published_or_final_version / Physics / Master / Master of Philosophy

微電列陣今已被廣泛用於各種生理和理的研究。通過把微電列陣連接到肌肉細胞，細胞外的電生理信號會被有效地記錄，我們進而對尖峰信號的傳播模式進行分析，以便了解肌肉收縮的模式。本文旨在對觀測到的電生理信號進行統計模型擬合，從而獲得對於肌肉收縮模式的統計推論。我們提出了三種方法用以提取尖峰信號的激活時間，分別為均值方差法、局部加權回歸法(LOWESS方法)和Butterworth濾波法。然後對抽取出來的尖峰信號應用隨機Hough轉換，識別出多個傳播的信號波，從而得到肌肉收縮的率。對於每個信號波，我們建立了兩個模型來描述信號的傳播模式，即圓形波陣面模型和線性波陣模型。通過這兩種模型擬合，表達信傳播特徵的參數可被估算，例如激發信號波的起源位和起始時間，信號的傳播方向以及速度等。利用根據兩種模型合成的模擬數據，我們證明了隨機霍夫轉換算法和模型擬合的有效性及準確性，並把文中提出的算法用於大鼠心肌培養細胞的一個數據集。由此數據集得出的結果可以用於監測細胞的電生理變化，從而闡明藥物或環條件對心肌細胞產生的影響。 / The microelectrode array (MEA) has been widely used in physiological and pharmacological research. By attaching the MEA system to muscle cells, extracellular electrophysiological signals can be recorded, and the spike-signal propagation pattern can be analyzed for understanding the muscle contraction pattern. This thesis aims at providing a statistical framework for analyzing the muscle contraction pattern from the observed electrophysiological signals. We first provides three methods for extracting the activation time of signal spikes: the mean-variance method, the LOWESS smoothing method, and the Butterworth filtering method. The randomized Hough transform is then applied to the signal spikes to identify the multiple propagating waves, which gives the rate of beating. For each propagating wave, we propose two models to describe the signal propagation pattern, namely the circular wavefront model and the linear wavefront model. By fitting these two models, parameters that characterize the signal propagation can be estimated, such as the origin and time of excitation, the direction of propagation, and the speed of propagation. We demonstrate the performances of the randomized Hough tranform algorithm and model fitting in two simulation studies, and apply these approaches to a real data set of cultured cardiac myocytes of rats. The result may be used to monitor the electrophysiological changes and thereby elucidate the drug effect or environmental condition on cardiomyocytes. / Detailed summary in vernacular field only. / Lu, Jiayi. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2013. / Includes bibliographical references (leaves 62-65). / Abstracts also in Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Motivating problem --- p.1 / Chapter 1.2 --- An overview of MEA --- p.2 / Chapter 1.3 --- Electrophysiology of cardiac myocytes --- p.5 / Chapter 1.4 --- Organization --- p.5 / Chapter 2 --- A generative model for MEA data --- p.7 / Chapter 2.1 --- Circular wavefront model --- p.9 / Chapter 2.2 --- Linear wavefront model --- p.11 / Chapter 3 --- Computing method for MEA signals --- p.13 / Chapter 3.1 --- Preliminaries --- p.13 / Chapter 3.1.1 --- Locally weighted scatterplot smoothing(LOWESS) --- p.13 / Chapter 3.1.2 --- Butterworth filter --- p.16 / Chapter 3.1.3 --- Hough transform --- p.16 / Chapter 3.1.4 --- Nonlinear minimization --- p.21 / Chapter 3.2 --- Overall procedure for MEA data analysis --- p.24 / Chapter 3.3 --- Extract the spike activation time --- p.25 / Chapter 3.4 --- Identification of multiple propagating waves --- p.28 / Chapter 3.5 --- Model fitting --- p.29 / Chapter 3.5.1 --- Circular wavefront model --- p.29 / Chapter 3.5.2 --- Linear wavefront model --- p.33 / Chapter 4 --- Simulation study based on synthesized data --- p.35 / Chapter 4.1 --- Wave detection using Hough transform --- p.35 / Chapter 4.1.1 --- Data synthesis from linear wavefront model --- p.35 / Chapter 4.1.2 --- Performance of the randomized Hough transform --- p.38 / Chapter 4.2 --- Model fitting for signal propagating pattern --- p.38 / Chapter 4.2.1 --- Data Synthesis from circular wavefront model --- p.38 / Chapter 4.2.2 --- Performance of the model fitting algorithm --- p.42 / Chapter 5 --- Real data application --- p.47 / Chapter 5.1 --- Data set --- p.47 / Chapter 5.2 --- Extract the spike activation time --- p.49 / Chapter 5.3 --- Identify multiple propagating waves --- p.52 / Chapter 5.4 --- Model fitting --- p.52 / Chapter 5.4.1 --- Fitting the circular wavefront model --- p.52 / Chapter 5.4.2 --- Fitting the linear wavefront model --- p.55 / Chapter 5.4.3 --- Comparison of the two models --- p.56 / Chapter 6 --- Conclusions and future directions --- p.60 / Chapter 6.1 --- Conclusions --- p.60 / Chapter 6.2 --- Future directions --- p.61

Orientadores: José Wilson Magalhães Bassani, Rosana Almada Bassani / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-24T11:00:56Z (GMT). No. of bitstreams: 1 Oshiyama_NataliaFerreira_D.pdf: 4721295 bytes, checksum: 5ed8a9a173462afb13315f133bf426f8 (MD5) Previous issue date: 2014 / Resumo: O potencial de ação (PA), variação do potencial elétrico através da membrana (Em), é gerado por fluxos iônicos através de canais e transportadores, cuja função e expressão pode ser alterada por hormônios, neurotransmissores, drogas e toxinas. Trata-se de um sistema complexo, para o qual os modelos computacionais constituem ferramenta importante de estudo. No presente trabalho, foi desenvolvido um modelo de PA e transporte de Ca2+ em células ventriculares de ratos neonatos, para o que foi necessário medir a concentração intracelular de Na+ ([Na+]i) e a corrente de Na+ (INa) em cardiomiócitos isolados, sobre as quais há pouca informação na literatura, e as correntes de Ca2+ (ICa), transiente de saída (Ito) e retificadora tardia (IK) de K+, além do próprio PA para melhorar a precisão do modelo. Medições em miócitos de ratos adultos foram realizadas para comparação. Foi observada menor excitabilidade das células de ratos neonatos, o que poderia ser explicado por um deslocamento da curva de ativação de INa de ~10 mV para a direita, i.e., a ativação dos canais de Na+ ocorreu em Em menos negativos e numa faixa mais ampla de Em em miócitos de neonatos do que em células de adultos. Outra diferença encontrada foi com relação à densidade de INa, ~2 vezes maior em células de neonatos. O maior influxo de Na+ poderia causar um aumento da [Na+]i durante a atividade em células de recém-nascidos, que foi confirmado pela medição de [Na+]i. No entanto, não houve aumento significativo quando ICa e o trocador Na+/Ca2+ (NCX) foram inibidos, o que indica que o aumento da [Na+]i se deve mais ao efluxo de Ca2+ via NCX do que ao influxo pelos canais de Na+ do sarcolema. Além disso, observou- se maior duração do PA em miócitos de neonatos, que poderia ser explicada pela menor densidade observada de correntes repolarizantes (Ito e IK). No entanto, não foi detectada diferença entre idades na densidade de ICa. Dados de simulações mostraram que o retículo sarcoplasmático (RS) é a principal fonte do Ca2+ ativador da contração e que a liberação fracional de Ca2+ do RS nos ratos neonatos é menor que nos adultos, confirmando dados experimentais deste laboratório. Portanto, o modelo poderá ser utilizado para predizer possíveis alterações eletrofisiológicas dos cardiomiócitos de ratos neonatos em diferentes condições. / Abstract: The action potential (AP), a change in electrical potential across the membrane (Em), is generated by ionic fluxes through channels and transporters, of which function and expression may be affected by hormones, neurotransmitters, drugs and toxins. Computational models constitute an important tool for the study of this highly non-linear and complex system. In this work, a model of AP and Ca2+ transport in ventricular cells of neonatal rats was developed. It was necessary to measure the intracellular Na+ concentration ([Na+]i) and the Na+ current (INa), for which information in the literature is scarce, and the Ca2+ current (ICa), as well as the outward transient (Ito) and delayed rectifier (IK) K+ currents, in addition to the AP itself, to improve the accuracy of the model. Measurements from adult rat myocytes were also made in order to compare these developmental phases. It was observed that neonatal rat cells are less excitable, which could be explained by a ~10 mV shift to the right of the channel activation curve, i.e., Na+ channels activation occured at less negative Em value and over a higher range of Em compared to adult cells. On the other hand, INa density was twice as great as that in adults. This might promote increase in [Na+]i during activity in cells from newborns, which was confirmed by measurement of [Na+]i. Nonetheless, significant Na+ accumulation was suppressed when ICa and the Na+ / Ca2+ exchanger (NCX) were inhibited, which indicates that the increase in [Na+]i probably depends more on Ca2+ efflux via NCX than on the influx through sarcolemmal Na+ channels. The longer AP duration in neonatal myocytes could be explained by the lower density of the repolarizing currents (Ito and IK). However, age-dependent difference in ICa density was not observed. Simulation data agreed with experimental data from this laboratory regarding the sarcoplasmic reticulum (SR) as the main source of Ca2+ during excitation-contraction coupling and the lower SR fractional release in neonatal than in adult myocytes. In conclusion, the present model may be used to predict possible electrophysiological alterations in developing cardiomyocytes under different conditions. / Doutorado / Engenharia Biomedica / Doutora em Engenharia Elétrica