Network controllability analysis of awake and asleep conditions in the brain Enhanced Publication

Network controllability analysis of awake and asleep conditions in the brain 大脑清醒和睡眠状态下的网络可控性分析 1 first-author 2 1 1 1 3 4 corresp Hao YAN, haoyan@xisu.edu.cn Hao YAN, haoyan@xisu.edu.cn haoyan@xisu.edu.cn https://orcid.org/0000-0002-1854-786X 西安外国语大学，人工智能与语言认知神经科学重点实验室，中国西安市，710121 西安外国语大学，人工智能与语言认知神经科学重点实验室，中国西安市，710121 Key Laboratory for Artificial Intelligence and Cognitive Neuroscience of Language, Xi’an International Studies University, Xi’an 710121, China Key Laboratory for Artificial Intelligence and Cognitive Neuroscience of Language, Xi’an International Studies University, Xi’an 710121, China 空军军医大学，西京医院神经外科，中国西安市，710032 空军军医大学，西京医院神经外科，中国西安市，710032 Department of Neurosurgery, Xijing Hosptial, Air Force Medical University, Xi’an 710032, China Department of Neurosurgery, Xijing Hosptial, Air Force Medical University, Xi’an 710032, China 西安电子科技大学，外国语学院，中国西安市，710121 西安电子科技大学，外国语学院，中国西安市，710121 Department of Linguistics, Xidian University, Xi’an 710121, China Department of Linguistics, Xidian University, Xi’an 710121, China 深圳市神经科学研究院，语言与大脑研究中心，中国深圳市，518107 深圳市神经科学研究院，语言与大脑研究中心，中国深圳市，518107 Center for Language and Brain, Shenzhen Institute of Neuroscience, Shenzhen 518107, China Center for Language and Brain, Shenzhen Institute of Neuroscience, Shenzhen 518107, China 睡眠和清醒之间的差异对人类的健康至关重要，清醒和睡眠之间的转换紊乱伴随脑部疾病，因此需要深入研究其具体特征。本研究引入网络可控性揭示大脑脑电活动中频率成分的功能特异性。具体来说，我们采用一个公开的颅内立体脑电图数据集。首先，记录受试者清醒和睡眠条件下的脑电信号，经过降噪、伪迹去除等预处理方法，通过带通滤波提取亚慢波（0.1~1 Hz）、δ（1~4 Hz）、θ（4~8 Hz）、α（8~13 Hz）、β（13~30 Hz）和γ（30~45 Hz）波段振荡。其次，利用锁相值（PLV）和不重叠滑动时间窗从时间窗脑电神经振荡中提取动态功能连通性。最后，在这些时变大脑网络上计算平均和模态网络的可控性。初步结果显示，清醒和睡眠状态下，不同频段脑电活动在额顶网络（FPN）、显著网络（SN）和默认模式网络（DMN）存在显著差异，即不同频率成分的脑电信号以不同网络控制策略参与大脑清醒和睡眠。网络可控性揭示了清醒和睡眠条件下的潜在大脑动力学，网络可控性和动态功能网络的结合为表征大脑清醒和睡眠阶段的区别提供了新的度量方法。 颅内立体脑电 相位锁定值 神经振荡 功能连接 网络可控性 The difference between sleep and wakefulness is critical for human health. Sleep takes up one third of our lives and remains one of the most mysterious conditions; it plays an important role in memory consolidation and health restoration. Distinct neural behaviors take place under awake and asleep conditions, according to neuroimaging studies. While disordered transitions between wakefulness and sleep accompany brain disease, further investigation of their specific characteristics is required. In this study, the difference is objectively quantified by means of network controllability. We propose a new pipeline using a public intracranial stereo-electroencephalography (stereo-EEG) dataset to unravel differences in the two conditions in terms of system neuroscience. Because intracranial stereo-EEG records neural oscillations covering large-scale cerebral areas, it offers the highest temporal resolution for recording neural behaviors. After EEG preprocessing, the EEG signals are band-passed into sub-slow (0.1‍–‍1 Hz), delta (1‍–‍4 Hz), theta (4‍–‍8 Hz), alpha (8‍–‍13 Hz), beta (13‍–‍30 Hz), and gamma (30‍–‍45 Hz) band oscillations. Then, dynamic functional connectivity is extracted from time-windowed EEG neural oscillations through phase-locking value (PLV) and non-overlapping sliding time windows. Next, average and modal network controllability are implemented on these time-varying brain networks. Based on this preliminary study, it appears that significant differences exist in the dorsolateral frontal-parietal network (FPN), salience network (SN), and default-mode network (DMN). The combination of network controllability and dynamic functional networks offers new insight for characterizing distinctions between awake and asleep stages in the brain. In other words, network controllability captures the underlying brain dynamics under both awake and asleep conditions. Sleep and wakefulness act as a double onion-like framework in the brain ( Ioannides, 2018 Ioannides, 2018 ). Entanglement of sleep and wakefulness is accompanied with neural pathology ( Ioannides, 2018 Ioannides, 2018 ; Sarasso et al., 2020 Sarasso et al., 2020 ; Andrillon et al., 2021 Andrillon et al., 2021 ). For example, one main cause of attentional lapses is the appearance of sleep-like slow waves in the awake state ( Andrillon et al., 2021 Andrillon et al., 2021 ). In focal brain injury, neural activity is disturbed by local sleep-liked behavior while awake ( Sarasso et al., 2020 Sarasso et al., 2020 ). Sleep problems can cause aberrant neural behaviors to appear in the DMN, central executive network (CEN), and SN in many psychiatric disorders ( Ioannides, 2018 Ioannides, 2018 ). Attention deficiency is closely linked to occurrences of sleep-like activity under awake condition ( Andrillon et al., 2021 Andrillon et al., 2021 ). In obstructive sleep apnea, aberrant neural activity appears during wakefulness, and its severity is related to the status of the insula and prefrontal cortex ( Wu et al., 2020 Wu et al., 2020 ). Multiple brain areas are involved in sleep and wakefulness, and it is necessary to look at the brain from the perspective of system neuroscience. Network representation depicts paramount interactions among neuron populations in the brain ( Liao et al., 2018 Liao et al., 2018 ; Xue et al., 2018 Xue et al., 2018 ). Graph theory is generally adopted as an important theoretical architecture because of its capability to estimate, model, and simulate the brain network. Structural connectivity represents the white matter wiring derived from diffusion tensor imaging, while functional connectivity emphasizes the synchronous reaction of separate brain areas in general. The latter also reflects the information flow of neural oscillations, and will be our focus in this paper. Network control detects the navigation of global dynamics through local perturbation along certain trajectories, and has attracted much attention in the area of brain network investigation ( Gu et al., 2015 Gu et al., 2015 ). Empirical evidence has shown the relationships between network controllability and working memory ( Beynel et al., 2020 Beynel et al., 2020 ), language perception ( Medaglia et al., 2021 Medaglia et al., 2021 ), cognitive control function ( Gu et al., 2015 Gu et al., 2015 ), creativity and intelligence ( Kenett et al., 2018 Kenett et al., 2018 ), executive function ( Cornblath et al., 2019 Cornblath et al., 2019 ), psychosis symptoms ( Parkes et al., 2021 Parkes et al., 2021 ), and schizophrenia ( Tang et al., 2022 Tang et al., 2022 ). While network control theory offers great help for exploiting working mechanisms, it mainly focuses on structural connectivity in the brain. More attention should be paid to the role of network controllability in functional interactions. The proposed pipeline is illustrated in Fig. 1 Fig. 1 . We derived time-varying functional connectivity to evaluate network controllability, and the involved cerebral areas were the SN, FPN, and DMN. Particularly, SN is comprised of the anterior insula (AI) and anterior cingulate cortex (ACC), FPN contains the middle frontal gyrus (MFG) and supramarginal gyrus (SMG), and DMN is constituted with the posterior cingulate cortex (PCC) and precuneus in this work. Neural oscillations offer subjective monitoring of the brain ( Doelling et al., 2021 Doelling et al., 2021 ). All EEG signals were band-passed into six separate sub-bands; we paid special attention to sub-slow (0.1‍–‍1 Hz) band oscillations, as they play critical roles in non-rapid eye movement stage 3 (N3) sleep. When phase-locking value and time windows were combined, we regarded the strength of instantaneous phase-locking as representing the adjacency elements in the brain networks. Both average controllability and modal controllability metrics were adopted to track and quantify brain dynamics under the corresponding four conditions as follows: rapid eye movement (REM), non-rapid eye movement stage 2 (N2), N3, and awake. The detailed methods are provided in the electronic supplementary materials for this paper. As demonstrated in Table S1, significant differences existed among all six frequency components in the targeted areas, based on modal controllability metrics ( P < 0.001). The PCC and precuneus demonstrated the frequency-dependent variations covering all four different conditions. Significant differences appeared in the AI under asleep condition but not awake condition. Since average controllability and modal controllability in turn depict easy and hard-to-attain states, respectively, in the frame of network control theory, we calculated the correlation between them. We carried out an analysis of variance (ANOVA) test to explore distinctions between them. As described in Table S2, there were significant differences among networks that were derived from six sub-band frequency components, when considering modal and average controllability metrics. Specifically, ACC, AI, MFG, and PCC all showed significant differences, especially under awake condition. In view of the brain state of certain areas that can be defined as the power of amplitude in recorded EEG signals, we calculated the power in all six sub-band EEG signals, and compared their ratios with each other. We found that there was a tiny sub-slow band EEG signal owing a ratio one ten-thousandth of that was derived from beta- or gamma-band EEG oscillation. Therefore, we removed sub-band EEG signals, and implemented average controllability on the networks derived from the remaining five sub-band EEG signals. Significant differences appeared in almost all six brain areas under all four conditions, as presented in Table S3. Exceptions were: MFG under N2 or REM conditions, SMG under REM, and AI under awake condition, where no significance appeared. Specifically, non-significant difference appeared in the MFG under N2 ( P =0.0821) or REM ( P =0.3076) condition, while the SMG ( P =0.0211) or precuneus ( P =0.0022) showed no significant differences under REM condition. Ultimately, we took three spatial levels of network connectivity into consideration for network controllability evaluation. It is found out that average network controllability is not only frequency-dependent but also spatially dependent. At the first level, individual brain areas were considered, including the AI, ACC, MFG, SMG, PCC, and precuneus. Then, at the second level, network controllability was estimated for the SN, FPN, and DMN, which are constituted by the AI and ACC, MFG and SMG, and PCC and precuneus, respectively. Finally, at the third level, the SN, FPN, and DMN made up the largest spatial SN-FPN-DMN network. Average controllability was implemented on the corresponding networks. Based on the preliminary results, we determined that extensive frequency-dependent network controllability existed in most enrolled individual brain areas. At the medial spatial network level, significant differences only existed in the SN network under REM, N3, and awake condition, while no significant difference appeared in the large SN-FPN-DMN network. To sum up, all six cerebral areas had certain frequency-dependent differences in network controllability under awake condition at the small spatial network level, whereas no such significant frequency-dependent differences emerged in the large spatial SN-FPN-DMN network. Normally, the brain is optimized to easily transition from sleep to wakefulness or vice versa. Awake and asleep states are apparently different from each other, and entanglement between them signifies brain disorder to some extent. However, quantified biomarkers to distinguish between the states are lacking. Klimesch (2018) Klimesch (2018) proposed the frequency framework of neural oscillations in the brain. According to our preliminary results, the existence of frequency-dependent network controllability provides one available biomarker, and reveals the foundation for seemingly effortless wake-to-sleep manipulation. As previously illustrated, a densely connected DMN controls easy tasks, while a weakly linked cognitive control network is responsible for difficult control issues. As nonstationary neural oscillations interact with each other across the large-scale cerebral cortex, trajectory controlling refers to driving the system from one condition to another desired condition, such as moving to the nearby states or distant states. The way in which they are integrated and segregated exhibits controlling trajectories under diverse cognitive conditions. With the assistance of network control theory, our results demonstrate the distinctions among frequency components, signifying their diverse roles in network adaption and manipulation. In general, modal controllability and average controllability are believed to act differently, as the former handles difficult tasks and the latter easy ones. The former requires integration and cooperation among more brain areas, while the latter brings neighboring areas together in order to achieve the goal. Certain brain areas prefer solving difficult problems, while others favor easy ones. Cheng et al. (2018) Cheng et al. (2018) reported that the PCC is critical in sleep and depressive problems. In modal controllability, when all six sub-band EEG signals are considered, the PCC exhibits significant differences under all four conditions, including asleep and awake states. Meanwhile, the precuneus shows significant differences from the N3 condition, while one distinction of N3 sleep is its dominance in slow-wave EEG oscillation. Kinreich et al. (2014) Kinreich et al. (2014) suggested that the AI plays important roles in the transition between wakefulness and sleep; it reflected significant differences among the three conditions other than the awake condition in this study. When modal controllability and average controllability are combined, the ACC, MFG, PCC, and precuneus show significant differences among brain networks; these differences are most apparent under awake condition. Besides, five sub-band EEG signals are enrolled while the sub-slow (0.1‍–‍1 Hz) band is removed due to its low amplitude. Average controllability shows significant differences in almost all six brain areas under the four distinct brain conditions. In this work, we employed intracranial stereo-EEG recordings and network controllability to investigate the roles that frequency components play in the brain. The difference between awake and asleep conditions is revealed in terms of network controllability. Frequency-dependent network controllability provides new insights into understanding the working mechanisms of the brain. Because stereo-EEG can detect temporal variations of neuron populations in large-scale brain areas, it offers a gold standard for evaluating neural activity. In detail, individual cortical regions demonstrate frequency-dependent network controllability to a greater extent than inter-regional interactions. Our findings demonstrate that network control theory offers a distinct systematic way of understanding frequency components in the brain. In consequence, system science should be considered as a future means of uncovering the network dynamics of frequency components in neural oscillation, which could lay a theoretical foundation for electrical neuromodulation in clinics. However, due to the limited number of participants involved in this study, more participants should be recruited to unravel the network control mystery of the brain in future work. Further investigation is necessary to uncover the individual systematic role that frequency components play in different brain regions for certain complex cognitive tasks. Materials and methods 1 Detailed methods are provided in the electronic supplementary materials of this paper. References Andrillon T , Burns A , Mackay T , et al. , 2021 . Predicting lapses of attention with sleep-like slow waves . Nat Commun , 12 : 3657 . https://doi.org/10.1038/s41467-021-23890-7 https://doi.org/10.1038/s41467-021-23890-7 Predicting lapses of attention with sleep-like slow waves Beynel L , Deng LF , Crowell CA , et al. , 2020 . Structural controllability predicts functional patterns and brain stimulation benefits associated with working memory . J Neurosci , 40 ( 35 ): 6770 - 6778 . https://doi.org/10.1523/JNEUROSCI.0531-20.2020 https://doi.org/10.1523/JNEUROSCI.0531-20.2020 Structural controllability predicts functional patterns and brain stimulation benefits associated with working memory Cheng W , Rolls ET , Ruan HT , et al. , 2018 . Functional connectivities in the brain that mediate the association between depressive problems and sleep quality . JAMA Psychiatry , 75 ( 10 ): 1052 - 1061 . https://doi.org/10.1001/jamapsychiatry.2018.1941 https://doi.org/10.1001/jamapsychiatry.2018.1941 Functional connectivities in the brain that mediate the association between depressive problems and sleep quality Cornblath EJ , Tang E , Baum GL , et al. , 2019 . Sex differences in network controllability as a predictor of executive function in youth . NeuroImage , 188 : 122 - 134 . https://doi.org/10.1016/j.neuroimage.2018.11.048 https://doi.org/10.1016/j.neuroimage.2018.11.048 Sex differences in network controllability as a predictor of executive function in youth Doelling KB , Assaneo MF , 2021 . Neural oscillations are a start toward understanding brain activity rather than the end . PLoS Biol , 19 ( 5 ): e3001234 . https://doi.org/10.1371/journal.pbio.3001234 https://doi.org/10.1371/journal.pbio.3001234 Neural oscillations are a start toward understanding brain activity rather than the end Gu S , Pasqualetti F , Cieslak M , et al. , 2015 . Controllability of structural brain networks . Nat Commun , 6 : 8414 . https://doi.org/10.1038/ncomms9414 https://doi.org/10.1038/ncomms9414 Controllability of structural brain networks Ioannides AA , 2018 . Neurofeedback and the neural representation of self: lessons from awake state and sleep . Front Hum Neurosci , 12 : 142 . https://doi.org/10.3389/fnhum.2018.00142 https://doi.org/10.3389/fnhum.2018.00142 Neurofeedback and the neural representation of self: lessons from awake state and sleep Kenett YN , Medaglia JD , Beaty RE , et al. , 2018 . Driving the brain towards creativity and intelligence: a network control theory analysis . Neuropsychologia , 118 ( Pt A ): 79 - 90 . https://doi.org/10.1016/j.neuropsychologia.2018.01.001 https://doi.org/10.1016/j.neuropsychologia.2018.01.001 Driving the brain towards creativity and intelligence: a network control theory analysis Kinreich S , Podlipsky I , Jamshy S , et al. , 2014 . Neural dynamics necessary and sufficient for transition into pre-sleep induced by EEG NeuroFeedback . NeuroImage , 97 : 19 - 28 . https://doi.org/10.1016/j.neuroimage.2014.04.044 https://doi.org/10.1016/j.neuroimage.2014.04.044 Neural dynamics necessary and sufficient for transition into pre-sleep induced by EEG NeuroFeedback Klimesch W , 2018 . The frequency architecture of brain and brain body oscillations: an analysis . Eur J Neurosci , 48 ( 7 ): 2431 - 2453 . https://doi.org/10.1111/ejn.14192 https://doi.org/10.1111/ejn.14192 The frequency architecture of brain and brain body oscillations: an analysis Liao ZL , Tan YF , Qiu YJ , et al. , 2018 . Interhemispheric functional connectivity for Alzheimer’s disease and amnestic mild cognitive impairment based on the triple network model . J Zhejiang Univ-Sci B (Biomed & Biotechnol) , 19 ( 12 ): 924 - 934 . https://doi.org/10.1631/jzus.B1800381 https://doi.org/10.1631/jzus.B1800381 Interhemispheric functional connectivity for Alzheimer’s disease and amnestic mild cognitive impairment based on the triple network model Medaglia JD , Harvey DY , Kelkar AS , et al. , 2021 . Language tasks and the network control role of the left inferior frontal gyrus . eNeuro , 8(5):ENEURO.0382-20.2021. https://doi.org/10.1523/ENEURO.0382-20.2021 https://doi.org/10.1523/ENEURO.0382-20.2021 Language tasks and the network control role of the left inferior frontal gyrus Parkes L , Moore TM , Calkins ME , et al. , 2021 . Network controllability in transmodal cortex predicts positive psychosis spectrum symptoms . Biol Psychiatry , 90 ( 6 ): 409- 418. https://doi.org/10.1016/j.biopsych.2021.03.016 https://doi.org/10.1016/j.biopsych.2021.03.016 Network controllability in transmodal cortex predicts positive psychosis spectrum symptoms Sarasso S , D'Ambrosio S , Fecchio M , et al. , 2020 . Local sleep-like cortical reactivity in the awake brain after focal injury . Brain , 143 ( 12 ): 3672 - 3684 . https://doi.org/10.1093/brain/awaa338 https://doi.org/10.1093/brain/awaa338 Local sleep-like cortical reactivity in the awake brain after focal injury Tang BQ , Zhang WJ , Deng SK , et al. , 2022 . Age-associated network controllability changes in first episode drug-naïve schizophrenia . BMC Psychiatry , 22 : 26 . https://doi.org/10.1186/s12888-021-03674-5 https://doi.org/10.1186/s12888-021-03674-5 Age-associated network controllability changes in first episode drug-naïve schizophrenia Wu Y , Zhao WR , Chen XY , et al. , 2020 . Aberrant awake spontaneous brain activity in obstructive sleep apnea: a review focused on resting-state EEG and resting-state fMRI . Front Neurol , 11 : 768 . https://doi.org/10.3389/fneur.2020.00768 https://doi.org/10.3389/fneur.2020.00768 Aberrant awake spontaneous brain activity in obstructive sleep apnea: a review focused on resting-state EEG and resting-state fMRI Xue SW , Lee TW , Guo YH , 2018 . Spontaneous activity in medial orbitofrontal cortex correlates with trait anxiety in healthy male adults . J Zhejiang Univ-Sci B (Biomed & Biotechnol) , 19 ( 8 ): 643 - 653 . https://doi.org/10.1631/jzus.B1700481 https://doi.org/10.1631/jzus.B1700481 Spontaneous activity in medial orbitofrontal cortex correlates with trait anxiety in healthy male adults Yan HE performed the data analysis and manuscript writing. Zhiqiang YAN performed the statistical analysis. Wenjia ZHANG contributed to the editing of the manuscript. Jie DONG contributed to the data analysis. Hao YAN contributed to the study design. All authors have read and approved the final manuscript, and therefore, have full access to all the data in the study and take responsibility for the integrity and security of the data. This work was supported by the National Natural Science Foundation of China (No. 62207021), the Basic Research Program for the Natural Science of Shaanxi Province of China (No. 2022JM134), and the Start-up Foundation from Xi’an International Studies University (No. KYQDF202138). We acknowledge the Open iEEG Atlas Project from the Montreal Neurological Institute (MNI), Canada. We are grateful to all the participants in this study. 10.1631/jzus.B2200393 1673-1581（2023）05-0458-05 何艳,闫志强,张文嘉等.大脑清醒和睡眠状态下的网络可控性分析[J].浙江大学学报（英文版）（B辑：生物医学和生物技术）,2023,24(05):458-462. 何艳,闫志强,张文嘉等.大脑清醒和睡眠状态下的网络可控性分析[J].浙江大学学报（英文版）（B辑：生物医学和生物技术）,2023,24(05):458-462. This work was supported by the National Natural Science Foundation of China (No. 62207021), the Basic Research Program for the Natural Science of Shaanxi Province of China (No. 2022JM134), and the Start-up Foundation from Xi’an International Studies University (No. KYQDF202138). 2022-08-04 2023-02-08 2023-05-15 2023-05-15 2023-04-14 Copyright © Zhejiang University and Springer-Verlag GmbH Germany, part of Springer Nature 2022 2023 浙江大学出版社 37538253 Yan HE, Zhiqiang YAN, Wenjia ZHANG, Jie DONG, and Hao YAN declare that they have no conflict of interest. All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2013. Informed consent was obtained from all patients for being included in the study. Compliance with ethics guidelines Supplementary information 1 Materials and methods; Tables S1‒S3 浙江大学学报（英文版）（B辑：生物医学和生物技术） 5 24