Research Information System
Biomedical Signal Processing and Control, vol.77, 2022 (SCI-Expanded)
https://www.sciencedirect.com/science/article/pii/S1746809422002622
ABSTRACT:
In this study, cognitive
and behavioral emotion regulation strategies (ERS) are classified by using
machine learning models driven by a new local EEG complexity approach so called
Frequency Specific Complexity (FSC) in resting-states (eyes-opened (EO), eyes-closed
(EC)). According to international 10–20 electrode placement system, FSC is
defined as entropy estimations in Alpha (8-12Hz) and Beta (12.5-30Hz) frequency
band intervals of non-overlapped short EEG segments to observe local EEG
complexity variations at 62 points on scalp surface. The healthy adults who use
both rumination and cognitive distraction frequently are included in the 1st
groups, while the others who use these strategies rarely are included in the
2nd group with respect to Cognitive Emotion Regulation Questionnaire (CERQ)
scores of them. EEG data and CERQ scores are downloaded from publicly available
data-base LEMON. In order to test the reliability of the proposed method, five
different supervised machine learning methods in addition to two Extreme
Learning Machine models are examined with 5-fold cross-validation for
discrimination of the contrasting groups. The highest classification accuracy
(CA) of 99.47% is provided by Class-specific Cost Regulation Extreme Learning
Machines in EC state. Regarding cortical regions (anterio-frontal, central,
temporal, parieto-occipital), the regional FSC estimations did not provide the
higher performance, however, corresponding statistical distribution shows the
decrease in EEG complexity at mostly anterior cortex in the 1st group
characterized by maladaptive rumination. In conclusion, FSC can be proposed to
investigate cognitive dysfunctions often caused by the use of rumination.
Highlights
· In
the present study, a new quantitative single-channel EEG marker called as
Frequency Specific Complexity for classification of
maladaptive rumination at resting-state.
· The
reliability of the proposed method has been provided by using seven different
5-fold cross-validated classifiers with
respect to both states (eyes-opened vs eyes-closed) and cortical regions.
· The
new findings show that maladaptive rumination cause decreases neuronal
complexity at mostly anterior regions.
DISCUSSION
AND CONCLUSION:
In
the present study, a new indicator so called FSC has been proposed for classification of
maladaptive rumination in resting-state, since electrophysiological findings
commonly reveal close relations between mental disorders and neuro-functional
network across the whole cortex even in the resting state [100], [101], [102]. Usefulness of WE and
AE in computing frequency band specific EEG complexity is tested for estimation
of accurate FSC. Then, the reliability of FSC is tested by examining
supervised machine learning models
and extreme learning machines for
detection of maladaptive rumination from resting-state EEG recordings. In
addition, statistical differences between the groups are computed through
one-way Anova tests according to the corresponding best results.
AE
provided considerable better estimations in comparison to WE due to nonlinear
nature of EEG segments. AE is a nonlinear statistical quantity of signal
irregularity such that its performance depends on the largeness of time-series.
Comparison of AE with WE has been studied in EEG analysis with several goals
such as detection of driving fatigue [103], [104] and
seizure [105].
Alpha
and Beta sub-band specific complexity estimations are averaged to provide FSC
in resting-state, since EEG analysis has frequently focused on Alpha and Beta
sub-bands for estimation of cognitive skills in resting-state in both past and
recent studies [106], [107], [108], [109]. Multi-channel FSC
estimations are considered as separate instants (i.e. features) in two-class
classification steps. Dimension of the features is identical to the number
recording channels that are included in regional features such SET-2, SET-3,
SET-4 and SET-5. So, the largest feature set is SET-1 including FSC values
estimated from 62-channels over a 1min duration trial divided into
short segments of 2s according to 20 individuals in each group. As
the number of recording channels increases, both dimension and number of the
features also increases in feature sets. Therefore, the best classification
performance is obtained through 5-fold cross validated CCR-ELM driven by the
largest feature set including FSC values estimated with AE at EC state. In
addition to higher CA, CCR-ELM is also faster
than the other machine learning models. It can be stated that the
classification performance is decreased as the number of features is decreased,
however, FSC values estimated with AE are able to detect neuro-functional
complexity differences originated from maladaptive rumination at both states
(EC, EO) in each feature set classified by any classifier except GNB.
In reference study, GNB was reported as less successful than both SVM and k-NN [84].
In
conclusion, the results reveal that maladaptive rumination is highly relevant
to FSC estimations (averaged EEG complexity with AE in Alpha and Beta sub-bands
of short EEG segments) at mostly EC state at mostly frontal, centro-parietal
and parietal locations. In detail, FSC values are decreased in maladaptive
ruminations characterized by repetitive thoughts associated with negative
feelings. In literature, the left frontal cortex is assumed to be responsible
for management of arousal and regulation of stress response, while the right
frontal cortex was reported to manage the fight or flight response [86]. However, the current
results prove that the whole cortex is more or less affected by cognitive
abilities in management of negative emotions. Since, the people with a high
tendency to rumination have low ability to manage negative emotions, it is the
important issue to correlate maladaptive rumination with resting-state EEG
based quantitative indicators.
Since,
frequently use of negative ERS called rumination induces deficits in cognitive
functioning, the most important hallmark symptoms of major depression disorder,
wherein individuals retrieve and repetitively rehearse autobiographical and
negatively valance content about both past and current issues [110], [111]. Moreover, persons
who use frequently rumination in daily life, have trouble with sustained
attention. Thus, maladaptive rumination is one of the most problematic
cognitive symptoms in adults. The present study propose a new quantitative tool
to understand the neural mechanism underlying
repetitive negative thoughts triggered by depressive feelings.