brain cortex.
Keywords: complexity, recurrence quantification analysis, EEG, age-related changes, event-related desynchronization.
Acknowledgements. This work has been supported by Russian Foundation for Basic Research (Grant 19-52-55001) and the Council on Grants of the President of the Russian Federation (Grant NSh-2594.2020.2). The author thanks Dr. N. Frolov and Prof. A. Hramov for useful discussions within the framework of this study.
For citation: Pitsik EN. Recurrence quantification analysis provides the link between age-related decline in motor brain response and complexity of the baseline EEG. Izvestiya VUZ. Applied Nonlinear Dynamics. 2021;29(3):386-397. DOI: 10.18500/0869-6632-2021-29-3-386-397
This is an open access article distributed under the terms of Creative Commons Attribution License (CC-BY 4.0).
E. N. Pitsik
Innopolis University, Russia E-mail: e.pitsik@innopolis.ru Received 30.10.2020, accepted 21.01.2021, published 31.05.2021
386
©ПщикЕ.Н, 2021
Научная статья УДК 51-7
DOI: 10.18500/0869-6632-2021-29-3-386-397
Рекуррентный анализ сложности предстимульных сигналов ЭЭГ и их связь с возрастными изменениями в двигательной активности мозга
Е. Н. Пицик
Университет Иннополис, Россия E-mail: e.pitsik@innopolis.ru Поступила в редакцию 30.10.2020, принята к публикации 21.01.2021, опубликована 31.05.2021
Аннотация. Целью данного исследования является изучение влияния процессов естественного старения на нейронные механизмы мозга, ответственные за обработку двигательной активности человека. Подобные биомаркеры возрастных изменений могут быть обнаружены с помощью математических методов анализа временных рядов и анализа сложности сигналов. Методы. В данном исследовании для анализа сложности предстимульных сигналов ЭЭГ в двух возрастных группах испытуемых были использованы меры рекуррентного анализа. Для оценки нейронной реакции во время выполнения движений, были применены традиционные методы частотно-временного анализа сигналов ЭЭГ. Результаты. Предложенный подход продемонстрировал следующее: 1) меры рекуррентного анализа показывают значительный рост сложности предстимульных сигналов ЭЭГ в группе возрастных испытуемых; 2) повышенная сложность ЭЭГ связана со сниженным нейронным ответом в а/ц-ритме (р < 0.01), измеренным с помощью частотно-временного анализа. Данные результаты позволяют сделать вывод о том, что повышенная сложности предстимульных сигналов ЭЭГ указывает на возрастное снижение нейронной пластичности. Заключение. Сложность предстимульных колебаний в а/ц-ритме может быть рассмотрена как релевантная мера для детектирования возрастных когнитивных и двигательных изменений. Кроме того, рекуррентный анализ продемонстрировал возможность оценить сложность предстимульных сигналов ЭЭГ и дать чёткую интерпретацию возрастным изменениям электрической активности коры головного мозга.
Ключевые слова: сложность, рекуррентный анализ, ЭЭГ, возрастные изменения, вызванная десинхронизация.
Благодарности. Данное исследование проведено при поддержке Российского Фонда Фундаментальных Исследований (грант 19-52-55001) и Совета по Грантам Президента Российской Федерации (грант НШ-2594.2020.2). Автор благодарит Н. С. Фролова и проф. А. Е. Храмова за плодотворные обсуждения в рамках данного исследования.
Для цитирования: Пицик Е. Н. Рекуррентный анализ сложности предстимульных сигналов ЭЭГ и их связь с возрастными изменениями в двигательной активности мозга//Известия вузов. ПНД. 2021. T. 29, № 3. С. 386-397. DOI: 10.18500/0869-6632-2021-29-3-386-397
Статья опубликована на условиях лицензии Creative Commons Attribution License (CC-BY 4.0).
Introduction
Studying the age-related changes in brain activity is a complex task of high social significance. Besides the commonly used clinical approaches to the diagnosis of neurological disorders, recent studies actively focus on the features of brain electrical activity extracted via the mathematical methods of time-series analysis [1]. Traditionally, hallmarks of the age-related pathological states are observed in the spectral power of electroencephalogram (EEG) rhythms. In particular, a well-known pattern of Alzheimer's disease is a generalized slowing of the EEG, i.e., a shift in the EEG spectral power towards the low-frequency 6- (4-8 Hz) and S-bands (1-4 Hz) [2,3].
Another group of studies focus on the dynamical properties of EEG reflecting the changes in brain behavior under healthy aging. In this context, the complexity features of EEG, which are known to reflect the current dynamical state of the EEG time series as well as to detect the transitions from periodic and chaotic behavior and vice versa [4,5], are of special interest. In particular, the entropy-based estimations of complexity provide an insight into various pathological processes, such as abnormalities in sleep-related EEG [6] and cardiac rhythm [7], epileptic brain activity [8], etc. Regarding the age-related changes, a decreased resting-state EEG complexity is a well-established marker of healthy aging [9,10].
In Ref. [11] the authors demonstrate that the response of EEG to the changes in environmental conditions weakens with age based on the analysis of dimensional complexity in the eyes-open and eyes-closed state. In Ref. [12] the authors show that the age-related slowing down in the motor initiation before the dominant hand movements is accompanied by the growth of the 6-activity within sensorimotor area and reconfiguration of the 6-band functional connectivity in elderly adults.
However, entropy-based methods have certain limitations. For instance, the fractal dimension methods, despite their simplicity and universality, are often criticized for sensitivity to the noisiness of the signal and narrow interval of the values, which can lead to identical results for different signals [13]. On the other hand, the entropy-based methods are known to be sensitive to the length of time series [14], which resulted in the development of several modifications depending on the research goal [15,16]. All these methods in their present form satisfy the most requirements of modern EEG studies. In our previous work [17], we used long-range correlations technique [18,19] to analyze age-related distinctions in EEG signals during execution of motor tasks. However, the application of new methods is also important to enhance the physiological interpretation of the brain signals complexity.
In the present study, the recurrence quantification analysis (RQA) is proposed to quantify the complexity of the baseline EEG preceding motor task [20]. RQA is a powerful tool that has been developed to explore the recurrent states of complex dynamical systems through the analysis of time series. The measures of RQA estimate the system's complexity using the natural property of dynamical systems to recur to their previous substantially similar states. RQA was widely used for analysis of climate data [21,22], in natural sciences and engineering [23-25], for natural language processing [26] and etc. One can witness a rapidly developing trend to apply RQA to the processing of biological signals. In particular, RQA measures demonstrated an advantage over the frequency domain analysis during the detection of pathological states on EEG, such as early stages of Autism Spectrum Disorder [27], multiple sclerosis [28] and stress-related conditions [29,30]. In addition, in our previous study, the RQA measures demonstrated an ability to quantify properly the features of motor-related brain activity [31]. It should be noted that RQA provides a clear interpretation of the system's complexity in terms of concepts of statistical physics and nonlinear dynamics.
Summarizing the above, the goal of the present paper is to demonstrate the relevance of the RQA measures in the analysis of age-related changes in brain behavior through the quantification of EEG complexity. We consider motor-related activity since recent studies indicate that the brain motor system is susceptible to the age-related degenerative changes [32,33]. The disintegration of the brain network connectivity leads to over-activation of the sensorimotor cortex, often addressed as a compensatory mechanism, which causes the different patterns of cognitive and motor-related activity processing in the elderly population [32,34,35]. Despite a large number of studies, the particular effects of aging on the motor-related response remain unclear. The main hypothesis of this study is that the complexity of pre-movement neural activity captured by EEG is connected with the ability of individuals to induce a proper motor-related cortical response, and healthy aging has a direct effect on this connection. To address this hypothesis, the relationships are investigated between the complexity of the pre-movement a/p (8-14 Hz) brain rhythm and motor-related cortical activation in young and elderly adult groups. The results demonstrate that the RQA-based complexity of the pre-movement EEG significantly increases with age. Finally, the correlation is uncovered between the pre-movement EEG complexity and the motor-related cortical activation. The significant differences dominate in the sensorimotor area.
1. Methods
1.1. Experimental design. The experiments were performed in two age groups consisting of 10 healthy elderly adults (65.0 ± 5.7 years, 6 females, 4 males) and a control group of 10 healthy young individuals (26.1 ± 5.2 years, 3 females, 7 males). Selected subjects had no medical history of head trauma, stroke, depression, and other neurological conditions that can affect the EEG features.
ml m
baseline i motor task pause
а
2 s LH 4-5 s
с
Pre-filtered EEG (1-100 Hz), mV
b
ERSP|j, a.u.
-Ад-
-0.5-2
6-8 s
С
4
time, sec
10
Filtered EEG (8-13 Hz), mV
(ЦПlMAlWp4^
0.8 1.2 time, sec
Fig. 1. a - the timeline of a single trial. Here, tc is the duration of the audio command: 0.55 s for the right hand and 0.75 s for the left. b - a raw EEG sample corresponding to the single trial of the right hand motor execution and c - wavelet energy for this trial averaged over ^-rhythm (8-13 Hz). Solid black and gray vertical lines correspond to the time interval of the audio command. d - sample of EEG baseline time series (upper plot) and corresponding RP (lower plot)
The experimental study was approved by the local ethics committee and performed in accordance with the Declaration of Helsinki. Each experimental session started with 5 minutes of eyes-open EEG recording. After that, the participants were instructed to perform fine motor tasks according to the audio instruction. Following either long or short audio signal (tc = 0.75 sec or tc = 0.55 sec for the left and right hand, respectively), the participants were asked to squeeze the corresponding hand into a fist and relax it after the second audio command (see Fig. 1, a). After the movement was completed, the 6-8 seconds long pause was given before the next motor task.
Each participant performed N = 60 trials (30 with each hand).
1.2. EEG recordings and preprocessing. For EEG acquisition, the EEG amplifier Encephalan-EEGR-19/26 (Medicom MTD, Russia) with a sampling rate 250 Hz was used. EEG dataset was recorded with 31 Ag/AgCl electrodes according to the "10-10" international system [36]. Acquired EEG signals were filtered using the 50 Hz Notch filter to avoid the power-line interference.
Next, the low-frequency artifacts were removed by applying the 5ife-order Butterworth filter in the range 1-100 Hz to the EEG dataset (see Fig. 1, b). The ocular and cardiac artifacts were also removed using the independent component analysis (ICA) [37]. Then, EEG recordings were sliced into epochs. Each epoch included 2s of the baseline EEG and 10s of the motor-related activity. Finally, a visual inspection of the EEG epochs was performed to exclude those containing the artifacts that could not be removed via described preprocessing steps, leaving 15 epochs for each hand.
The preprocessing steps, including filtering, ICA, and epoching, were performed using the MNE package for Python 3.7 (ver. 0.20.0) [38]. The experimental data underlying the results presented in the study are available online at [39].
1.3. Recurrence quantification analysis. The RQA was applied to the EEG signals filtered in the a/^-rhythm (8-14 Hz, see Fig. 1, d) to access the complexity of the brain activity preceding motor task. For proper estimation of the EEG complexity via RQA, one should consider the reconstructed phase space trajectory of the EEG time series using Taken's theorem [40]. The embedding parameters were chosen as follows: the time delay of EEG filtered in a/^-rhythm is t = 5 defined with the mutual information method [41], and the embedding dimension is m = 3 according to the Kennel's false nearest neighbors method [42].
In the reconstructed phase space trajectory Xk, the two states are considered as similar if they are in each other neighborhood defined by the threshold e [43]. Recurrence matrix can be constructed by evaluation of all neighboring states of the trajectory as follows:
Rij(e) = 0(e - ||xk(tt) - xk(tj)||), (1)
where i,j = 1, ...,T — 2t, 0 is the Heaviside function, and || ■ || is the norm (Euclidean norm in the present study). The recurrence plot (RP) is a visualization of the recurrence matrix that shows recurrent/non-recurrent states as black and white dots, respectively. Note that the value of the recurrence threshold e is crucial for the assessment of the correct structure of the RP. In the present study, the e was chosen as the 3rd percentile of the pairwise distance distribution according to Kraemer et al. [44]. RP consists of various structures formed by the recurrent "black" dots and non-recurrent "white" dots, that represent different aspects of the system's dynamics in the state space. For example, the "black" diagonal lines quantified by the measure of determinism (DET) mean that the part of the system's trajectory moves almost parallel to another part for a time equal to the line length [43]:
Em . ip(0
DET = 'mm-—. (2)
££1 ip (0
Therefore, DET represents the ratio of the recurrent dots that are included in the diagonal lines to the all recurrent dots of RP. In Eq. 2, ¿min is the minimal length of diagonal line (¿min = 20 in this study) and the P(I) is the histogram of diagonal lines of length I.
Another measure of complexity is a recurrence time entropy (RTE) based on the "white" vertical lines analysis, which uses Shannon's definition of entropy and reflects the transitions of the considered system from regular to chaotic state and vice versa [44]. RTE is defined as follows:
1 Tmax
RTE = — -— £ p(tw) In p(tw), (3)
in 1 max , _..
zw —1
where p(tw) is the probability of recurrence time tw, and Tmax is the largest recurrence time.
The RQA measures were applied to the 2 s time window of EEG before the first audio signal which corresponded to the pre-movement activity. DET and RTE were computed for each of 31 EEG channels and averaged over 30 epochs for each of 20 subjects (15 LH epochs and 15 RH epochs). Between-subject comparison of DET and RTE in spatial domain aimed at uncovering age-related changes in the brain signals complexity was performed using the non-parametric clustering permutation test following Ref. [45]. Effect was evaluated at each sensor-pair using two-tailed F-test for independent samples (DF1 = 1, DF2 = 18, p = 0.05, Fcritical = 4.41).
The RQA was performed using pyunicorn package for Python 3.7 (ver.0.6.0) [46].
1.4. Time-frequency analysis. To support the findings regarding EEG signals complexity, its link with the motor-related changes of EEG spectral power was explored in both age groups. A priory knowledge about the cortical activation during movements execution implies a pronounced event-related desynchronization of a/p-oscillations (ERDp) in the contralateral area of the primary motor cortex [47-51].
First, time-series of the a/p-band event-related spectral power (ERSPp) were evaluated in symmetrical C4 (in right hemisphere for LH) and C3 (in left hemisphere for RH) sensors using Morlet-based wavelet transform [52,53] implemented in the MNE package (see an example of ERSPp time series in Fig. 1, c). Baseline correction was applied by subtracting the mean of the 2 s baseline preceding audio command and then dividing by the mean of the baseline ("percent" mode). To investigate the effect of age, a between-subject comparison of the motor-related ERSPp averaged over the time interval
1-4 s was performed in each condition (LH and RH) via the unpaired t-test (DF = 9). The normality of the a/^-band spectral power samples was validated via the Shapiro-Wilk test. Then, the correlation between the complexity of the pre-movement EEG and corresponding values of the averaged a/^-band spectral power was analyzed using Pearson's coefficient and linear regression.
2. Results and Discussion
2.1. Pre-movement EEG complexity versus age. Fig. 2, a shows the results of the sensor-level between-subject comparison of the pre-movement EEG complexity. Both RQA measures demonstrated the presence of significant clusters, with 26 channels in DET (except F8, F7, and those placed over occipital cortex) and 13 channels for RTE (Tp8, Tp7, Pz, Oz, Fc-, C-, Cp-channels). These differences revealed via non-parametric permutation test can be interpreted as almost whole-head increased complexity of the pre-movement a/^-band oscillations in elderly adults (reduced DET and increased RTE). Specifically, the pronounced differences are observed in the bilateral motor cortex with a skew towards the left hemisphere.
Next, we selected RQA data from the channels included in significant clusters only, and obtained the mean values of RQA complexity measures DET and RTE for each subject:
DET = £ DET, (4)
Sdet sesDET
RTE = ^ RTES, (5)
1 Srte sesRTE
where Sdet and Srte are the sensor clusters of significant DET and RTE changes respectively,
DET
ERSR. left hand, C4
Î
EA YA
a
b
elderly adult
c
EA YA young adult ......baseline level
d
Fig. 2. a - Between-subject differences in DET (upper row) and RTE (lower row) calculated via a non-parametric cluster-based permutation test. The black dots correspond to the sensors composing significant clusters. b - Dependencies between the age and the RQA measures of complexity. Here, '***' indicates p = 0.001 and '****' indicates p = 0.0001 according to Mann-Whitney U-test. c - ERSPa/^ averaged across the subjects (mean ± SE) for the left-hand (top panel) and right-hand (bottom panel) movements. The shaded area corresponds to the time interval imotor, where the motor execution takes place. d - The distributions of ERD^ for both age groups in the case of left-hand (top panel) and right-hand (bottom panel) movement
Nsdet and Nsrte are the sizes of these clusters, DETS and RTES are measures of EEG signal complexity at the sth sensor.
Fig. 2, b shows the obtained dependencies. In particular, elderly group demonstrated decreased DET compared with young individuals with p = 0.0001 according to Mann-Whitney test. At the same time, RTE significantly increases with age (p = 0.001).
According to the demonstrated dependencies, the age-related changes in the a/p-band EEG are associated with the increased complexity of these signals. In previous research, the entropy characteristics were considered that demonstrated the age-related decrease of the complexity of the resting state brain activity of different modality [54-56]. The age-related changes in brain activity were usually associated with reduced complexity due to the destructive changes in the cortical network structure [57]. However, in the present paper, the periods of rest between sequential motor executions are considered, which significantly deviates from the resting-state EEG due to the presence of the traces of task specific activity. Note that in our previous work considering the motor-related activity in the young population, the reduced EEG complexity was associated with ERD pattern during motor execution [31]. Besides, the motor-related ERD/S was previously connected with lower/higher EEG complexity estimated via Kolmogorov entropy [48]. We can conclude that in the young adult group the increase of the EEG oscillation regularity in the task-specific area corresponds to the better responsivity and stabilization of the brain state associates with the motor execution. At the same time, in the elderly population the motor-related baseline EEG demonstrated increased complexity in a/p-rhythm also can be an illustration of poor neural response on stimuli or actions under healthy aging.
2.2. Motor-related spectral power and age. Fig. 2, c shows the ERSPp averaged across the subjects in each group and illustrates the motor-related desynchronization of the p-rhythm soon after the audio command. The gray area in Fig. 2, c highlights to the time interval imotor = [1,4]s corresponding to the motor task execution. To compare the level of the motor-related desynchronization between the age groups and experimental conditions, ERSPp was averaged within imotor time interval so that ERDp was obtained.
Fig. 2, d shows the distributions of the ERPSp for both age groups in the case of the left-hand (C4 sensor top panel) and the right-hand (C3 sensor bottom panel) movements. As expected, young adults demonstrated more pronounced desynchronization compared with elderly adults. The mixed-design ANOVA (see Table) yielded the age as the only significant factor that affects the motor-related neural response (F = 9.049, p = 0.008). The within-subject factor of motor task (right or left hand) did not have significant influence on the ERDp value (F = 0.255, p = 0.641), as well as the interaction between the factors (F = 1.471, p = 0.241). This observation confirms the known neurophysiological effects of healthy aging. Several studies on age-related changes evidenced that unilateral motor activation is less pronounced in elderly individuals. Instead, they employed a compensatory mechanism involving larger cortical areas including the regions of ipsilateral motor and prefrontal cortices [58-60]. The white matter tract disruption and volume decrease underlie such functional reorganization of brain regions [61,62].
Table. Time-averaged ERDp, two-way mixed ANOVA
Cases df1 df2 Mean Square F p
Age (between-subject factor) 1 18 0.308 9.049 0.008
Motor task (within-subject factor) 1 18 0.003 0.225 0.641
Motor task x Age 1 18 0.020 1.471 0.241
2.3. Relationship between the pre-movement complexity and the motor-related spectral power. Fig. 3 demonstrates the results of the correlation between the RQA complexity measures and ERDp. The correlation was evaluated at the same EEG sensors in the motor area (C4 for left-hand movements and C3 for right-hand movements).
ERDi
left hand, C4
0.0 -0.2 -0.4 -0.6
ERDi
right hand, C3
rAY=-0.74 p = 0.0002 Ft2 = 0.553
0.0-0.2-0.4-0.6-
rAY = -0.65 p = 0.0018 R2 = 0.426
a
ERD
И
0.0 -0.2 -0.4 -0.6
0.60
left hand, C4 rAY= 0.67 p = 0.0013 R2 = 0.447
0.65
0.70
0.75
0.60
0.65
0.70
0.75
DET
DET
^PQ right hand, C3 ^ rAY=0.51 0.0H P = 0.0204 R2 = 0.264
-0.2 -0.4 -0.6
4.1
4.2
4.3
4.4
4.5
4.1
4.2
4.3
4.4
4.5
4.6
Ь
RTE
• elderly adult x young adult
RTE
Fig. 3. Correlations between the averaged p-band spectral power ERDp and pre-movement EEG complexity measured by DET (a) and RTE (b). Left and right columns illustrate correlations in the case of the left-hand and right-hand movements, respectively. The black line represents the linear law fitted by the linear regression with a coefficient of determination R2. The shaded area highlights the 95% confidence interval. Pearson's correlation coefficient tay and its p-value are shown on the corresponding scatter plots
The results demonstrate that the considered variables have significant linear relationships. The Pearson's correlation show that DET is negatively correlated with ERDp (for the left hand movement, C4: rAY = —0.74, p = 0.0002, for the right hand movement, C3: rAY = —0.65, p = 0.0018), whereas RTE demonstrated positive trend (for the left hand movement, C4: tay = 0.67, p = 0.0013, for the right hand movement, C3: tay = 0.51, p = 0.0204). Despite the demonstrated regression was rather moderate, the Pearson's p-value indicated the significance of linear relationship between the considered variables. We assume that the computed R2 values are affected by the relatively small and individual deviations caused by non-age factors.
In summary, the results demonstrated the correlation between the pre-movement brain p-rhythm's complexity estimated via the RQA and the value of motor-related p-band desynchronization ERPSp. In particular, the higher regularity of the pre-movement EEG precedes the more pronounced suppression of the ERDp. Both measures were considered as the biomarkers of the well-established neural response in the brain motor cortex, which were mostly manifest in young adults. These results provide the link between the known effects of aging and verify the RQA-based complexity analysis as a proper approach to reveal the markers of the age-related changes in the human EEG recordings.
Conclusions
The Recurrence Quantification Analysis was applied to evaluate the pre-movement EEG complexity of the aging brain and to explore its relationship with a motor-related neural response. The RQA-based measures demonstrated a strong age-related increase in the baseline level of EEG complexity.
In particular, pre-movement EEG of the elderly adults demonstrated the low value DET and high value of RTE, suggesting that sequential execution of the fine motor tasks affect the structure of the baseline EEG in a different way in different age groups. Elderly adults also showed a weaker motor-related ц-band desynchronization compared with young individuals. The results of the two-way mixed ANOVA verified that age was the only significant factor affecting neural activation on the brain motor cortex. Finally, the motor-related neural response was found to be correlated with the complexity of the pre-movement EEG. This observation confirmed the prior hypothesis that the age-related abnormalities in the complexity of pre-movement EEG anticipate the quality of motor-related neural response. Besides, the results of the RQA provided a robust and clearly interpreted measurement of the EEG signal complexity.
References
1. Adeli H, Ghosh-Dastidar S. Automated EEG-Based Diagnosis of Neurological Disorders: Inventing the Future of Neurology. CRC Press; 2010. 423 p. DOI: 10.1201/9781439815328.
2. Musaeus C, Engedal K, H0gh P, Jelic V, Morup M, Naik M, Oeksengaard A, Snaedal J, Wahlund L, Waldemar G, Andersen B. EEG theta power is an early marker of cognitive decline in dementia due to Alzheimer's disease. Journal of Alzheimer's Disease. 2018;64(4):1359-1371. DOI: 10.3233/JAD-180300.
3. Smailovic U, Jelic V. Neurophysiological markers of Alzheimer's disease: Quantitative EEG approach. Neurology and Therapy. 2019;8(2):37-55. DOI: 10.1007/s40120-019-00169-0.
4. Liu J, He T, Zheng C, Huang Y. Measuring EEG complexity for studying the state of mental load. Journal of Biomedical Engineering. 1997;14(1):33-37.
5. Cao Z, Lin C. Inherent fuzzy entropy for the improvement of EEG complexity evaluation. IEEE Transactions on Fuzzy Systems. 2018;26(2):1032-1035. DOI: 10.1109/TFUZZ.2017.2666789.
6. Acharya U, Bhat S, Faust O, Adeli H, Chua E, Lim W, Koh J. Nonlinear dynamics measures for automated EEG-based sleep stage detection. European Neurology. 2015;74(5-6):268-287. DOI: 10.1159/000441975.
7. Billeci L, Varanini M. Characterizing electrocardiographic changes during pre-seizure periods through temporal and spectral features. In: 2017 Computing in Cardiology (CinC). Rennes, France: IEEE; 2017. P. 1-4. DOI: 10.22489/CinC.2017.098-282.
8. Gao X, Yan X, Gao P, Gao X, Zhang S. Automatic detection of epileptic seizure based on approximate entropy, recurrence quantification analysis and convolutional neural networks. Artificial Intelligence in Medicine. 2020;102:101711. DOI: 10.1016/j.artmed.2019.101711.
9. Smits F, Porcaro C, Cottone C, Cancelli A, Rossini P, Tecchio F. Electroencephalographic fractal dimension in healthy ageing and Alzheimer's disease. PLOS One. 2016;11(2):e0149587. DOI: 10.1371/journal.pone.0149587.
10. Ishii R, Canuet L, Aoki Y, Hata M, Iwase M, Ikeda S, Nishida K, Ikeda M. Healthy and pathological brain aging: From the perspective of oscillations, functional connectivity, and signal complexity. Neuropsychobiology. 2017;75(4):151-161. DOI: 10.1159/000486870.
11. Barry R, Clarke A, Johnstone S, Brown C. EEG differences in children between eyes-closed and eyes-open resting conditions. Clinical Neurophysiology. 2009;120(10):1806-1811. DOI: 10.1016/j.clinph.2009.08.006.
12. Frolov N, Pitsik E, Maksimenko V, Grubov V, Kiselev A, Wang Z, Hramov A. Age-related slowing down in the motor initiation in elderly adults. Plos One. 2020;15(9):e0233942. DOI: 10.1371/journal.pone.0233942.
13. Kesic S, Spasic S. Application of Higuchi's fractal dimension from basic to clinical neurophysiology: A review. Computer Methods and Programs in Biomedicine. 2016;133:55-70. DOI: 10.1016/j.cmpb.2016.05.014.
14. Richman J, Moorman J. Physiological time-series analysis using approximate entropy and sample entropy. American Journal of Physiology Heart and Circulatory Physiology. 2000;278(6):H2039-H2049. DOI: 10.1152/ajpheart.2000.278.6.H2039.
15. Mizuno T, Takahashi T, Cho R, Kikuchi M, Murata T, Takahashi K, Wada Y. Assessment of EEG
dynamical complexity in Alzheimer's disease using multiscale entropy. Clinical Neurophysiology. 2010;121(9):1438-1446. DOI: 10.1016/j.clinph.2010.03.025.
16. Simons S, Abasolo D, Escudero J. Classification of Alzheimer's disease from quadratic sample entropy of electroencephalogram. Healthcare Technology Letters. 2015;2(3):70-73. DOI: 10.1049/htl.2014.0106.
17. Pavlov A, Pitsik E, Frolov N, Badarin A, Pavlova O, Hramov A. Age-related distinctions in EEG signals during execution of motor tasks characterized in terms of long-range correlations. Sensors (Basel). 2020;20(20):5843. DOI: 10.3390/s20205843.
18. Peng C, Buldyrev S, Havlin S, Simons M, Stanley H, Goldberger A. Mosaic organization of DNA nucleotides. Physical Review E. 1994;49(2):1685-1689. DOI: 10.1103/PhysRevE.49.1685.
19. Frolov N, Grubov V, Maksimenko V, Luttjohann A, Makarov V, Pavlov A, Sitnikova E, Pisarchik A, Kurths J, Hramov A. Statistical properties and predictability of extreme epileptic events. Scientific Reports. 2019;9(1):7243. DOI: 10.1038/s41598-019-43619-3.
20. Webber Jr C, Marwan N. Recurrence Quantification Analysis: Theory and Best Practices. Understanding Complex Systems. Springer International Publishing Switzerland; 2015. DOI: 10.1007/978-3-319-07155-8.
21. Trauth M, Asrat A, Duesing W, Foerster V, Kraemer K, Marwan N, Maslin M, Schaebitz F. Classifying past climate change in the Chew Bahir basin, southern Ethiopia, using recurrence quantification analysis. Climate Dynamics. 2019;53(5-6):2557-2572. DOI: 10.1007/s00382-019-04641-3.
22. Panagoulia D, Vlahogianni E. Recurrence quantification analysis of extremes of maximum and minimum temperature patterns for different climate scenarios in the Mesochora catchment in Central-Western Greece. Atmospheric Research. 2018;205:33-47. DOI: 10.1016/j.atmosres.2018.02.004.
23. Hou Y, Aldrich C, Lepkova K, Kinsella B. Detection of under deposit corrosion in a CO2 environment by using electrochemical noise and recurrence quantification analysis. Electrochimica Acta. 2018;274:160-169. DOI: 10.1016/j.electacta.2018.04.037.
24. Stender M, Oberst S, Tiedemann M, Hoffmann N. Complex machine dynamics: systematic recurrence quantification analysis of disk brake vibration data. Nonlinear Dynamics. 2019;97(4):2483-2497. DOI: 10.1007/s11071-019-05143-x.
25. Mitra V, Sarma B, Sarma A, Janaki M, Sekar Iyengar A, Marwan N, Kurths J. Investigation of complexity dynamics in a DC glow discharge magnetized plasma using recurrence quantification analysis. Physics of Plasmas. 2016;23(6):062312. DOI: 10.1063/1.4953903.
26. Allen L, Likens A, McNamara D. Recurrence quantification analysis: A technique for the dynamical analysis of student writing. In: Proceedings of the 30th International Florida Artificial Intelligence Research Society Conference. Marco Island, United States: AAAI Press; 2017. P. 240-245.
27. Bosl W, Tager-Flusberg H, Nelson C. EEG analytics for early detection of autism spectrum disorder: A data-driven approach. Scientific Reports. 2018;8(1):6828. DOI: 10.1038/s41598-018-24318-x.
28. Carrubba S, Minagar A, Chesson Jr A, Frilot 2nd C, Marino A. Increased determinism in brain electrical activity occurs in association with multiple sclerosis. Neurological Research. 2012;34(3):286-290. DOI: 10.1179/1743132812Y.0000000010.
29. Dick O, Svyatogor I, Reznikova T, Fedoryaka D, Nozdrachev A. Analysis of EEG patterns in subjects with panic attacks. Human Physiology. 2020;46(2):163-174. DOI: 10.1134/S0362119720010065.
30. Fan M, Tootooni M, Sivasubramony R, Miskovic V, Rao P, Chou C. Acute stress detection using recurrence quantification analysis of electroencephalogram (EEG) signals. In: Brain Informatics and Health. BIH 2016. vol. 9919 of Lecture Notes in Computer Science. Omaha, NE, USA: Springer, Cham; 2016. p. 252-261. DOI: 10.1007/978-3-319-47103-7_25.
31. Pitsik E, Frolov N, Kraemer K, Grubov V, Maksimenko V, Kurths J, Hramov A. Motor execution reduces EEG signals complexity: Recurrence quantification analysis study. Chaos: An Interdisciplinary Journal of Nonlinear Science. 2020;30(2):023111. DOI: 10.1063/1.5136246.
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
39
Michely J, Volz L, Hoffstaedter F, Tittgemeyer M, Eickhoff S, Fink G, Grefkes C. Network connectivity of motor control in the ageing brain. NeuroImage: Clinical. 2018;18:443-455. DOI: 10.1016/j.nicl.2018.02.001.
Burianova H, Marstaller L, Rich A, Williams M, Savage G, Ryan M, Sowman P. Motor neuroplasticity: A MEG-fMRI study of motor imagery and execution in healthy ageing. Neuropsychologia. 2020;146:107539. DOI: 10.1016/j.neuropsychologia.2020.107539. Mitchell T, Starrs F, Soucy J, Thiel A, Paquette C. Impaired sensorimotor processing during complex gait precedes behavioral changes in middle-aged adults. The Journals of Gerontology: Series A. 2019;74(12):1861-1869. DOI: 10.1093/gerona/gly210.
Quandt F, Bonstrup M, Schulz R, Timmermann J, Zimerman M, Nolte G, Hummel F. Spectral variability in the aged brain during fine motor control. Frontiers in Aging Neuroscience. 2016;8:305. DOI: 10.3389/fnagi.2016.00305.
Nuwer M, Comi G, Emerson R, Fuglsang-Frederiksen A, Guerit J, Hinrichs H, Ikeda A, Luccas F, Rappelsburger P. IFCN standards for digital recording of clinical EEG. Electroencephalography and Clinical Neurophysiology. 1998;106(3):259-261. DOI: 10.1016/s0013-4694(97)00106-5. Hyvarinen A, Oja E. Independent component analysis: algorithms and applications. Neural Networks. 2000;13(4-5):411-430. DOI: 10.1016/S0893-6080(00)00026-5. Gramfort A, Luessi M, Larson E, Engemann D, Strohmeier D, Brodbeck C, Goj R, Jas M, Brooks T, Parkkonen L, Hamalainen M. MEG and EEG data analysis with MNE-Python. Frontiers in Neuroscience. 2013;7:267. DOI: 10.3389/fnins.2013.00267.
Frolov N, Pitsik E, Grubov V, Kiselev A, Maksimenko V, Hramov A. EEG dataset for the analysis of age-related changes in motor-related cortical activity during a series of fine motor tasks performance [Electronic resource]; 2020. Available from: https://figshare.com/articles/EEG_dataset_for_the_analysis_of_age-related_ changes_in_motor-related_cortical_activity_during_a_series_of_fine_motor_ tasks_performance/12301181/1. DOI: 10.6084/m9.figshare.12301181. Takens F. Detecting strange attractors in turbulence. In: Dynamical Systems and Turbulence, Warwick 1980. vol. 898 of Lecture Notes in Mathematics. Springer, Berlin, Heidelberg; 1981. p. 366-381. DOI: 10.1007/BFb0091924.
Roulston M. Estimating the errors on measured entropy and mutual information. Physica D: Nonlinear Phenomena. 1999;125(3-4):285-294. DOI: 10.1016/S0167-2789(98)00269-3. Kennel M, Brown R, Abarbanel H. Determining embedding dimension for phase-space reconstruction using a geometrical construction. Physical Review A. 1992;45(6):3403-3411. DOI: 10.1103/PhysRevA.45.3403.
Marwan N, Carmen Romano M, Thiel M, Kurths J. Recurrence plots for the analysis of complex systems. Physics Reports. 2007;438(5-6):237-329. DOI: 10.1016/j.physrep.2006.11.001. Kraemer K, Donner R, Heitzig J, Marwan N. Recurrence threshold selection for obtaining robust recurrence characteristics in different embedding dimensions. Chaos: An Interdisciplinary Journal of Nonlinear Science. 2018;28(8):085720. DOI: 10.1063/1.5024914.
Maris E, Oostenveld R. Nonparametric statistical testing of EEG-and MEG-data. Journal of Neuroscience Methods. 2007;164(1):177-190. DOI: 10.1016/j.jneumeth.2007.03.024. Donges J, Heitzig J, Beronov B, Wiedermann M, Runge J, Feng Q, Tupikina L, Stolbova V, Donner R, Marwan N, Dijkstra H, Kurths J. Unified functional network and nonlinear time series analysis for complex systems science: The pyunicorn package. Chaos: An Interdisciplinary Journal of Nonlinear Science. 2015;25(11):113101. DOI: 10.1063/1.4934554.
Pfurtscheller G, Brunner C, Schlogl A, Lopes Da Silva F. Mu rhythm (de)synchronization and EEG single-trial classification of different motor imagery tasks. NeuroImage. 2006;31(1):153-159. DOI: 10.1016/j.neuroimage.2005.12.003.
Gao L, Wang J, Chen L. Event-related desynchronization and synchronization quantification in motor-related EEG by Kolmogorov entropy. Journal of Neural Engineering. 2013;10(3):036023. DOI: 10.1088/1741-2560/10/3/036023.
Maksimenko V, Pavlov A, Runnova A, Nedaivozov V, Grubov V, Koronovskii A, Pchelintseva S,
Pitsik E, Pisarchik A, Hramov A. Nonlinear analysis of brain activity, associated with motor action and motor imaginary in untrained subjects. Nonlinear Dynamics. 2018;91(4):2803—2817. DOI: 10.1007/s11071-018-4047-y.
50. Maksimenko V, Kurkin S, Pitsik E, Musatov V, Runnova A, Efremova T, Hramov A, Pisarchik A. Artificial neural network classification of motor-related EEG: An increase in classification accuracy by reducing signal complexity. Complexity. 2018;2018:1-10. DOI: 10.1155/2018/9385947.
51. Chholak P, Niso G, Maksimenko V, Kurkin S, Frolov N, Pitsik E, Hramov A, Pisarchik A. Visual and kinesthetic modes affect motor imagery classification in untrained subjects. Scientific Reports. 2019;9(1):9838. DOI: 10.1038/s41598-019-46310-9.
52. Hramov A, Koronovskii A, Makarov V, Pavlov A, Sitnikova E. Wavelets in Neuroscience. Springer Series in Synergetics. Springer-Verlag Berlin Heidelberg; 2015. DOI: 10.1007/978-3-662-43850-3.
53. Pavlov A, Hramov A, Koronovskii A, Sitnikova E, Makarov V, Ovchinnikov A. Wavelet analysis in neurodynamics. Physics Uspekhi. 2012;55(9):845-875. DOI: 10.3367/UFNe.0182.201209a.0905.
54. Sun J, Wang B, Niu Y, Tan Y, Fan C, Zhang N, Xue J, Wei J, Xiang J. Complexity analysis of EEG, MEG, and fMRI in mild cognitive impairment and Alzheimer's disease: A review. Entropy (Basel). 2020;22(2):239. DOI: 10.3390/e22020239.
55. Jia Y, Gu H, Luo Q. Sample entropy reveals an age-related reduction in the complexity of dynamic brain. Scientific Reports. 2017;7(1):7990. DOI: 10.1038/s41598-017-08565-y.
56. Li X, Zhu Z, Zhao W, Sun Y, Wen D, Xie Y, Liu X, Niu H, Han Y. Decreased resting-state brain signal complexity in patients with mild cognitive impairment and Alzheimer's disease: a multi-scale entropy analysis. Biomedical Optics Express. 2018;9(4):1916-1929. DOI: 10.1364/BOE.9.001916.
57. Ruiz-Gomez S, Gomez C, Poza J, Martínez-Zarzuela M, Tola-Arribas M, Cano M, Hornero R. Measuring alterations of spontaneous EEG neural coupling in Alzheimer's disease and mild cognitive impairment by means of cross-entropy metrics. Frontiers in Neuroinformatics. 2018;12:76. DOI: 10.3389/fninf.2018.00076.
58. Mattay V, Fera F, Tessitore A, Hariri A, Das S, Callicott J, Weinberger D. Neurophysiological correlates of age-related changes in human motor function. Neurology. 2002;58(4):630-635. DOI: 10.1212/wnl.58.4.630.
59. Sailer A, Dichgans J, Gerloff C. The influence of normal aging on the cortical processing of a simple motor task. Neurology. 2000;55(7):979-985. DOI: 10.1212/wnl.55.7.979.
60. Lariviere S, Xifra-Porxas A, Kassinopoulos M, Niso G, Baillet S, Mitsis G, Boudrias M. Functional and effective reorganization of the aging brain during unimanual and bimanual hand movements. Human Brain Mapping. 2019;40(10):3027-3040. DOI: 10.1002/hbm.24578.
61. Guttmann C, Jolesz F, Kikinis R, Killiany R, Moss M, Sandor T, Albert M. White matter changes with normal aging. Neurology. 1998;50(4):972-978. DOI: 10.1212/WNL.50.4.972.
62. Webb C, Rodrigue K, Hoagey D, Foster C, Kennedy K. Contributions of white matter connectivity and BOLD modulation to cognitive aging: A lifespan structure-function association study. Cerebral Cortex. 2020;30(3):1649-1661. DOI: 10.1093/cercor/bhz193.
Пицик Елена Николаевна - родилась в Саратове (1994), окончила факультет КНиИТ Саратовского университета по направлению «Программная инженерия» (2015, бакалавриат) и по направлению «Информатика в образовании» (2017, магистратура). После окончания СГУ работала в НОЦ «Системы искусственного интеллекта и нейротехнологии» на базе Саратовского технического университета. С 2019 года работает в лаборатории нейробиологии и когнитивных технологий Университета Иннополис в качестве младшего научного сотрудника. Область научных интересов: анализ временных рядов, нейробиология, количественный анализ рецидивов, двигательная активность мозга. Опубликовала более 20 научных статей по указанным направлениям.
Россия, 420500 Иннополис, Университетская, 1
Лаборатория нейробиологии и когнитивных технологий, Университет Иннополис E-mail: pitsikelena@gmail.com
