- Email: [email protected]
hyperactivity disorder: A multiscale entropy analysis
Accepted Manuscript Title: Complexity analysis of brain activity in attention-deficit/hyperactivity disorder: A multiscale entropy analysis Author: Chenxi Li Yanni Chen Youjun Li Jue Wang Tian Liu PII: DOI: Reference:
S0361-9230(16)30050-8 http://dx.doi.org/doi:10.1016/j.brainresbull.2016.03.007 BRB 8979
To appear in:
Brain Research Bulletin
Received date: Revised date: Accepted date:
27-11-2015 24-2-2016 15-3-2016
Please cite this article as: Li Chenxi, Yanni Chen, Youjun Li, Jue Wang, Tian Liu, Complexity analysis of brain activity in attentiondeficit/hyperactivity disorder: A multiscale entropy analysis, Brain Research Bulletin http://dx.doi.org/10.1016/j.brainresbull.2016.03.007 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Complexity Analysis of Brain Activity in AttentionDeficit/Hyperactivity
Disorder:
A Multiscale
Entropy
Analysis Chenxi Li1, 2, Yanni Chen3, Youjun Li1, 2, Jue Wang1, 2* and Tian Liu1, 2* 1
The Key Laboratory of Biomedical Information Engineering of Ministry of Education, Institute of
Biomedical Engineering, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, China; 2
National Engineering Research Center of Health Care and Medical Devices, Xi’an Jiaotong
University Branch, Xi’an 710049, China 3
Xi’an Children’s Hospital, Xi’an 710003, China1
*
Corresponding author.
Jue Wang Address: Xianning West Road 28#, Xi’an, Shaanxi 710049, P.R. China. Tel. /fax: +86 029 82663497. E‐mail：[email protected] Tian Liu Address: Xianning West Road 28#, Xi’an, Shaanxi 710049, P.R. China. Tel. /fax: +86 029 82663497.E‐mail：[email protected]
* Corresponding author at: The Key Laboratory of Biomedical Information Engineering of Ministry of Education, Institute of Biomedical Engineering, School of Life Science and Technology, Xi’an Jiaotong University, National Engineering Research Center of Health Care and Medical Devices, Xi’an Jiaotong University Branch Jue Wang
Tian Liu
Tel. /fax: +86 029 82663497.
Tel. /fax: +86 029 82663497.
E-mail：[email protected]
E-mail：[email protected]
Highlights
1. The different EEG complexity pattern revealed that the connectivity pattern of brain was significantly changed in ADHD children. 2. The global connectivity was weaker and the local connectivity was enhanced in brain of children with ADHD. 3. The novel MSE method may be a new index to identify and understand the neural mechanism of ADHD.
Abstract The multiscale entropy (MSE) is a novel method for quantifying the intrinsic dynamical complexity of physiological systems over several scales. To evaluate this method as a promising way to explore the neural mechanisms in ADHD, we calculated the MSE in EEG activity during the designed task. EEG data were collected from 13 outpatient boys with a confirmed diagnosis of ADHD and 13 ageand gender-matched normal control children during their doing multi-source interference task (MSIT). We estimated the MSE by calculating the sample entropy values of delta, theta, alpha and beta frequency bands over twenty time scales using coarse-grained procedure. The results showed increased complexity of EEG data in delta and theta frequency bands and decreased complexity in alpha frequency bands in ADHD children. The findings of this study revealed aberrant neural connectivity of kids with ADHD during interference task. The results showed that MSE method may be a new index to identify and understand the neural mechanism of ADHD.
Keywords: ADHD, EEG, Complexity, Multiscale entropy, Connectivity
1. Introduction Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder with cardinal signs of inattention, impulsivity, and hyperactivity. About 5% ~ 8% of children are estimated to meet the diagnosis of ADHD worldwide (Bush, 2010). And ADHD is 2.3 times more
common in boys than girls (Bauermeister et al., 2007). ADHD children are at high risk of having learning disability and delinquency problems, which is a huge burden to the society, family and the children (Konrad and Eickhoff, 2010). Various sources have been used to illustrate that ADHD is associated with an atypical neurodevelopmental process, including neuroimaging, neuropsychological, genetics and neurochemical (Bush, 2010). Structural magnetic resonance imaging (sMRI) studies in ADHD children have observed abnormalities in late developing fronto-striatal, fronto-temporo-parietal and fronto-cerebellar network (Cubillo et al., 2012). Similar to that, functional magnetic resonance imaging (fMRI) studies have provided plenty of evidences on prefrontal cortex dysfunction (Smith et al., 2008), dorsal anterior midcingulate cortex (daMCC) abnormal (Bush, 2008), parietal cortex underactived (Tamm et al., 2006), cerebellum dysfunction (Durston et al., 2007), and abnormal activation of stratum (Scheres et al., 2007). These findings offered ample demonstrations that ADHD is related to disconnection and dysfunctions of brains. The human brain is a complex nonlinear system. Its activity exhibits complex fluctuations, both in spatial and temporal (Takahashi et al., 2009). Electroencephalogram (EEG) is the recording of electrical activity of the brain along the scalp, and it is believed to have the finest temporal resolution. With its easy operation and low cost, EEG is particularly suitable for investigating inherently complex biological signals arising from brain systems. Earlier EEG studies revealed the differences between children with ADHD and normal controls. Willis et al demonstrated theta activity was increased while alpha and beta activities were decreased in ADHD (Willis and Weiler, 2005). Meanwhile, the increased theta/beta power ratio (TBR) was reported as a meaningful EEG feature of ADHD (Barry et al., 2003). However, Arns et al concluded from their meta-analysis that the TBR cannot be considered as a reliable diagnostic measure for ADHD (Arns et al., 2013). The inconsistent research conclusions might come from the power analysis’s internal deficit of being influenced by noise from outside and inside body easily, i.e. EOG, MEG, the machine itself and so on. Therefore, only power analysis can hardly reveal the inherent neural mechanism of ADHD. Over the past decades, novel nonlinear approached based on entropy have been widely used to measure the complexity of physiological signal, including EEG signal (Fernández et al., 2013), electrocardiographic (ECG) (Šliupaitė et al., 2015) and biological codes (Costa et al., 2005).
Approximate entropy calculates the probability of new mode emerging in the signal along with the changing to evaluate the system complexity (Pincus et al., 1992, Richman et al., 2000). The sample entropy is more accurate to measure the probability of newly emerged mode, which is an advanced method of approximate entropy algorithm (Pincus et al., 1992). These traditional methods determine the probability of finding specific patterns or resemblance in a time series, thereby examine the irregularity or predictability in only one time series (Mizuno et al., 2010). Although widely used, the two methods are limited in their scope to only short-range temporal dynamics (Costa et al., 2005). Costa et al proposed Multiscale entropy (MSE) as an extension of sample entropy, quantifying the complexity of physiological signal by measuring the entropy across multiple time scales by coarsegraining procedure (Costa et al., 2005, Costa et al., 2002). This extension to larger temporal scales may reflect information of long-range temporal dynamics (Mizuno et al., 2010). The atypical MSE pattern may reflect disease condition of the brain and provide useful information into the network controlling mechanisms underlying physiological dynamics (Costa et al., 2002). Due to those advantages above, MSE method has been widely used on studies of mental disorders. Tomoyuki et al demonstrated the complexity increased at larger scales factors in Alzheimer’s but significantly reduced over smaller scales (Mizuno et al., 2010). Autism disease has been associated with significantly decreased MSE complexity (Catarino et al., 2011). And MSE study of drug-naive schizophrenia found elevated complexity at higher time scales in centro-temporal (Takahashi et al., 2010). It has been successfully employed as a useful biomarker for early detection of risk for autism spectrum disorder abnormalities in infants’ cognitive development (Catarino et al., 2011, Bosl et al., 2011). And also, atypical cortical dynamics in schizophrenia was found to be easily characterized and understand by MSE analytic method. (Takahashi et al., 2010). To characterize the complexity pattern of brain function of ADHD and Control, we tested the subjects’ attention related performance using multi-source interference task (MSIT) (Bush et al., 2003). The multi-source interference task (MSIT) combines multiple dimensions of cognitive interference in a single task (i.e., Stroop tasks, Eriksen Flanker-type tasks, and Simon effect task variants) (Bush and Shin, 2006). It is known as an effective way to active the network of brain regions involving attention and cognitive control with decision-making, target detection, response selection, stimulus/response competition and so on (Bush and Shin, 2006). Various studies using MSIT have been reported in patients with ADHD (Matsubara et al., 2014, Nakashima et al., 2014);
for instance, children with ADHD showed abnormal prefrontal activation in response to multiple interference control (Nakashima et al., 2014). In the present study, first, our aim was to investigate the MSE pattern of EEG in a population of ADHD children and healthy control subjects. We hypothesize that ADHD group will present significantly changed MSE pattern compared to control group. Second, we conduct a traditional power analysis and try to explore relevancy between MSE and power spectrum at different frequencies. To test this hypothesis, EEG was used to calculate the complexity and power spectrum of children with ADHD and normal Control during the MSIT performance.
2. Materials and methods 2.1. Participants EEG data were collected from 13 outpatient boys with a confirmed diagnosis of ADHD (age range, 6-13years; mean age, 8.5 ± 3.17 years; mean IQ score, 93 ± 16.8) and 13 age- and gendermatched healthy control children (age range, 6-13years; mean age, 7.9 ± 1.98 years; mean IQ score, 98 ± 18.3). All the participants are right handed. According to the Declaration of Helsinki, all parents of participants were informed with written consent. The study was approved by ethics committees of Xi’an Jiaotong Unversity School of Medicine. All participants were given full-scale and verbal IQ test with scores > 80 (Wechsler intelligence Scale for Children, Fourth Edition). All the 26 parents filled translated versions of the Conner’s Parent Rating Scales-Revised and the ADHD Rating Scale-VI for current and childhood-onset ADHD symptoms. The ADHD Rating Scale-IV is linked directly to the diagnostic criteria for ADHD of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) (Dupaul et al., 2006), which has been wildly used as a reliable method to diagnose ADHD (Konrad et al., 2012, Uddin et al., 2008). ADHD group had higher DSM-IV scores (mean, 13.23 ± 3.22) than that of typical control group (mean, 5.23 ± 2.09) (P < 0.01).Subjects with higher DSM-IV scores suffer from more serious ADHD. The ADHD children were include if they had (1)
6 of 9 DSM-IV symptoms of inattention as well as
6
of 9 symptoms of hyperactivity/impulsivity; (2) no record of brain injuries, medical history, and other neuropsychiatric disorders; and (3) no signs of cognitive deficit, learning disabilities, as well as communication problems. Exclusion criteria for normal control group were the same, and also
they do not have ADHD disease or have individuals in their families. 2.2. Task Description The multi-source interference task (MSIT) is believed to have the ability activating the cingulofroto-paretal (CFP) network (Stroop, 1935). The MSIT task is designed using system STIM2 provided by Neuroscan company. The operation is practicable for the both groups and can be finished within 15 minutes. Subjects were given a button with two keys, of which the left one and right one respectively represented the number 1 and 2. In this study, the interference trails were included only. In the trails, there are 3 numbers on the screen combined by 1 and 2, the total set of the number combinations are 112, 212, 221 and 211. During the interference task, subjects were instructed to use the index and middle finger of right hand to respond to the target number among the three, the target numbers were never placed congruently with their button box (Stroop effect (Eriksen and Eriksen, 1974)), and the Flanker stimuli (Bush and Shin., 2006)were always potential target. One example of the stimuli in a single trail is shown in Figure 1. Before the experiment, each subject underwent some adaptive exercises. All the interference trials would appear at the center of the screen every 3 seconds with 15 second of fixation (a white dot in the center of the screen). During fixation, subjects completed 80 random trials. 2.3. EEG data collection EEG data was acquired from 19 electrodes using the International ten-twenty system. The scalp locations are: Fp1, Fp2, F3, F4, F7, F8, Fz, T3, T4, T7, T8, C3, Cz, P3, P4, Pz, O1 and O2 (Neuroscan; Compumedics Limited, Charlotte, NC). The impedance of all sits was < 5 kΩ. Reference electrode was at the left mastoid, and the ground electrode was between Fpz and Fz. Eyemovement were monitored vertically and horizontally separately, the vertical electro-oculogram was recorded from electrode attached above and below the left eye, and a horizontal electro-oculogram was recorded from the outer canthi of both eyes. EEG was sampled with 32 bits of accuracy. The analog/digital sampling rate was 1,000 Hz. All subjects were seated inside an acoustically and electromagnetically shielded room. They were asked to do their best to keep body steady during the experiments. 2.4. EEG recording All subjects were able to react to the task in about 2 seconds (average reaction times of the two groups were 1,703 ± 522 and 1,162 ± 462ms), so we defined the period from target stimulus onset
to 2 seconds as the response period. Only correct trials were segmented, and each subject had 10 epochs (total, 20 seconds; 20,000 samples) of artifact free data (no ocular and muscular artifacts). Filter analysis was performed on every EEG data segment, and EEG components of the following 4 frequency bands were obtained: delta (0.5 – 3 Hz), theta (4 – 7 Hz), alpha (8 – 13 Hz), and beta (13 – 30 Hz). For each participant, MSE as well as relative power was calculated on each frequency band. 2.5. Multiscale entropy (MSE) The MSE method was developed by Costa et al in 2005, and it has been used to quantifies the complexity of signal by calculating the sample entropy (SampEn) over multiple temporal scales which was realized by coarse-grained procedure (Costa et al, 2005, Costa et al., 2002). Sample entropy was proposed by Richman and Moorman in 2000 (Richman and Moorman, 2000). The SampEn measures the regularity of time series, it is a modified version of Approximate Entropy (ApEn) (Richman et al., 2004). Given an one-dimensional EEG discrete time series x
,…
and the scale factor , the
time series is calculated in to consecutive and nonoverlapping time series procedure:
=
∑
,1
j
by coarse-grained
N/ τ . Scale one is simply the original time series
representing the short-range temporal scale and larger scale factor represents longer temporal scales. And then calculates the SampEn of each series
. SampEn was defined as the negative of the
logarithmic condition probability that two similar sequences of m consecutive data points will 1 (Catarino et al., 2011, Richman and Moorman, 2000).
remain similar at the next point SampEn (m, r, N) = -ln ,i
r /
, where
j}/{number of all probable pairs = N
distance between vectors
and
m
= {number of pairs (i, j) with 1 / N
m }, where
. Therein, m is dimension of vectors
and
denotes the and r is
the tolerable distance between the two vectors and N represents the length of the time series. Studies by Richman in 2004 have proved that dimension m = 1 or m = 2 and tolerable distance r = 0.1~0.25 times the S.D. (Standard deviation) of the time series have the best statistical validity (Richman and Moorman, 2000). In this study, we used the parameter m = 2, r = 0.15×S.D., N = 20000 data points and we set the scale factor as 20. After the coarse-grained procedure, we still have 20000/20 = 1000 data points, which has been proved enough to obtain reliable estimation of sample entropy value
(Richman and Moorman, 2000). 2.6. Power analysis Besides MSE analysis, a power analysis was performed as a conventional EEG measurement so as to explore the relationship between the system complexity and EEG power spectrum. We used the pwelch function in MATLAB (version 7.12). The data were separated into nine equal lengths, each with 50% overlap. Four standard bands, delta (0.5 – 3 Hz), theta (4 – 7 Hz), alpha (8 – 13 Hz), and beta (13 – 30 Hz) were included in the power analysis. For statistical analysis, the relative power change was then (power in each frequency divided by total power across all frequency bands) calculated. 2.7. Statistical analysis The statistical analysis was carried out by SPSS 13.0 for Windows. Kolmogorov-Smirnov normality test was used to confirm the distribution normality of MSE value for every group and every channel. Mauchly’s Test of Sphericity was used to test whether the data are appropriately for the ANOVA or a data correction is needed, then the Greenhouse-Geisser adjustment was applied to the degrees of freedom for all analysis. The alpha significance value was set as
0.05 for all
analysis (i.e., both for repeated ANOVA and post-hoc t-tests). To test for differences in behavioral results, a two - groups (ADHD vs. Control)
1 condition
(MSIT) analysis of variance (ANOVA) was done for response time. For MSE analysis, we applied 4-way repeated-measures analysis of variance (ANOVA) to test the group differences with group (ADHD vs. Control) as the between-subjects factor, and the scale factor (sf: τ=20), frequency bands (delta, theta, alpha, beta) as well as channel (FP1, FP2, F7, F3, FZ, F4, F8, T7, T8, FT7, FT8, CZ, C3, C4, PZ, P3, P4, O1, O2) as the within-subjects factors. Next for every frequency band, we tested the group difference with group (ADHD vs. Control) as the between-subjects factor, and the SF (τ = 20), and channel (FP1, FP2, F7, F3, FZ, F4, F8, T7, T8, FT7, FT8, CZ, C3, C4, PZ, P3, P4, O1, O2) as the within-subjects factors. And then for every electrode in every frequency band, repeated measures ANOVA with group (ADHD vs. Control) as the between-subject factor and sf (τ = 20) as the with-in subject factor. Subsequently, a repeated measures analysis of variance (ANOVA) was performed to test for group differences in relative power analysis. The group (ADHD and control) was the betweensubjects factor, frequency bands (delta, theta, alpha, and beta) and channels were the within-subjects
factors. For every band under every channel, 2-way repeated ANOVA was performed with group (ADHD vs. Control) as between-subject factor, and frequency band as within-subject factors. In case of the significant group-by-band interaction, then independent t-test was used to find group difference for each frequency band in each channel. To investigate the relevancy of MSE and power spectra, we presented Spearman’s rank order correlations between relative power and MSE values for the two groups.
3. Results 3.1 Behavioral performance The mean reaction times for the ADHD and the Control groups were 1,703 ± 522 ms and 1,162 ± 462 ms. During the interference trails, reaction time of the ADHD were significantly longer than that was for Controls (F1,18 = 6.019, p = 0.025).
3.2. Power analysis The results revealed that during the MSIT tasks, the brain regions in frontal and central region exhibited relatively lower power in delta frequency band but showed relatively higher power in frontal, temporal and central region in theta and alpha frequency bands. The results showed no main significant effects for group (F1, 342 = 3.584, p = 0.059) and channel (F18, 342 = 0.717, p = 0.794). Also, no significant group-by-channel (F18, 342 = 1.112, p = 0.338) and group-by-band-by-channel (F32.862, 624.383 = 1.241, p = 0.194) interactions were found. However, there was a significant groupby-fb (frequency band) interaction (F1.826, 624.383 = 35.336, p < 0.001). We used 2-way repeated ANOVA for investigating the group-by-fb (frequency band) interaction (Table 1). This interaction was explored further using independent t-test to find group difference separately for each frequency band and channel. The results are shown in Figure 2, indicating a significant group-by-band interaction in Fp1 (F1.788, 32.192 = 4.611; p = 0.020), F3 (F2.219, 39.948
= 4.741; p = 0.012), Fz (F1.463, 26.340 =4.586; p = 0.046), FT7 (F1.632, 29.383 = 8.905; p = 0.008),
C3 (F1.370, 24.653 = 9.288; p = 0.003), C4 (F1.553, 27.957 = 6.806; p = 0.018), and Pz (F1.785, 32.136 = 6.970; p = 0.004). Post-hoc independent t-test showed significant decreases in delta and beta and increases in theta and alpha frequency bands in ADHD, compared to Control group (Figure 2).
3.3. MSE analysis The MSE results showed no main effect of group difference (F1, 1368 = 3.152, p = 0.076) in 4way ANOVA. However when testing the group difference in each frequency band, the result showed there exist main effect of group difference in delta (F1, 342 = 60.703, p < 0.001) and alpha (F1, 342 = 29.552, p < 0.001) frequency bands that relative to Control group, the mean sample entropy (SampEn) values of ADHD group was increased in delta frequency band (see in Figure3) and decreased in alpha frequency band (see in Figure 4) as the scale factor increases, but the differential in theta and beta is not that significant. Figure 3 and figure 4 reflect group differences of SampEn values in delta and alpha frequency bands in each channel for increasing scale factor. The MSE results showed no main effect of group difference (F1, 1368 = 3.152, p = 0.076) in 4way ANOVA. However when testing the group difference in each frequency band, the result showed there exist main effect of group difference in delta (F1, 342 = 60.703, p < 0.001) and alpha (F1, 342 = 29.552, p < 0.001) frequency bands that relative to Control group, the mean sample entropy (SampEn) values of ADHD group was increased in delta frequency band (see in Figure3) and decreased in alpha frequency band (see in Figure 4) as the scale factor increases, but the differential in theta and beta is not that significant. Figure 3 and figure 4 reflect group differences of SampEn values in delta and alpha frequency bands in each channel for increasing scale factor. 3.4 Relationship between power analysis and MSE In order to explore the relationship between MSE and the power analysis, we divided the scale factors into four components as Tomoyuki did in AD research (i.e. sf 1 – 5, sf 6 – 10, sf 11 – 15, and sf 16 – 20). According to the statistical analysis, the result showed no significant relationship between MSE and power analysis. Table 4 and 5 demonstrate the example (Fp1) of correlation result between MSE value and the relative power for ADHD and normal control subjects.
4. Discussion The presented study found that the reaction time were significantly longer for ADHD children than that of Control group, thus we inferred that the behavioral result might related to inattention, which is the core clinical symptom of ADHD. In our power analysis, we found significantly increased relative theta and alpha power in frontal,
temporal and central region of brain as well as decreased relative delta power in frontal and central region during MIST task. These findings are in line with many previous power analysis of EEG in ADHD (Willis and Weiler, 2005, Clarke et al., 2001, Fernández et al., 2009) that the ADHD children display an atypical neural pattern during the cognitive information processing procedure. Although power spectrum analysis, i.e. increased theta/beta power ratio (TBR), has been widely used as a biomarker of ADHD, Arns proposed in the meta-analysis study that TBR cannot be considered a reliable diagnostic measure of ADHD (Arns et al., 2012). This inconstancy may lead to misdiagnose of ADHD disease. In the present study, we used the novel MSE analysis to investigate complexity of EEG signal in ADHD children during their MSIT task. MSE method has already been demonstrated as a sufficient way to quantify the complexity of dynamical signals over several time scales (Bosl et al., 2011), and can be used as a reliable index of some mental disorders (Mizuno et al., 2010, Takahashi et al., 2010). So we used the MSE method to explore the neurophysiological evidence for the abnormal neural pattern of activated condition, and thus identifying whether it can be a new nonlinear dynamic index for analyzing EEG signal in children with ADHD. Although early studies reported relatively decreased EEG complexity of ADHD subjects during tasks (Willis and Weiler, 2005, Fernández et al., 2009), our findings of EEG complexity of ADHD were different. In our MSE analysis study, we measured the complexity of EEG over multiple time scales in four frequency bands and we found that the complexity was significantly elevated in delta frequency band and decreased in alpha frequency band at larger scale factors in ADHD compared to normal subjects. Abnormal complexity in neurophysiologic signals might reflect aberrant neural connectivity pattern in mental disorders (Takahashi, 2013, Sporns et al., 2000), thus we hypothesize that our MSE findings of changed EEG complexity in ADHD may reflect the aberrant neural connectivity pattern. Empirical and computational studies suggested that changes in functional connectivity may underlie specific perceptual and cognitive state and involve the integration of information across specialized areas of the brain, and the interplay of functional areas and integration can be quantified particular in neural complexity (Sporns et al., 2000), so they inferred that the disconnection of brain is associated with an increase of complexity (Takahashi, 2013, Sporns et al., 2000). In the current study, we found that the complexity of ADHD was significantly increased in delta frequency band and decreased in alpha band at lager scale factors
than normal children, which indicated that the connectivity pattern was significantly changed in ADHD children. Previous research revealed that specific frequency bands reflect different functional state of brain activities (Takahashi, 2013). Higher frequency band and lower frequency band originate from small neuronal populations and larger populations (Takahashi, 2013, Schnitzler and Gross, 2005), which means the complexity pattern of lower frequency band may reflect global character of the brain in ADHD while the in higher frequency may reflect the local connectivity feature of brain. In our study, we found different change pattern of complexity in delta and alpha frequency bands. Based on the conclusions above, we suspect that the global connectivity was weaker and the local connectivity was enhanced in brain of children with ADHD. Supporting this is the work by Tomasi et al, they quantified short-and long-range functional density in the brain with ADHD and normal controls and showed a pattern of hyper short-range connectivity and hypo long-range connectivity of brain in children with ADHD (Tomasi and Volkow, 2012). Also the research about small-world properties of brain network in ADHD concluded that the global characteristic became weaken and the local feature was enhanced in ADHD (Liu et al., 2014, Wang et al., 2009, Wang et al., 2013). These previous neuroimaging results support our hypotheses that altered MSE pattern of EEG reflect the atypical neural connectivity in ADHD, and ADHD is possibly in association with reduced long-range connectivity and increased local brain connectivity. Comparing MSE and power analysis, we found no significant relevancy between relative power and MSE, that is to say, there is no relationship between the complexity measures and power spectrum. This is supported by the MSE study of autism that altered complexity pattern are not a reflection of changes in power spectrum (Catarino et al., 2011).
5. Conclusions In summary, in the present study, we explored the MSE method to analyze the EEG signal of children with and without ADHD during their MIST task. The results of the present study reveal a changed complexity pattern for ADHD children compared to normal controls in delta and alpha frequency bands. In addition, there was no significant correlation between complexity and power analysis for the two group on MSIT task. These findings reveal that the brain connectivity was
changed in ADHD, suggesting that ADHD may be related to long-range disconnection as well as enhanced local connection. Therefore, the MSE method may provide a new view to characterize and analyze aberrant brain network in ADHD and a new aspect to understand neurophysiological mechanism of ADHD. However, due to the small sample size, findings presented in the study might be limited. Further studies should be carried out to confirm that our results are not a particular case. The potential effect of gender and severity of the disease are needed to be taken into consideration. In future studies, the resting state are necessarily needed to clarify this outstanding of this method. Also, there’s one issue need to notice that whether the MSE method would be sensitive to the different neuronal activities due to different tasks (Catarino et al., 2011). So future study should involve different difficulty levels and forms of tasks to investigate the sensitivity of the MSE method to the tasks. If future studies confirm that complexity pattern measured by MSE can indeed be a reliable clinical EEG index of ADHD, this might be a new way to diagnose ADHD.
Funding The author(s) received no financial support for the research, authorship, and/or publication of this article. Declaration of Conflicting Interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
6. Acknowledgement The study was supported by grant 61503295 from Youth Project of National Natural Science Foundation of China and grant 2015JQ8318 from Natural Science Foundation of Shaanxi Province. We thank the Xi’an Children’s Hospital for their help of data collection and our colleagues who provided help to this article.
Reference: Arns, M., Conners, C. K., & Kraemer, H. C. (2012). A decade of EEG theta/beta ratio research in ADHD: a meta-analysis. Journal of attention disorders, 1087054712460087. Barry, R. J., Clarke, A. R., & Johnstone, S. J. (2003). A review of electrophysiology in attentiondeficit/hyperactivity disorder: I. Qualitative and quantitative electroencephalography. Clinical neurophysiology, 114(2), 171-183. Bauermeister, J. J., Shrout, P. E., Chávez, L., Rubio‐Stipec, M., Ramírez, R., Padilla, L. & Canino, G. (2007). ADHD and gender: are risks and sequela of ADHD the same for boys and girls? Journal of Child Psychology and Psychiatry, 48(8), 831-839. Bosl, W., Tierney, A., Tager-Flusberg, H., & Nelson, C. (2011). EEG complexity as a biomarker for autism spectrum disorder risk. BMC medicine, 9(1), 1. Bush, G., Shin, L. M., Holmes, J., Rosen, B. R., & Vogt, B. A. (2003). The Multi-Source Interference Task: validation study with fMRI in individual subjects. Molecular psychiatry, 8(1), 60-70. Bush, G., & Shin, L. M. (2006). The Multi-Source Interference Task: an fMRI task that reliably activates the cingulo-frontal-parietal cognitive/attention network. NATURE PROTOCOLSELECTRONIC EDITION-, 1(1), 308. Bush, G. (2009). Dorsal anterior midcingulate cortex: roles in normal cognition and disruption in attention-deficit/hyperactivity disorder (pp. 207-218). Oxford University Press, NY, USA. Bush,
G.
(2010).
Attention-deficit/hyperactivity
disorder
and
attention
networks.
Neuropsychopharmacology, 35(1), 278-300. Catarino, A., Churches, O., Baron-Cohen, S., Andrade, A., & Ring, H. (2011). Atypical EEG complexity in autism spectrum conditions: a multiscale entropy analysis. Clinical neurophysiology, 122(12), 2375-2383. Clarke, A. R., Barry, R. J., McCarthy, R., & Selikowitz, M. (2001). EEG-defined subtypes of children with attention-deficit/hyperactivity disorder.Clinical Neurophysiology, 112(11), 20982105. Costa, M., Goldberger, A. L., & Peng, C. K. (2002). Multiscale entropy analysis of complex physiologic time series. Physical review letters, 89(6), 068102. Costa, M., Goldberger, A. L., & Peng, C. K. (2005). Multiscale entropy analysis of biological signals.
Physical review E, 71(2), 021906. Cubillo, A., Halari, R., Smith, A., Taylor, E., & Rubia, K. (2012). A review of fronto-striatal and fronto-cortical brain abnormalities in children and adults with Attention Deficit Hyperactivity Disorder (ADHD) and new evidence for dysfunction in adults with ADHD during motivation and attention. Cortex, 48(2), 194-215. Dupaul, G. J., Power, T. J., Anastopoulos, A. D., & Reid, R., (2006). ADHD Rating Scale-IV: Checklists, norms, and clinical interpretation. Journal of Psychoeducational Assessment 24 (2), 172178. Durston, S., Davidson, M. C., Mulder, M. J., Spicer, J. A., Galvan, A., Tottenham, N., & Casey, B. J. (2007). Neural and behavioral correlates of expectancy violations in attention ‐ deficit hyperactivity disorder. Journal of Child Psychology and Psychiatry, 48(9), 881-889. Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the identification of a target letter in a nonsearch task. Perception & psychophysics, 16(1), 143-149. Fernández, A., Quintero, J., Hornero, R., Zuluaga, P., Navas, M., Gómez, C. & Ortiz, T. (2009). Complexity analysis of spontaneous brain activity in attention-deficit/hyperactivity disorder: diagnostic implications. Biological psychiatry, 65(7), 571-577. Fernández, A., Gómez, C., Hornero, R., & López-Ibor, J. J. (2013). Complexity and schizophrenia. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 45, 267-276. Konrad, A., Dielentheis, T. F., El Masri, D., Dellani, P. R., Stoeter, P., Vucurevic, G., & Winterer, G. (2012). White matter abnormalities and their impact on attentional performance in adult attentiondeficit/hyperactivity disorder. European archives of psychiatry and clinical neuroscience, 262(4), 351-360. Konrad, K., & Eickhoff, S. B. (2010). Is the ADHD brain wired differently? A review on structural and functional connectivity in attention deficit hyperactivity disorder. Human brain mapping, 31(6), 904-916. Lake, D. E., Richman, J. S., Griffin, M. P., & Moorman, J. R. (2002). Sample entropy analysis of neonatal heart rate variability. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 283(3), R789-R797. Liu, T., Chen, Y., Lin, P., & Wang, J. (2014). Small-world brain functional networks in children with attention-deficit/hyperactivity disorder revealed by EEG synchrony. Clinical EEG and neuroscience,
1550059414523959. Matsubara, T., Matsuo, K., Nakashima, M., Nakano, M., Harada, K., Watanuki, T. & Watanabe, Y. (2014). Prefrontal activation in response to emotional words in patients with bipolar disorder and major depressive disorder. Neuroimage, 85, 489-497. Mizuno, T., Takahashi, T., Cho, R. Y., Kikuchi, M., Murata, T., Takahashi, K., & Wada, Y. (2010). Assessment of EEG dynamical complexity in Alzheimer’s disease using multiscale entropy. Clinical Neurophysiology, 121(9), 1438-1446. Nakashima, M., Matsuo, K., Hashimoto, A., Nakano, M., Fujii, Y., Matsushige, T. & Sugiyama, S. (2014). Prefrontal Abnormality in Children with ADHD during Cognitive Interference Control: a Functional NIRS Study. The bulletin of the Yamaguchi Medical School, 61(3), 37-47. Nazari, M. A., Wallois, F., Aarabi, A., & Berquin, P. (2011). Dynamic changes in quantitative electroencephalogram during continuous performance test in children with attentiondeficit/hyperactivity disorder. International journal of psychophysiology, 81(3), 230-236. Pincus, S. M., & Huang, W. M. (1992). Approximate entropy: statistical properties and applications. Communications in Statistics-Theory and Methods, 21(11), 3061-3077. Richman, J. S., & Moorman, J. R. (2000). Physiological time-series analysis using approximate entropy and sample entropy. American Journal of Physiology-Heart and Circulatory Physiology, 278(6), H2039-H2049. Richman, J. S., Lake, D. E., & Moorman, J. R., (2004). Sample entropy. Methods in Enzymology 384 (1), 172-184. Scheres, A., Milham, M. P., Knutson, B., & Castellanos, F. X. (2007). Ventral striatal hyporesponsiveness during reward anticipation in attention-deficit/hyperactivity disorder. Biological psychiatry, 61(5), 720-724. Schnitzler, A., & Gross, J. (2005). Normal and pathological oscillatory communication in the brain. Nature reviews neuroscience, 6(4), 285-296. Šliupaitė, A., Navickas, Z., & Vainoras, A. (2015). Evaluation of Complexity of ECG Parameters Using Sample Entropy and Hankel Matrix. Elektronika ir Elektrotechnika, 92(4), 107-110. Smith, A. B., Taylor, E., Brammer, M., Halari, R., & Rubia, K. (2008). Reduced activation in right lateral prefrontal cortex and anterior cingulate gyrus in medication ‐ naïve adolescents with attention deficit hyperactivity disorder during time discrimination. Journal of Child Psychology and
Psychiatry, 49(9), 977-985. Sporns, O., Tononi, G., & Edelman, G. M. (2000). Connectivity and complexity: the relationship between neuroanatomy and brain dynamics. Neural Networks, 13(8), 909-922. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of experimental psychology, 18(6), 643. Takahashi, T., Cho, R. Y., Murata, T., Mizuno, T., Kikuchi, M., Mizukami, K. & Wada, Y. (2009). Age-related variation in EEG complexity to photic stimulation: a multiscale entropy analysis. Clinical Neurophysiology, 120(3), 476-483. Takahashi, T., Cho, R. Y., Mizuno, T., Kikuchi, M., Murata, T., Takahashi, K., & Wada, Y. (2010). Antipsychotics reverse abnormal EEG complexity in drug-naive schizophrenia: a multiscale entropy analysis. Neuroimage, 51(1), 173-182. Takahashi, T. (2013). Complexity of spontaneous brain activity in mental disorders. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 45, 258-266. Tamm, L., Menon, V., & Reiss, A. L. (2006). Parietal attentional system aberrations during target detection in adolescents with attention deficit hyperactivity disorder: event-related fMRI evidence. American Journal of Psychiatry. Tomasi, D., & Volkow, N. D. (2012). Abnormal functional connectivity in children with attentiondeficit/hyperactivity disorder. Biological psychiatry, 71(5), 443-450. Uddin, L. Q., Kelly, A. C., Biswal, B. B., Margulies, D. S., Shehzad, Z., Shaw, D., ... & Milham, M. P. (2008). Network homogeneity reveals decreased integrity of default-mode network in ADHD. Journal of neuroscience methods, 169(1), 249-254. Willis, W. G., & Weiler, M. D. (2005). Neural substrates of childhood attention-deficit/hyperactivity disorder: Electroencephalographic and magnetic resonance imaging evidence. Developmental Neuropsychology, 27(1), 135-182. Wang, L., Zhu, C., He, Y., Zang, Y., Cao, Q., Zhang, H. & Wang, Y. (2009). Altered small‐world brain functional networks in children with attention‐deficit/hyperactivity disorder. Human brain mapping, 30(2), 638-649. Wang, P., Zhu, D., Li, X., Chen, H., Jiang, X., Sun, L. & Wang, Y. (2013, April). Identifying functional connectomics abnormality in attention deficit hyperactivity disorder. In Biomedical Imaging (ISBI), 2013 IEEE 10th International Symposium on (pp. 544-547). IEEE.
Figure1. Illustration of the MSIT. During the interference trails, subjects were instructed to press the button to figure out the number that differs from the other 2 numbers (Flanker stimuli), but the target never matched the button location (Stroop effect). In this example, the number 2 is different from the 2 numbers, so the correct answer would be to press button 2.
Figure 2. Relative power analysis of ADHD children and normal control subjects. Post-hoc analysis between ADHD and Control: *p 0.05
Figure 3. Multiscale entropy (MSE) analysis in delta of 13 ADHD (blue line) children and normal control (red line) for each electrode. Each panel presents the average of MSE values of each group. *p < 0.05
Figure 4. Multiscale entropy (MSE) analysis in alpha of 13 ADHD (blue line) children and normal control (red line) for each electrode. Each panel presents the average of MSE values of each group. *p < 0.05
Table 1 Group-by-fb interaction significance values for each channel in power analysis. Group-by-band
Group-by-band interaction
Channel Fp1
F1.788, 32.192 = 4.611; p = 0.020*
Fp2
F1.858, 33.446 = 1.212; p = 0.308
F7
F1.624, 29.227 = 2.841; p = 0.084
F3
F2.219, 39.948 = 4.741; p = 0.012*
Fz
F1.463, 26.340 =4.586; p = 0.044*
F4
F2.015, 36.271 = 1.134; p = 0.333
F8
F1.523, 27.410 = 1.600; p = 0.221
T7
F1.420, 25.567 = 2.168; p = 0.147
T8
F1.559, 28.061 = 1.168; p = 0.314
FT7
F1.632, 29.383 = 3.475; p = 0.050*
FT8
F1.512, 27.214 = 1.561; p = 0.228
Cz
F1.728, 31.102 = 8.771; p = 0.004
C3
F1.370, 24.653 = 9.288; p = 0.003*
C4
F1.553, 27.957 = 6.806; p = 0.018*
Pz
F1.785, 32.136 = 6.970; p = 0.004*
P3
F1.541, 27.740 = 2.005; p = 0.162
P4
F1.629, 29.320 = 0.528; p = 0.559
O1
F1.305, 23.482 = 2.570; p = 0.115
O2
F1.290, 23.217 = 1.613; p = 0.221
*These channels showed a significance (p .05) group-by-fb interaction.
Table 2 Group-by-sf interaction significance values for each channel in delta frequency band. Group-by-sf
Group-by-sf interaction
channel Fp1
F1.313, 23.634 = 9.545; p = 0.006*
Fp2
F1.395,25.115 = 4.341; p = 0.052
F7
F1.263, 22.736 = 6.980; p = 0.017*
F3
F1.249, 22.482 = 4.155; p = 0.056
Fz
F1.771, 31,872 = 7.196; p = 0.015*
F4
F1.750, 31.502 = 1.466; p = 0.242
F8
F1.503, 27.048 = 0.580; p = 0.456
T7
F1.183, 21,288 = 9.783; p = 0.006*
T8
F1.713, 30.831 = 0.855; p = 0.368
FT7
F1.218, 21.928 = 6.471; p = 0.020*
FT8
F1.501, 27.011 = 0.583; p = 0.455
Cz
F1.449, 26.075 = 1.849; p = 0.191
C3
F1.577, 28.378 = 9.954; p = 0.005*
C4
F1.866, 33.946 = 17.547; p = 0.001*
Pz
F1.910, 34,377 = 2.480; p = 0.133
P3
F1.647, 29.655 = 1.659; p = 0.214
P4
F1.482, 26.684 = 0.002; p = 0.962
O1
F1.795, 32.317 = 0.125; p = 0.728
O2
F1.711, 30.801 = 0.049; p = 0.827
*These channels showed a significant (p
0.05) group-by-sf interaction.
Table 3 Group-by-sf interaction significance values for each channel in alpha frequency band. Group-by-sf
Group-by-sf interaction
channel Fp1
F1.466, 26.385 = 3.879; p = 0.064
Fp2
F1.278, 23.06 = 0.270; p = 0.610
F7
F2.384 42.906 = 0.373; p = 0.549
F3
F1.572, 28.291 = 2.215; p = 0.154
Fz
F1.320, 23.769 = 7.780; p = 0.012*
F4
F1.222, 21.988 = 4.047; p = 0.059
F8
F1.236, 22.241 = 0.932; p = 0.347
T7
F1.913, 34.437 = 6.057; p = 0.024*
T8
F1.176, 21.164 = 0.803; p = 0.382
FT7
F2.165, 38.974 = 0.016; p = 0.900
FT8
F1.212, 21.814 = 1.130; p = 0.302
Cz
F1.138, 20.486 = 14.225; p = 0.001*
C3
F1.138, 20.486 = 6.522; p = 0.020*
C4
F1.083, 19.492 = 2.298; p = 0.147
Pz
F1.706, 19.373 = 8.516; p = 0.009*
P3
F1.203, 21,648 = 0.690; p = 0.417
P4
F1.151, 20.712 = 0.193; p = 0.665
O1
F1.196, 21.520 = 1.042; p = 0.321
O2
F1.310, 23.587 = 0.445; p = 0.513
*These channels showed a significant (p
0.05) group-by-sf interaction.
Table 4 Correlations between MSE and relative power among ADHD subjects for Fp1. Relative power
MSE sf: 1-5
MSE sf: 6-10
MSE sf: 11-15
MSE sf: 16-20
Delta
-0.159
-0.159
0.402
0.598
0.662
0.662
0.249
0.068
0.829
0.829
0.549
0.341
0.003
0.003
0.100
0.334
-0.037
-0.037
-0.146
-0.232
0.920
0.920
0.687
0.519
-0.744
-0.744
-0.695
-0.622
0.014
0.014
0.026
0.055
Theta
Alpha
Beta
The upper number in each represents r-values for correlations between relative frequency power value and averaged MSE value across scale factors, and the lower number represents the p-value
Table 5 Correlations between MSE and relative power among normal control subjects for Fp1. Relative power
MSE sf: 1-5
MSE sf: 6-10
MSE sf: 11-15
MSE sf: 16-20
Delta
-0.576
-0.552
-0.539
-0.636
0.082
0.098
0.108
0.048
0.709
0.685
0.758
0.770
0.022
0.029
0.011
0.009
0.842
0.867
0.830
0.867
0.002
0.001
0.003
0.001
0.333
0.321
0.285
0.382
0.347
0.365
0.425
0.276
Theta
Alpha
Beta
The upper number in each represents r-values for correlations between relative frequency power value and averaged MSE value across scale factors, and the lower number represents the p-value
S0361-9230(16)30050-8 http://dx.doi.org/doi:10.1016/j.brainresbull.2016.03.007 BRB 8979
To appear in:
Brain Research Bulletin
Received date: Revised date: Accepted date:
27-11-2015 24-2-2016 15-3-2016
Please cite this article as: Li Chenxi, Yanni Chen, Youjun Li, Jue Wang, Tian Liu, Complexity analysis of brain activity in attentiondeficit/hyperactivity disorder: A multiscale entropy analysis, Brain Research Bulletin http://dx.doi.org/10.1016/j.brainresbull.2016.03.007 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Complexity Analysis of Brain Activity in AttentionDeficit/Hyperactivity
Disorder:
A Multiscale
Entropy
Analysis Chenxi Li1, 2, Yanni Chen3, Youjun Li1, 2, Jue Wang1, 2* and Tian Liu1, 2* 1
The Key Laboratory of Biomedical Information Engineering of Ministry of Education, Institute of
Biomedical Engineering, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, China; 2
National Engineering Research Center of Health Care and Medical Devices, Xi’an Jiaotong
University Branch, Xi’an 710049, China 3
Xi’an Children’s Hospital, Xi’an 710003, China1
*
Corresponding author.
Jue Wang Address: Xianning West Road 28#, Xi’an, Shaanxi 710049, P.R. China. Tel. /fax: +86 029 82663497. E‐mail：[email protected] Tian Liu Address: Xianning West Road 28#, Xi’an, Shaanxi 710049, P.R. China. Tel. /fax: +86 029 82663497.E‐mail：[email protected]
* Corresponding author at: The Key Laboratory of Biomedical Information Engineering of Ministry of Education, Institute of Biomedical Engineering, School of Life Science and Technology, Xi’an Jiaotong University, National Engineering Research Center of Health Care and Medical Devices, Xi’an Jiaotong University Branch Jue Wang
Tian Liu
Tel. /fax: +86 029 82663497.
Tel. /fax: +86 029 82663497.
E-mail：[email protected]
E-mail：[email protected]
Highlights
1. The different EEG complexity pattern revealed that the connectivity pattern of brain was significantly changed in ADHD children. 2. The global connectivity was weaker and the local connectivity was enhanced in brain of children with ADHD. 3. The novel MSE method may be a new index to identify and understand the neural mechanism of ADHD.
Abstract The multiscale entropy (MSE) is a novel method for quantifying the intrinsic dynamical complexity of physiological systems over several scales. To evaluate this method as a promising way to explore the neural mechanisms in ADHD, we calculated the MSE in EEG activity during the designed task. EEG data were collected from 13 outpatient boys with a confirmed diagnosis of ADHD and 13 ageand gender-matched normal control children during their doing multi-source interference task (MSIT). We estimated the MSE by calculating the sample entropy values of delta, theta, alpha and beta frequency bands over twenty time scales using coarse-grained procedure. The results showed increased complexity of EEG data in delta and theta frequency bands and decreased complexity in alpha frequency bands in ADHD children. The findings of this study revealed aberrant neural connectivity of kids with ADHD during interference task. The results showed that MSE method may be a new index to identify and understand the neural mechanism of ADHD.
Keywords: ADHD, EEG, Complexity, Multiscale entropy, Connectivity
1. Introduction Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder with cardinal signs of inattention, impulsivity, and hyperactivity. About 5% ~ 8% of children are estimated to meet the diagnosis of ADHD worldwide (Bush, 2010). And ADHD is 2.3 times more
common in boys than girls (Bauermeister et al., 2007). ADHD children are at high risk of having learning disability and delinquency problems, which is a huge burden to the society, family and the children (Konrad and Eickhoff, 2010). Various sources have been used to illustrate that ADHD is associated with an atypical neurodevelopmental process, including neuroimaging, neuropsychological, genetics and neurochemical (Bush, 2010). Structural magnetic resonance imaging (sMRI) studies in ADHD children have observed abnormalities in late developing fronto-striatal, fronto-temporo-parietal and fronto-cerebellar network (Cubillo et al., 2012). Similar to that, functional magnetic resonance imaging (fMRI) studies have provided plenty of evidences on prefrontal cortex dysfunction (Smith et al., 2008), dorsal anterior midcingulate cortex (daMCC) abnormal (Bush, 2008), parietal cortex underactived (Tamm et al., 2006), cerebellum dysfunction (Durston et al., 2007), and abnormal activation of stratum (Scheres et al., 2007). These findings offered ample demonstrations that ADHD is related to disconnection and dysfunctions of brains. The human brain is a complex nonlinear system. Its activity exhibits complex fluctuations, both in spatial and temporal (Takahashi et al., 2009). Electroencephalogram (EEG) is the recording of electrical activity of the brain along the scalp, and it is believed to have the finest temporal resolution. With its easy operation and low cost, EEG is particularly suitable for investigating inherently complex biological signals arising from brain systems. Earlier EEG studies revealed the differences between children with ADHD and normal controls. Willis et al demonstrated theta activity was increased while alpha and beta activities were decreased in ADHD (Willis and Weiler, 2005). Meanwhile, the increased theta/beta power ratio (TBR) was reported as a meaningful EEG feature of ADHD (Barry et al., 2003). However, Arns et al concluded from their meta-analysis that the TBR cannot be considered as a reliable diagnostic measure for ADHD (Arns et al., 2013). The inconsistent research conclusions might come from the power analysis’s internal deficit of being influenced by noise from outside and inside body easily, i.e. EOG, MEG, the machine itself and so on. Therefore, only power analysis can hardly reveal the inherent neural mechanism of ADHD. Over the past decades, novel nonlinear approached based on entropy have been widely used to measure the complexity of physiological signal, including EEG signal (Fernández et al., 2013), electrocardiographic (ECG) (Šliupaitė et al., 2015) and biological codes (Costa et al., 2005).
Approximate entropy calculates the probability of new mode emerging in the signal along with the changing to evaluate the system complexity (Pincus et al., 1992, Richman et al., 2000). The sample entropy is more accurate to measure the probability of newly emerged mode, which is an advanced method of approximate entropy algorithm (Pincus et al., 1992). These traditional methods determine the probability of finding specific patterns or resemblance in a time series, thereby examine the irregularity or predictability in only one time series (Mizuno et al., 2010). Although widely used, the two methods are limited in their scope to only short-range temporal dynamics (Costa et al., 2005). Costa et al proposed Multiscale entropy (MSE) as an extension of sample entropy, quantifying the complexity of physiological signal by measuring the entropy across multiple time scales by coarsegraining procedure (Costa et al., 2005, Costa et al., 2002). This extension to larger temporal scales may reflect information of long-range temporal dynamics (Mizuno et al., 2010). The atypical MSE pattern may reflect disease condition of the brain and provide useful information into the network controlling mechanisms underlying physiological dynamics (Costa et al., 2002). Due to those advantages above, MSE method has been widely used on studies of mental disorders. Tomoyuki et al demonstrated the complexity increased at larger scales factors in Alzheimer’s but significantly reduced over smaller scales (Mizuno et al., 2010). Autism disease has been associated with significantly decreased MSE complexity (Catarino et al., 2011). And MSE study of drug-naive schizophrenia found elevated complexity at higher time scales in centro-temporal (Takahashi et al., 2010). It has been successfully employed as a useful biomarker for early detection of risk for autism spectrum disorder abnormalities in infants’ cognitive development (Catarino et al., 2011, Bosl et al., 2011). And also, atypical cortical dynamics in schizophrenia was found to be easily characterized and understand by MSE analytic method. (Takahashi et al., 2010). To characterize the complexity pattern of brain function of ADHD and Control, we tested the subjects’ attention related performance using multi-source interference task (MSIT) (Bush et al., 2003). The multi-source interference task (MSIT) combines multiple dimensions of cognitive interference in a single task (i.e., Stroop tasks, Eriksen Flanker-type tasks, and Simon effect task variants) (Bush and Shin, 2006). It is known as an effective way to active the network of brain regions involving attention and cognitive control with decision-making, target detection, response selection, stimulus/response competition and so on (Bush and Shin, 2006). Various studies using MSIT have been reported in patients with ADHD (Matsubara et al., 2014, Nakashima et al., 2014);
for instance, children with ADHD showed abnormal prefrontal activation in response to multiple interference control (Nakashima et al., 2014). In the present study, first, our aim was to investigate the MSE pattern of EEG in a population of ADHD children and healthy control subjects. We hypothesize that ADHD group will present significantly changed MSE pattern compared to control group. Second, we conduct a traditional power analysis and try to explore relevancy between MSE and power spectrum at different frequencies. To test this hypothesis, EEG was used to calculate the complexity and power spectrum of children with ADHD and normal Control during the MSIT performance.
2. Materials and methods 2.1. Participants EEG data were collected from 13 outpatient boys with a confirmed diagnosis of ADHD (age range, 6-13years; mean age, 8.5 ± 3.17 years; mean IQ score, 93 ± 16.8) and 13 age- and gendermatched healthy control children (age range, 6-13years; mean age, 7.9 ± 1.98 years; mean IQ score, 98 ± 18.3). All the participants are right handed. According to the Declaration of Helsinki, all parents of participants were informed with written consent. The study was approved by ethics committees of Xi’an Jiaotong Unversity School of Medicine. All participants were given full-scale and verbal IQ test with scores > 80 (Wechsler intelligence Scale for Children, Fourth Edition). All the 26 parents filled translated versions of the Conner’s Parent Rating Scales-Revised and the ADHD Rating Scale-VI for current and childhood-onset ADHD symptoms. The ADHD Rating Scale-IV is linked directly to the diagnostic criteria for ADHD of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) (Dupaul et al., 2006), which has been wildly used as a reliable method to diagnose ADHD (Konrad et al., 2012, Uddin et al., 2008). ADHD group had higher DSM-IV scores (mean, 13.23 ± 3.22) than that of typical control group (mean, 5.23 ± 2.09) (P < 0.01).Subjects with higher DSM-IV scores suffer from more serious ADHD. The ADHD children were include if they had (1)
6 of 9 DSM-IV symptoms of inattention as well as
6
of 9 symptoms of hyperactivity/impulsivity; (2) no record of brain injuries, medical history, and other neuropsychiatric disorders; and (3) no signs of cognitive deficit, learning disabilities, as well as communication problems. Exclusion criteria for normal control group were the same, and also
they do not have ADHD disease or have individuals in their families. 2.2. Task Description The multi-source interference task (MSIT) is believed to have the ability activating the cingulofroto-paretal (CFP) network (Stroop, 1935). The MSIT task is designed using system STIM2 provided by Neuroscan company. The operation is practicable for the both groups and can be finished within 15 minutes. Subjects were given a button with two keys, of which the left one and right one respectively represented the number 1 and 2. In this study, the interference trails were included only. In the trails, there are 3 numbers on the screen combined by 1 and 2, the total set of the number combinations are 112, 212, 221 and 211. During the interference task, subjects were instructed to use the index and middle finger of right hand to respond to the target number among the three, the target numbers were never placed congruently with their button box (Stroop effect (Eriksen and Eriksen, 1974)), and the Flanker stimuli (Bush and Shin., 2006)were always potential target. One example of the stimuli in a single trail is shown in Figure 1. Before the experiment, each subject underwent some adaptive exercises. All the interference trials would appear at the center of the screen every 3 seconds with 15 second of fixation (a white dot in the center of the screen). During fixation, subjects completed 80 random trials. 2.3. EEG data collection EEG data was acquired from 19 electrodes using the International ten-twenty system. The scalp locations are: Fp1, Fp2, F3, F4, F7, F8, Fz, T3, T4, T7, T8, C3, Cz, P3, P4, Pz, O1 and O2 (Neuroscan; Compumedics Limited, Charlotte, NC). The impedance of all sits was < 5 kΩ. Reference electrode was at the left mastoid, and the ground electrode was between Fpz and Fz. Eyemovement were monitored vertically and horizontally separately, the vertical electro-oculogram was recorded from electrode attached above and below the left eye, and a horizontal electro-oculogram was recorded from the outer canthi of both eyes. EEG was sampled with 32 bits of accuracy. The analog/digital sampling rate was 1,000 Hz. All subjects were seated inside an acoustically and electromagnetically shielded room. They were asked to do their best to keep body steady during the experiments. 2.4. EEG recording All subjects were able to react to the task in about 2 seconds (average reaction times of the two groups were 1,703 ± 522 and 1,162 ± 462ms), so we defined the period from target stimulus onset
to 2 seconds as the response period. Only correct trials were segmented, and each subject had 10 epochs (total, 20 seconds; 20,000 samples) of artifact free data (no ocular and muscular artifacts). Filter analysis was performed on every EEG data segment, and EEG components of the following 4 frequency bands were obtained: delta (0.5 – 3 Hz), theta (4 – 7 Hz), alpha (8 – 13 Hz), and beta (13 – 30 Hz). For each participant, MSE as well as relative power was calculated on each frequency band. 2.5. Multiscale entropy (MSE) The MSE method was developed by Costa et al in 2005, and it has been used to quantifies the complexity of signal by calculating the sample entropy (SampEn) over multiple temporal scales which was realized by coarse-grained procedure (Costa et al, 2005, Costa et al., 2002). Sample entropy was proposed by Richman and Moorman in 2000 (Richman and Moorman, 2000). The SampEn measures the regularity of time series, it is a modified version of Approximate Entropy (ApEn) (Richman et al., 2004). Given an one-dimensional EEG discrete time series x
,…
and the scale factor , the
time series is calculated in to consecutive and nonoverlapping time series procedure:
=
∑
,1
j
by coarse-grained
N/ τ . Scale one is simply the original time series
representing the short-range temporal scale and larger scale factor represents longer temporal scales. And then calculates the SampEn of each series
. SampEn was defined as the negative of the
logarithmic condition probability that two similar sequences of m consecutive data points will 1 (Catarino et al., 2011, Richman and Moorman, 2000).
remain similar at the next point SampEn (m, r, N) = -ln ,i
r /
, where
j}/{number of all probable pairs = N
distance between vectors
and
m
= {number of pairs (i, j) with 1 / N
m }, where
. Therein, m is dimension of vectors
and
denotes the and r is
the tolerable distance between the two vectors and N represents the length of the time series. Studies by Richman in 2004 have proved that dimension m = 1 or m = 2 and tolerable distance r = 0.1~0.25 times the S.D. (Standard deviation) of the time series have the best statistical validity (Richman and Moorman, 2000). In this study, we used the parameter m = 2, r = 0.15×S.D., N = 20000 data points and we set the scale factor as 20. After the coarse-grained procedure, we still have 20000/20 = 1000 data points, which has been proved enough to obtain reliable estimation of sample entropy value
(Richman and Moorman, 2000). 2.6. Power analysis Besides MSE analysis, a power analysis was performed as a conventional EEG measurement so as to explore the relationship between the system complexity and EEG power spectrum. We used the pwelch function in MATLAB (version 7.12). The data were separated into nine equal lengths, each with 50% overlap. Four standard bands, delta (0.5 – 3 Hz), theta (4 – 7 Hz), alpha (8 – 13 Hz), and beta (13 – 30 Hz) were included in the power analysis. For statistical analysis, the relative power change was then (power in each frequency divided by total power across all frequency bands) calculated. 2.7. Statistical analysis The statistical analysis was carried out by SPSS 13.0 for Windows. Kolmogorov-Smirnov normality test was used to confirm the distribution normality of MSE value for every group and every channel. Mauchly’s Test of Sphericity was used to test whether the data are appropriately for the ANOVA or a data correction is needed, then the Greenhouse-Geisser adjustment was applied to the degrees of freedom for all analysis. The alpha significance value was set as
0.05 for all
analysis (i.e., both for repeated ANOVA and post-hoc t-tests). To test for differences in behavioral results, a two - groups (ADHD vs. Control)
1 condition
(MSIT) analysis of variance (ANOVA) was done for response time. For MSE analysis, we applied 4-way repeated-measures analysis of variance (ANOVA) to test the group differences with group (ADHD vs. Control) as the between-subjects factor, and the scale factor (sf: τ=20), frequency bands (delta, theta, alpha, beta) as well as channel (FP1, FP2, F7, F3, FZ, F4, F8, T7, T8, FT7, FT8, CZ, C3, C4, PZ, P3, P4, O1, O2) as the within-subjects factors. Next for every frequency band, we tested the group difference with group (ADHD vs. Control) as the between-subjects factor, and the SF (τ = 20), and channel (FP1, FP2, F7, F3, FZ, F4, F8, T7, T8, FT7, FT8, CZ, C3, C4, PZ, P3, P4, O1, O2) as the within-subjects factors. And then for every electrode in every frequency band, repeated measures ANOVA with group (ADHD vs. Control) as the between-subject factor and sf (τ = 20) as the with-in subject factor. Subsequently, a repeated measures analysis of variance (ANOVA) was performed to test for group differences in relative power analysis. The group (ADHD and control) was the betweensubjects factor, frequency bands (delta, theta, alpha, and beta) and channels were the within-subjects
factors. For every band under every channel, 2-way repeated ANOVA was performed with group (ADHD vs. Control) as between-subject factor, and frequency band as within-subject factors. In case of the significant group-by-band interaction, then independent t-test was used to find group difference for each frequency band in each channel. To investigate the relevancy of MSE and power spectra, we presented Spearman’s rank order correlations between relative power and MSE values for the two groups.
3. Results 3.1 Behavioral performance The mean reaction times for the ADHD and the Control groups were 1,703 ± 522 ms and 1,162 ± 462 ms. During the interference trails, reaction time of the ADHD were significantly longer than that was for Controls (F1,18 = 6.019, p = 0.025).
3.2. Power analysis The results revealed that during the MSIT tasks, the brain regions in frontal and central region exhibited relatively lower power in delta frequency band but showed relatively higher power in frontal, temporal and central region in theta and alpha frequency bands. The results showed no main significant effects for group (F1, 342 = 3.584, p = 0.059) and channel (F18, 342 = 0.717, p = 0.794). Also, no significant group-by-channel (F18, 342 = 1.112, p = 0.338) and group-by-band-by-channel (F32.862, 624.383 = 1.241, p = 0.194) interactions were found. However, there was a significant groupby-fb (frequency band) interaction (F1.826, 624.383 = 35.336, p < 0.001). We used 2-way repeated ANOVA for investigating the group-by-fb (frequency band) interaction (Table 1). This interaction was explored further using independent t-test to find group difference separately for each frequency band and channel. The results are shown in Figure 2, indicating a significant group-by-band interaction in Fp1 (F1.788, 32.192 = 4.611; p = 0.020), F3 (F2.219, 39.948
= 4.741; p = 0.012), Fz (F1.463, 26.340 =4.586; p = 0.046), FT7 (F1.632, 29.383 = 8.905; p = 0.008),
C3 (F1.370, 24.653 = 9.288; p = 0.003), C4 (F1.553, 27.957 = 6.806; p = 0.018), and Pz (F1.785, 32.136 = 6.970; p = 0.004). Post-hoc independent t-test showed significant decreases in delta and beta and increases in theta and alpha frequency bands in ADHD, compared to Control group (Figure 2).
3.3. MSE analysis The MSE results showed no main effect of group difference (F1, 1368 = 3.152, p = 0.076) in 4way ANOVA. However when testing the group difference in each frequency band, the result showed there exist main effect of group difference in delta (F1, 342 = 60.703, p < 0.001) and alpha (F1, 342 = 29.552, p < 0.001) frequency bands that relative to Control group, the mean sample entropy (SampEn) values of ADHD group was increased in delta frequency band (see in Figure3) and decreased in alpha frequency band (see in Figure 4) as the scale factor increases, but the differential in theta and beta is not that significant. Figure 3 and figure 4 reflect group differences of SampEn values in delta and alpha frequency bands in each channel for increasing scale factor. The MSE results showed no main effect of group difference (F1, 1368 = 3.152, p = 0.076) in 4way ANOVA. However when testing the group difference in each frequency band, the result showed there exist main effect of group difference in delta (F1, 342 = 60.703, p < 0.001) and alpha (F1, 342 = 29.552, p < 0.001) frequency bands that relative to Control group, the mean sample entropy (SampEn) values of ADHD group was increased in delta frequency band (see in Figure3) and decreased in alpha frequency band (see in Figure 4) as the scale factor increases, but the differential in theta and beta is not that significant. Figure 3 and figure 4 reflect group differences of SampEn values in delta and alpha frequency bands in each channel for increasing scale factor. 3.4 Relationship between power analysis and MSE In order to explore the relationship between MSE and the power analysis, we divided the scale factors into four components as Tomoyuki did in AD research (i.e. sf 1 – 5, sf 6 – 10, sf 11 – 15, and sf 16 – 20). According to the statistical analysis, the result showed no significant relationship between MSE and power analysis. Table 4 and 5 demonstrate the example (Fp1) of correlation result between MSE value and the relative power for ADHD and normal control subjects.
4. Discussion The presented study found that the reaction time were significantly longer for ADHD children than that of Control group, thus we inferred that the behavioral result might related to inattention, which is the core clinical symptom of ADHD. In our power analysis, we found significantly increased relative theta and alpha power in frontal,
temporal and central region of brain as well as decreased relative delta power in frontal and central region during MIST task. These findings are in line with many previous power analysis of EEG in ADHD (Willis and Weiler, 2005, Clarke et al., 2001, Fernández et al., 2009) that the ADHD children display an atypical neural pattern during the cognitive information processing procedure. Although power spectrum analysis, i.e. increased theta/beta power ratio (TBR), has been widely used as a biomarker of ADHD, Arns proposed in the meta-analysis study that TBR cannot be considered a reliable diagnostic measure of ADHD (Arns et al., 2012). This inconstancy may lead to misdiagnose of ADHD disease. In the present study, we used the novel MSE analysis to investigate complexity of EEG signal in ADHD children during their MSIT task. MSE method has already been demonstrated as a sufficient way to quantify the complexity of dynamical signals over several time scales (Bosl et al., 2011), and can be used as a reliable index of some mental disorders (Mizuno et al., 2010, Takahashi et al., 2010). So we used the MSE method to explore the neurophysiological evidence for the abnormal neural pattern of activated condition, and thus identifying whether it can be a new nonlinear dynamic index for analyzing EEG signal in children with ADHD. Although early studies reported relatively decreased EEG complexity of ADHD subjects during tasks (Willis and Weiler, 2005, Fernández et al., 2009), our findings of EEG complexity of ADHD were different. In our MSE analysis study, we measured the complexity of EEG over multiple time scales in four frequency bands and we found that the complexity was significantly elevated in delta frequency band and decreased in alpha frequency band at larger scale factors in ADHD compared to normal subjects. Abnormal complexity in neurophysiologic signals might reflect aberrant neural connectivity pattern in mental disorders (Takahashi, 2013, Sporns et al., 2000), thus we hypothesize that our MSE findings of changed EEG complexity in ADHD may reflect the aberrant neural connectivity pattern. Empirical and computational studies suggested that changes in functional connectivity may underlie specific perceptual and cognitive state and involve the integration of information across specialized areas of the brain, and the interplay of functional areas and integration can be quantified particular in neural complexity (Sporns et al., 2000), so they inferred that the disconnection of brain is associated with an increase of complexity (Takahashi, 2013, Sporns et al., 2000). In the current study, we found that the complexity of ADHD was significantly increased in delta frequency band and decreased in alpha band at lager scale factors
than normal children, which indicated that the connectivity pattern was significantly changed in ADHD children. Previous research revealed that specific frequency bands reflect different functional state of brain activities (Takahashi, 2013). Higher frequency band and lower frequency band originate from small neuronal populations and larger populations (Takahashi, 2013, Schnitzler and Gross, 2005), which means the complexity pattern of lower frequency band may reflect global character of the brain in ADHD while the in higher frequency may reflect the local connectivity feature of brain. In our study, we found different change pattern of complexity in delta and alpha frequency bands. Based on the conclusions above, we suspect that the global connectivity was weaker and the local connectivity was enhanced in brain of children with ADHD. Supporting this is the work by Tomasi et al, they quantified short-and long-range functional density in the brain with ADHD and normal controls and showed a pattern of hyper short-range connectivity and hypo long-range connectivity of brain in children with ADHD (Tomasi and Volkow, 2012). Also the research about small-world properties of brain network in ADHD concluded that the global characteristic became weaken and the local feature was enhanced in ADHD (Liu et al., 2014, Wang et al., 2009, Wang et al., 2013). These previous neuroimaging results support our hypotheses that altered MSE pattern of EEG reflect the atypical neural connectivity in ADHD, and ADHD is possibly in association with reduced long-range connectivity and increased local brain connectivity. Comparing MSE and power analysis, we found no significant relevancy between relative power and MSE, that is to say, there is no relationship between the complexity measures and power spectrum. This is supported by the MSE study of autism that altered complexity pattern are not a reflection of changes in power spectrum (Catarino et al., 2011).
5. Conclusions In summary, in the present study, we explored the MSE method to analyze the EEG signal of children with and without ADHD during their MIST task. The results of the present study reveal a changed complexity pattern for ADHD children compared to normal controls in delta and alpha frequency bands. In addition, there was no significant correlation between complexity and power analysis for the two group on MSIT task. These findings reveal that the brain connectivity was
changed in ADHD, suggesting that ADHD may be related to long-range disconnection as well as enhanced local connection. Therefore, the MSE method may provide a new view to characterize and analyze aberrant brain network in ADHD and a new aspect to understand neurophysiological mechanism of ADHD. However, due to the small sample size, findings presented in the study might be limited. Further studies should be carried out to confirm that our results are not a particular case. The potential effect of gender and severity of the disease are needed to be taken into consideration. In future studies, the resting state are necessarily needed to clarify this outstanding of this method. Also, there’s one issue need to notice that whether the MSE method would be sensitive to the different neuronal activities due to different tasks (Catarino et al., 2011). So future study should involve different difficulty levels and forms of tasks to investigate the sensitivity of the MSE method to the tasks. If future studies confirm that complexity pattern measured by MSE can indeed be a reliable clinical EEG index of ADHD, this might be a new way to diagnose ADHD.
Funding The author(s) received no financial support for the research, authorship, and/or publication of this article. Declaration of Conflicting Interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
6. Acknowledgement The study was supported by grant 61503295 from Youth Project of National Natural Science Foundation of China and grant 2015JQ8318 from Natural Science Foundation of Shaanxi Province. We thank the Xi’an Children’s Hospital for their help of data collection and our colleagues who provided help to this article.
Reference: Arns, M., Conners, C. K., & Kraemer, H. C. (2012). A decade of EEG theta/beta ratio research in ADHD: a meta-analysis. Journal of attention disorders, 1087054712460087. Barry, R. J., Clarke, A. R., & Johnstone, S. J. (2003). A review of electrophysiology in attentiondeficit/hyperactivity disorder: I. Qualitative and quantitative electroencephalography. Clinical neurophysiology, 114(2), 171-183. Bauermeister, J. J., Shrout, P. E., Chávez, L., Rubio‐Stipec, M., Ramírez, R., Padilla, L. & Canino, G. (2007). ADHD and gender: are risks and sequela of ADHD the same for boys and girls? Journal of Child Psychology and Psychiatry, 48(8), 831-839. Bosl, W., Tierney, A., Tager-Flusberg, H., & Nelson, C. (2011). EEG complexity as a biomarker for autism spectrum disorder risk. BMC medicine, 9(1), 1. Bush, G., Shin, L. M., Holmes, J., Rosen, B. R., & Vogt, B. A. (2003). The Multi-Source Interference Task: validation study with fMRI in individual subjects. Molecular psychiatry, 8(1), 60-70. Bush, G., & Shin, L. M. (2006). The Multi-Source Interference Task: an fMRI task that reliably activates the cingulo-frontal-parietal cognitive/attention network. NATURE PROTOCOLSELECTRONIC EDITION-, 1(1), 308. Bush, G. (2009). Dorsal anterior midcingulate cortex: roles in normal cognition and disruption in attention-deficit/hyperactivity disorder (pp. 207-218). Oxford University Press, NY, USA. Bush,
G.
(2010).
Attention-deficit/hyperactivity
disorder
and
attention
networks.
Neuropsychopharmacology, 35(1), 278-300. Catarino, A., Churches, O., Baron-Cohen, S., Andrade, A., & Ring, H. (2011). Atypical EEG complexity in autism spectrum conditions: a multiscale entropy analysis. Clinical neurophysiology, 122(12), 2375-2383. Clarke, A. R., Barry, R. J., McCarthy, R., & Selikowitz, M. (2001). EEG-defined subtypes of children with attention-deficit/hyperactivity disorder.Clinical Neurophysiology, 112(11), 20982105. Costa, M., Goldberger, A. L., & Peng, C. K. (2002). Multiscale entropy analysis of complex physiologic time series. Physical review letters, 89(6), 068102. Costa, M., Goldberger, A. L., & Peng, C. K. (2005). Multiscale entropy analysis of biological signals.
Physical review E, 71(2), 021906. Cubillo, A., Halari, R., Smith, A., Taylor, E., & Rubia, K. (2012). A review of fronto-striatal and fronto-cortical brain abnormalities in children and adults with Attention Deficit Hyperactivity Disorder (ADHD) and new evidence for dysfunction in adults with ADHD during motivation and attention. Cortex, 48(2), 194-215. Dupaul, G. J., Power, T. J., Anastopoulos, A. D., & Reid, R., (2006). ADHD Rating Scale-IV: Checklists, norms, and clinical interpretation. Journal of Psychoeducational Assessment 24 (2), 172178. Durston, S., Davidson, M. C., Mulder, M. J., Spicer, J. A., Galvan, A., Tottenham, N., & Casey, B. J. (2007). Neural and behavioral correlates of expectancy violations in attention ‐ deficit hyperactivity disorder. Journal of Child Psychology and Psychiatry, 48(9), 881-889. Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the identification of a target letter in a nonsearch task. Perception & psychophysics, 16(1), 143-149. Fernández, A., Quintero, J., Hornero, R., Zuluaga, P., Navas, M., Gómez, C. & Ortiz, T. (2009). Complexity analysis of spontaneous brain activity in attention-deficit/hyperactivity disorder: diagnostic implications. Biological psychiatry, 65(7), 571-577. Fernández, A., Gómez, C., Hornero, R., & López-Ibor, J. J. (2013). Complexity and schizophrenia. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 45, 267-276. Konrad, A., Dielentheis, T. F., El Masri, D., Dellani, P. R., Stoeter, P., Vucurevic, G., & Winterer, G. (2012). White matter abnormalities and their impact on attentional performance in adult attentiondeficit/hyperactivity disorder. European archives of psychiatry and clinical neuroscience, 262(4), 351-360. Konrad, K., & Eickhoff, S. B. (2010). Is the ADHD brain wired differently? A review on structural and functional connectivity in attention deficit hyperactivity disorder. Human brain mapping, 31(6), 904-916. Lake, D. E., Richman, J. S., Griffin, M. P., & Moorman, J. R. (2002). Sample entropy analysis of neonatal heart rate variability. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 283(3), R789-R797. Liu, T., Chen, Y., Lin, P., & Wang, J. (2014). Small-world brain functional networks in children with attention-deficit/hyperactivity disorder revealed by EEG synchrony. Clinical EEG and neuroscience,
1550059414523959. Matsubara, T., Matsuo, K., Nakashima, M., Nakano, M., Harada, K., Watanuki, T. & Watanabe, Y. (2014). Prefrontal activation in response to emotional words in patients with bipolar disorder and major depressive disorder. Neuroimage, 85, 489-497. Mizuno, T., Takahashi, T., Cho, R. Y., Kikuchi, M., Murata, T., Takahashi, K., & Wada, Y. (2010). Assessment of EEG dynamical complexity in Alzheimer’s disease using multiscale entropy. Clinical Neurophysiology, 121(9), 1438-1446. Nakashima, M., Matsuo, K., Hashimoto, A., Nakano, M., Fujii, Y., Matsushige, T. & Sugiyama, S. (2014). Prefrontal Abnormality in Children with ADHD during Cognitive Interference Control: a Functional NIRS Study. The bulletin of the Yamaguchi Medical School, 61(3), 37-47. Nazari, M. A., Wallois, F., Aarabi, A., & Berquin, P. (2011). Dynamic changes in quantitative electroencephalogram during continuous performance test in children with attentiondeficit/hyperactivity disorder. International journal of psychophysiology, 81(3), 230-236. Pincus, S. M., & Huang, W. M. (1992). Approximate entropy: statistical properties and applications. Communications in Statistics-Theory and Methods, 21(11), 3061-3077. Richman, J. S., & Moorman, J. R. (2000). Physiological time-series analysis using approximate entropy and sample entropy. American Journal of Physiology-Heart and Circulatory Physiology, 278(6), H2039-H2049. Richman, J. S., Lake, D. E., & Moorman, J. R., (2004). Sample entropy. Methods in Enzymology 384 (1), 172-184. Scheres, A., Milham, M. P., Knutson, B., & Castellanos, F. X. (2007). Ventral striatal hyporesponsiveness during reward anticipation in attention-deficit/hyperactivity disorder. Biological psychiatry, 61(5), 720-724. Schnitzler, A., & Gross, J. (2005). Normal and pathological oscillatory communication in the brain. Nature reviews neuroscience, 6(4), 285-296. Šliupaitė, A., Navickas, Z., & Vainoras, A. (2015). Evaluation of Complexity of ECG Parameters Using Sample Entropy and Hankel Matrix. Elektronika ir Elektrotechnika, 92(4), 107-110. Smith, A. B., Taylor, E., Brammer, M., Halari, R., & Rubia, K. (2008). Reduced activation in right lateral prefrontal cortex and anterior cingulate gyrus in medication ‐ naïve adolescents with attention deficit hyperactivity disorder during time discrimination. Journal of Child Psychology and
Psychiatry, 49(9), 977-985. Sporns, O., Tononi, G., & Edelman, G. M. (2000). Connectivity and complexity: the relationship between neuroanatomy and brain dynamics. Neural Networks, 13(8), 909-922. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of experimental psychology, 18(6), 643. Takahashi, T., Cho, R. Y., Murata, T., Mizuno, T., Kikuchi, M., Mizukami, K. & Wada, Y. (2009). Age-related variation in EEG complexity to photic stimulation: a multiscale entropy analysis. Clinical Neurophysiology, 120(3), 476-483. Takahashi, T., Cho, R. Y., Mizuno, T., Kikuchi, M., Murata, T., Takahashi, K., & Wada, Y. (2010). Antipsychotics reverse abnormal EEG complexity in drug-naive schizophrenia: a multiscale entropy analysis. Neuroimage, 51(1), 173-182. Takahashi, T. (2013). Complexity of spontaneous brain activity in mental disorders. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 45, 258-266. Tamm, L., Menon, V., & Reiss, A. L. (2006). Parietal attentional system aberrations during target detection in adolescents with attention deficit hyperactivity disorder: event-related fMRI evidence. American Journal of Psychiatry. Tomasi, D., & Volkow, N. D. (2012). Abnormal functional connectivity in children with attentiondeficit/hyperactivity disorder. Biological psychiatry, 71(5), 443-450. Uddin, L. Q., Kelly, A. C., Biswal, B. B., Margulies, D. S., Shehzad, Z., Shaw, D., ... & Milham, M. P. (2008). Network homogeneity reveals decreased integrity of default-mode network in ADHD. Journal of neuroscience methods, 169(1), 249-254. Willis, W. G., & Weiler, M. D. (2005). Neural substrates of childhood attention-deficit/hyperactivity disorder: Electroencephalographic and magnetic resonance imaging evidence. Developmental Neuropsychology, 27(1), 135-182. Wang, L., Zhu, C., He, Y., Zang, Y., Cao, Q., Zhang, H. & Wang, Y. (2009). Altered small‐world brain functional networks in children with attention‐deficit/hyperactivity disorder. Human brain mapping, 30(2), 638-649. Wang, P., Zhu, D., Li, X., Chen, H., Jiang, X., Sun, L. & Wang, Y. (2013, April). Identifying functional connectomics abnormality in attention deficit hyperactivity disorder. In Biomedical Imaging (ISBI), 2013 IEEE 10th International Symposium on (pp. 544-547). IEEE.
Figure1. Illustration of the MSIT. During the interference trails, subjects were instructed to press the button to figure out the number that differs from the other 2 numbers (Flanker stimuli), but the target never matched the button location (Stroop effect). In this example, the number 2 is different from the 2 numbers, so the correct answer would be to press button 2.
Figure 2. Relative power analysis of ADHD children and normal control subjects. Post-hoc analysis between ADHD and Control: *p 0.05
Figure 3. Multiscale entropy (MSE) analysis in delta of 13 ADHD (blue line) children and normal control (red line) for each electrode. Each panel presents the average of MSE values of each group. *p < 0.05
Figure 4. Multiscale entropy (MSE) analysis in alpha of 13 ADHD (blue line) children and normal control (red line) for each electrode. Each panel presents the average of MSE values of each group. *p < 0.05
Table 1 Group-by-fb interaction significance values for each channel in power analysis. Group-by-band
Group-by-band interaction
Channel Fp1
F1.788, 32.192 = 4.611; p = 0.020*
Fp2
F1.858, 33.446 = 1.212; p = 0.308
F7
F1.624, 29.227 = 2.841; p = 0.084
F3
F2.219, 39.948 = 4.741; p = 0.012*
Fz
F1.463, 26.340 =4.586; p = 0.044*
F4
F2.015, 36.271 = 1.134; p = 0.333
F8
F1.523, 27.410 = 1.600; p = 0.221
T7
F1.420, 25.567 = 2.168; p = 0.147
T8
F1.559, 28.061 = 1.168; p = 0.314
FT7
F1.632, 29.383 = 3.475; p = 0.050*
FT8
F1.512, 27.214 = 1.561; p = 0.228
Cz
F1.728, 31.102 = 8.771; p = 0.004
C3
F1.370, 24.653 = 9.288; p = 0.003*
C4
F1.553, 27.957 = 6.806; p = 0.018*
Pz
F1.785, 32.136 = 6.970; p = 0.004*
P3
F1.541, 27.740 = 2.005; p = 0.162
P4
F1.629, 29.320 = 0.528; p = 0.559
O1
F1.305, 23.482 = 2.570; p = 0.115
O2
F1.290, 23.217 = 1.613; p = 0.221
*These channels showed a significance (p .05) group-by-fb interaction.
Table 2 Group-by-sf interaction significance values for each channel in delta frequency band. Group-by-sf
Group-by-sf interaction
channel Fp1
F1.313, 23.634 = 9.545; p = 0.006*
Fp2
F1.395,25.115 = 4.341; p = 0.052
F7
F1.263, 22.736 = 6.980; p = 0.017*
F3
F1.249, 22.482 = 4.155; p = 0.056
Fz
F1.771, 31,872 = 7.196; p = 0.015*
F4
F1.750, 31.502 = 1.466; p = 0.242
F8
F1.503, 27.048 = 0.580; p = 0.456
T7
F1.183, 21,288 = 9.783; p = 0.006*
T8
F1.713, 30.831 = 0.855; p = 0.368
FT7
F1.218, 21.928 = 6.471; p = 0.020*
FT8
F1.501, 27.011 = 0.583; p = 0.455
Cz
F1.449, 26.075 = 1.849; p = 0.191
C3
F1.577, 28.378 = 9.954; p = 0.005*
C4
F1.866, 33.946 = 17.547; p = 0.001*
Pz
F1.910, 34,377 = 2.480; p = 0.133
P3
F1.647, 29.655 = 1.659; p = 0.214
P4
F1.482, 26.684 = 0.002; p = 0.962
O1
F1.795, 32.317 = 0.125; p = 0.728
O2
F1.711, 30.801 = 0.049; p = 0.827
*These channels showed a significant (p
0.05) group-by-sf interaction.
Table 3 Group-by-sf interaction significance values for each channel in alpha frequency band. Group-by-sf
Group-by-sf interaction
channel Fp1
F1.466, 26.385 = 3.879; p = 0.064
Fp2
F1.278, 23.06 = 0.270; p = 0.610
F7
F2.384 42.906 = 0.373; p = 0.549
F3
F1.572, 28.291 = 2.215; p = 0.154
Fz
F1.320, 23.769 = 7.780; p = 0.012*
F4
F1.222, 21.988 = 4.047; p = 0.059
F8
F1.236, 22.241 = 0.932; p = 0.347
T7
F1.913, 34.437 = 6.057; p = 0.024*
T8
F1.176, 21.164 = 0.803; p = 0.382
FT7
F2.165, 38.974 = 0.016; p = 0.900
FT8
F1.212, 21.814 = 1.130; p = 0.302
Cz
F1.138, 20.486 = 14.225; p = 0.001*
C3
F1.138, 20.486 = 6.522; p = 0.020*
C4
F1.083, 19.492 = 2.298; p = 0.147
Pz
F1.706, 19.373 = 8.516; p = 0.009*
P3
F1.203, 21,648 = 0.690; p = 0.417
P4
F1.151, 20.712 = 0.193; p = 0.665
O1
F1.196, 21.520 = 1.042; p = 0.321
O2
F1.310, 23.587 = 0.445; p = 0.513
*These channels showed a significant (p
0.05) group-by-sf interaction.
Table 4 Correlations between MSE and relative power among ADHD subjects for Fp1. Relative power
MSE sf: 1-5
MSE sf: 6-10
MSE sf: 11-15
MSE sf: 16-20
Delta
-0.159
-0.159
0.402
0.598
0.662
0.662
0.249
0.068
0.829
0.829
0.549
0.341
0.003
0.003
0.100
0.334
-0.037
-0.037
-0.146
-0.232
0.920
0.920
0.687
0.519
-0.744
-0.744
-0.695
-0.622
0.014
0.014
0.026
0.055
Theta
Alpha
Beta
The upper number in each represents r-values for correlations between relative frequency power value and averaged MSE value across scale factors, and the lower number represents the p-value
Table 5 Correlations between MSE and relative power among normal control subjects for Fp1. Relative power
MSE sf: 1-5
MSE sf: 6-10
MSE sf: 11-15
MSE sf: 16-20
Delta
-0.576
-0.552
-0.539
-0.636
0.082
0.098
0.108
0.048
0.709
0.685
0.758
0.770
0.022
0.029
0.011
0.009
0.842
0.867
0.830
0.867
0.002
0.001
0.003
0.001
0.333
0.321
0.285
0.382
0.347
0.365
0.425
0.276
Theta
Alpha
Beta
The upper number in each represents r-values for correlations between relative frequency power value and averaged MSE value across scale factors, and the lower number represents the p-value