|
| 作者 | 李欢冬
|
| 学位类别 | 工学博士
|
| 答辩日期 | 2015-12-05
|
| 授予单位 | 中国科学院大学
|
| 授予地点 | 北京
|
| 导师 | 蒋田仔
|
| 关键词 | 丙泊酚
麻醉
功能磁共振
功能连接
功能整合
脑岛
默认网络
特征选择
多体素模式识别
支持向量机
|
| 学位专业 | 模式识别与智能系统
|
| 中文摘要 | 丙泊酚是目前临床应用最广泛的静脉全身麻醉药，但是其确切的作用机制还未完全阐明。全身麻醉机制的研究一直是麻醉学研究的一个热点。由于技术水平的限制，之前对全麻药的作用机制研究多集中在分子和细胞水平，观察全麻药物对离子通道和受体的影响。这些传统手段很难在大脑对全麻药的作用区域进行精确的定位。近20年，脑功能成像技术的发展为麻醉机制的研究提供了新的途径。本研究基于功能磁共振影像，以健康志愿者为研究对象，运用功能网络分析体系以及机器学习的方法来探索全身麻醉药物丙泊酚的大脑中枢的作用机制。本文的主要工作有：
-
研究丙泊酚对脑岛三个功能子区功能连接模式的影响。我们利用清醒状态的静息态功能数据将具有复杂结构和功能的脑岛分割成背侧前脑岛，腹侧前脑岛和后侧脑岛三个子区，以分割得到的这三个子区作为种子点进行功能连接分析。三个子区分别主要参与大脑的认知处理、情感处理、感觉运动处理。我们发现脑岛认知网络与脑岛感觉运动网络从轻度镇静时就受到丙泊酚的明显抑制，深度镇静时继续加深，而脑岛情感网络受丙泊酚的影响很小。
-
深入探讨丙泊酚镇静逐步加深对默认网络的影响。通过图论的方法，我们深入地观测默认网络在不同麻醉浓度下全局与局部网络拓扑属性的变化（度、聚类系数、最短路径长度、全局效率和局部效率）。我们发现默认网络是对丙泊酚敏感的网络，在低剂量的药物浓度时就受到比较严重的抑制了。随着浓度的升高，这种抑制作用增强。其中前内侧前额皮层、右侧额上回、左侧颞下回与左侧背侧顶叶的网络属性随麻醉浓度逐渐改变，是默认网络中受到丙泊酚影响的核心节点，腹内侧前额皮层与海马旁回是受药物影响最少的区域。
-
研究丙泊酚对全脑功能网络的作用。通过观察全脑网络随丙泊酚剂量增大过程中的变化，我们发现全脑网络的全局网络属性没有显著的改变，但是hub节点的分布发生了大的迁移，从清醒状态时主要位于额叶和顶叶的多模式联络皮层逐渐往后迁移。此外，我们发现在深度镇静时改变的功能连接主要是与这些hub节点相关，并且主要是网络之间而不是网络内部连接的改变。我们猜测这种网络间信息整合的破坏，hub节点的迁移可能与意识水平的下降有关。
-
利用功能连接模式区分丙泊酚引导的不同意识水平。在本项工作中，我们基于功能连接，使用前向选择后项剔除的支持向量机分别考察了清醒、轻度镇静以及深度镇静三种状态数据之间的分类状况。通过权值分析，我们反推出了在分类模型中贡献最显著的生物标记。结果显示仅需18（总共1128条）条功能连接就能将清醒状态与深度镇静状态分开，分类精确度高达96.4%，分类权重最大的脑区主要位于小脑、辅助运动区、颞下回等区域，显著特征主要位于默认网络、执行控制网络、凸显网络三大网络间，提示我们这些网络以及脑区在丙泊酚中枢作用机制中的重要地位。
|
| 英文摘要 | Propofol has been widely used clinically as an intravenous anesthetics. But we still do not know the precise mechanism by which anesthetic agents affect the brain activity. The study of general anesthesia mechanism is hot in anesthesiology. Due to technical limitations, previous research was concentrated on molecular and cellular level and studied the effects of general anesthetics on the entire range of ion channels and neurotransmitters. These traditional methods are difficult to locate the exact target of general anesthetics. Over the past 20 years, the development of functional brain imaging techniques provided a new way to study the mechanisms of anesthesia. In the present study, we will apply methodologies of brain network and machine learning to explore the mechanism of propofol based on functional magnetic resonance imaging data from healthy volunteers. The main contributions of this thesis include following issues:
-
Investigating the changes of the functional connectivity pattern of insular subdivisions induced by increasing doses of propofol. We parcellated the structurally and functionally complex insula into three subregions based on the data at awake state, and used these parcellated subregions as seeds for functional connectivity analysis. The dosal anterior insula (dAI) is more involved in high-level cognitive processes, the ventral anterior insula (vAI) is associated with affective processes and the posterior insula (PI) is associated with sensorimotor processes. In present work, we found the insular cognitive network and insular sensorimotor network were propofol-sensitive and progressively inhibited with propofol in a dose-dependent manner. However, the insular affective network under the influence of propofol was small.
-
Observe the changes in default mode netwok (DMN) induced by increasing doses of propofol and progressively deepening sedation. Applying graph theoretical analysis, we provided detailed observation and comprehensive investigation of global and local changes (degree, clustering coefficient, shortest path length, global efficiency and local efficiency) of DMN during wakefulness, light sedation and deep sedation. We found that DMN is propofol-sensitive. A small dose of propofol can significantly inhibit the DMN, affecting the level of consciousness. As the dose was increased, the inhibitory effect was enhanced. Specially, propofol had a dose-dependent inhibitive effects on the anterior medial prefrontal cortex (aMPFC), right superior frontal gyrus, left inferior temporal gyrus and left lateral parietal lobe. Ventral medial prefrontal cortex (vMPFC) and parahippocampus (PHC) were less sensitive to propofol, and could be significantly inhibited by a higher concentration of propofol.
-
Studying how propofol affects the whole-brain functional network. Applying graph theoretical analysis, we investigated functional connectivity and topological properties of whole-brain network during wakefulness and light sedation and deep sedation. We found no significant changes of the global network topology properties(degree, clustering coefficient, shortest path length, global efficiency and local efficiency). However, we found that hub distributions were drastically affected by propofol. In normal awake state, hubs were in frontal and parietal areas. As the dose was increased, hubs transferred from heteromodal association cortex in the anterior part of the brain to unimodal association and primary cortices in the posterior part. In addition, we observed systemic decreases in functional connectivity, particular inter-network connections. Most of the functional connectivity was hub-related. this study provides imaging evidence that propofol may suppress consciousness through reconfiguring hub structures and disrupting the functional integration.
-
Identifying intrinsic connectivity patterns during propofol-induced loss of consciousness. In this study, we applied the support vector machine with forward-back search strategy (SVM-FoBa) algorithm to classify different states of consciousness (wakefulness, light sedation and deep sedation) using intrinsic network patterns. Besides, with the feature subset directly extracted from the original features, through model understanding (e.g., weight analysis) we can readily identify the informative features. Classification accuracy of wakefulness and deep sedation achieved 96.4% based on only 18 features (total 1128). Most of these informative features were functional connectivity within and between DMN, executive control network (ECN), and salience network (SN). The most discriminative brain regions were cerebellum, supplementary motor area (SMA) and inferior temporal gyrus. The findings suggest that these networks and regions may play an important role in the neurobiological basis of consciousness and propofol.
|
| 学科主题 | 工学
|
| 语种 | 中文
|
| 源URL | [http://ir.ia.ac.cn/handle/173211/10555]  |
| 专题 | 毕业生_硕士学位论文
|
| 作者单位 | 中国科学院自动化研究所模式识别国家重点实验室脑网络组研究中心
|
推荐引用方式
GB/T 7714 |
李欢冬. 丙泊酚中枢作用机制的功能磁共振影像研究[D]. 北京. 中国科学院大学. 2015.
|
