信號複雜度於腦波與心電圖之相關性分析

Correlation analysis between ECG and EEG signals based on signal complexity

林佩芬(Pei-Feng Lin)

指導教授 ： 曹建和

國立臺灣大學/電機資訊學院/生醫電子與資訊學研究所/博士(2015年)

https://doi.org/10.6342/NTU.2015.02480

摘要

緒論
生命的祕密至今無解。各種律動組合成人體的運作，是維持生命的要素。這些律動彼此交互影響、形成迴路、並共同與擾動的、多變的環境發生應對。他們提供組成生命所需有序的功能。回顧歷史，心臟在許多不同文化裡被視為是情緒和智慧的本源。大多數的神經科學家認為意識及思考只是大腦及其相關的神經生理功能。然而，心臟在腦形成以前已經開始跳動。傳統說法，心與腦經由神經及荷爾蒙兩種路徑互相溝通，其溝通是否也能經由個別律動內所隱藏的動態變化，則令人好奇。心律變異是心臟自主神經系統的指標，腦波動態變化則被證明與意識活動有關。本論文始於複習非線性訊號分析。生物訊號屬於多尺度、非線性且非穩定性序列，混沌理論是否可行不需執著，非隨機的複雜性的確存在於生物信號中。我選擇以探討規則的統計為中心的複雜性分析來探討心腦連結問題。同步記錄及非同步記錄的腦波及心電圖都將被使用。實驗對象使用一群老人，基本上心臟功能健康、而意識功能則從健全到失智拉開分布。腦的電磁活動運轉非常迅速，類穩定的腦波信號是非常短的，處於幾十秒的尺度。因此我採用符號動力學方法來分析同步信號。腦波的來源至今未有定論，其中新近被討論的皮層慢電位，其頻率範圍接近心臟的。我採用非線性、直覺性的方法來著手探討皮層慢電位。並且對於腦波之電位及即時頻率兩個成分將分開探討。

研究方法及材料
實驗對象包含89個老年門診病患，分為三個族群，38個血管型失智症、22個阿茲海默症、以及29個智能健全的對照組。多尺度熵分析(Multiscale entropy)用以分析非同步的腦波心電圖，符號動力學方法用以分析同步的腦波心電圖。另外分散信號的方法用以將波幅及即時頻率分開。傅立葉頻譜低頻對高頻比(LF/HF)用以代表交感副交感神經平衡指數。

結果與討論
在非同步信號經由多尺度熵分析，我發現心腦信號之複雜動態間，的確存在線性關係。但短暫的同步信號之間並無關聯證據。這可能因為同步於腦波的心律變異太短，無法穩定表現個案特質。不過，符號動力學方法顯示: 每個分散的腦波局部高峰，其電位值的變異與智能相關，但是腦波的及時頻率並不相關。這表是局部同步激發的皮質神經元數量而非其激發的時機與智能相關。熵的值，或說複雜度，亦即規則性的強弱並不代表健康度。不同方法測量的是生物信號的不同尺度。在腦波分析方面，分散信號的方法所得到的資訊，並不亞於全波形的分析。失智的病理表現可能是連續性而非階梯式的。

關鍵字

信號複雜性 ； 心電圖 ； 腦波 ； 多尺度熵分析 ； 符號動力學 ； 振幅 ； 及時頻率

並列摘要

Introduction
The secret of life remains extremely concealed. There are all sorts of rhythms in human bodies and they are central to life. The rhythms interact with each other as well as the outside fluctuating, noisy environment under the control of innumerable feedback systems. They provide an orderly function that enables life. The heart has been considered the source of emotional experience and wisdom in many cultures throughout the ages. Most neuroscientists consider consciousness or even thought is merely an epiphenomenon of the human brain function and its associated neurophysiology. However, the heart begins to beat before the brain is formed. Conventionally, both neural and humoral pathways connect the heart with the brain. Whether the interplay between the heart and brain could be explored through their rhythms is the question. Heart rate variability is recognized as the indicator of cardiac autonomic function. The dynamics of human electroencephalography (EEG) dynamics has been proved to be related to cognitive activities. This dissertation starts with reviewing the nonlinear methods in analyzing biological rhythms, which are multiscale, nonlinear and non-stationary. Regardless of whether chaos is present, deterministic complexity exists in biological rhythms. Regularity based complexity was chosen after comparisons. The goal is to find correlations between EEG and electrocardiography (ECG) through regularity based complexity analysis. Both simultaneous and non-simultaneous data would be examined. The experimental subjects are from a geriatric sample with varied cognitive abilities and basically healthy hearts. The electromagnetic activity of the brain works at an extremely fast speed, and the quasi-stationary epochs of EEG are, in general, short lasting, in the order of tens of seconds. Therefore symbolic techniques were introduced when exploring the very short simultaneous EEG and R-R interval (RRI) data. The origin of EEG remains unknown. Slow cortical potential (SCP), one component of EEG, is in the frequency range similar to that of the heart, and would be explored in an intuitive nonlinear way. In addition, the amplitude and instantaneous frequency of EEG would be separately approached.

Methods
The sample consisted of 89 geriatric outpatients in three patient groups: 38 fresh cases of vascular dementia (VD), 22 fresh cases of Alzheimer’s disease (AD) and 29 controls. Multiscale entropy (MSE) analysis was applied to the non-simultaneous EEG and RRI data. Symbolic analysis was applied to the simultaneous EEG and RRI data. Discrete events (local peaks) of EEG were extracted to separate the amplitude and instantaneous frequency. The low-to-high frequency power (LF/HF) ratio of RRI was calculated to represent sympatho-vagal balance.

Results and Discussions
MSE revealed correlations between the signal complexity of brain and cardiac activities in non-simultaneous data. Linear correlation between the MSE value and the score of the mini-mental state examination was first found. Symbolic dynamics failed to correlate the heart to the brain. This is due to that the RRI is too short to represent the characteristics of a subject. The symbolic analysis revealed important information that the EEG dynamics which relates to either the cognitive functions or the underlying pathologies of dementia are embedded within the dynamics of the amount of but not the interval between each synchronized firing of adjacent cerebral neurons. Just like RRI of ECG, discrete events of EEG also provided important information. The relative value of complexity does not indicate health condition straightly. It depends on the method and the scale or dimension that particular method measures. Discrete events provide no less information than continuous waveforms of EEG. Pathological condition is continuous rather than stepwise.

並列關鍵字

signal complexity ； ECG ； EEG ； Multiscale entropy ； symbolic dynamics ； amplitude ； instantaneous frequency

參考文獻

Selig FA, Tonolli ER, Silva EV, Godoy MF. Heart rate variability in preterm and term neonates. Arq Bras Cardiol. 2011;96:443-9.

Abasolo D, Hornero R, Espino P, Escudero J, Gomez C. Electroencephalogram background activity characterization with approximate entropy and auto mutual information in Alzheimer's disease patients. Conf Proc IEEE Eng Med Biol Soc. 2007;2007:6192-5.

Abasolo D, Hornero R, Espino P, Poza J, Sanchez CI, de la Rosa R. Analysis of regularity in the EEG background activity of Alzheimer's disease patients with Approximate Entropy. Clin Neurophysiol. 2005;116:1826-34.

Adler G, Brassen S, Jajcevic A. EEG coherence in Alzheimer's dementia. J Neural Transm. 2003;110:1051-8.

Agostini C, Gurrieri C. Chemokine/cytokine cocktail in idiopathic pulmonary fibrosis. Proc Am Thorac Soc. 2006;3:357-63.