CRCNS: Machine Learning for Analysis of fMRI

CRCNS：用于分析 fMRI 的机器学习

• 批准号: 7051239
• 负责人: DIMITRIOS SAMARAS
• 金额: $ 25.31万
• 依托单位: STATE UNIVERSITY NEW YORK STONY BROOK
• 依托单位国家: 美国
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-09-01 至 2010-12-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7051239
• 关键词: behavioral /social science research tag; bioimaging /biomedical imaging; clinical research; cocaine; computer data analysis; drug addiction; functional magnetic resonance imaging; genotype; human subject; neuropsychology; reinforcer; substance abuse related behavior

DESCRIPTION (provided by applicant): Compromised inhibitory control is a hallmark neuropsychological deficit underlying complex disorders and behaviors as drug addiction and aggression. Individuals with compromised mechanisms of control are difficult to identify unless they are subjected to challenging conditions directly. This is possibly due to the subtle ways in which poor inhibitory control is expressed in their baseline functioning, making classification and most importantly, prediction of future behavior, extremely challenging. We propose novel computational techniques to analyze brain-behavior relationships underlying mechanisms of inhibitory control, focusing on performing classification of hard-to-categorize groups of subjects based on brain activation response patterns to behavioral challenges of inhibitory control using functional magnetic resonance imaging (fMRl). These classification methods are applied on two distinct datasets: one of substance dependent individuals and the other of individuals with a particular genotype conferring vulnerability to poor inhibitory control. We hypothesize that unique patterns of variability in brain function can assist in identification of brain mechanisms rooted in compromised inhibitory control. Such patterns will increase our understanding of brain connectivity and circuitry as we move iteratively between a-priori and exploratory means of describing circuits of inhibitory control. Machine Learning techniques have been shown to be successful in discovering optimal features and patterns in complex high dimensional datasets. The diversity of the underlying questions when studying inhibitory control and the subtlety of the effects that can be used for classification, motivate us to propose an integrated machine learning framework for the joint exploration of spatial, temporal and functional information for the analysis of fMRI signals, thus allowing the testing of hypotheses and development of applications that are not supported by traditional analysis methods. We hypothesize that: 1) A differential spatial brain pattern will indicate a diagnosis of drug addiction and a membership in one or another level ol MAOA genotype. Spatial information from static 3D contrast maps will be input to PCA-based and Voxel-based methods, Adaboosting and Learning with Side information 2) A temporally accounted intrasubject pattern of response to the inhibitory control challenge conditions in the fMRI paradigms will reveal group membership in both data sets. Temporal fMRl information wilt be used for Hidden Markov Models, Conditional Random Fields, etc. 3) A connectivity map corresponding to brain circuits functionally subserving inhibitory control wilt be revealed with indications of directionality of influence between brain regions by analyzing functional information with Dynamic Bayesian Networks, Dynamically Multi-Linked HMMs, etc.

描述（由申请人提供）：受损的抑制性控制是一种标志性的神经心理缺陷，基本的复杂疾病和行为是药物成瘾和侵略性。除非直接遭受具有挑战性的条件，否则很难识别具有受损的控制机制的个体。这可能是由于其基线功能中表达较差的抑制性控制的微妙方式，最重要的是对未来行为的预测，极具挑战性。我们提出了新型的计算技术来分析抑制性控制的基本机制的脑行为关系，重点是基于脑激活响应模式对使用功能磁共振成像（FMRL）的抑制性控制行为挑战进行难以分类的受试者组进行分类。这些分类方法应用于两个不同的数据集：一个依赖物质的个体，另一个具有特定基因型的个体，赋予了抑制性较差的控制。我们假设大脑功能的独特变异性模式可以帮助识别植根于受损抑制性控制的大脑机制。当我们在A-Priori和描述抑制性控制电路的探索方式之间迭代时，这种模式将增加我们对大脑连通性和电路的理解。机器学习技术已被证明成功地发现了复杂的高维数据集中的最佳功能和模式。在研究抑制性控制和可用于分类的效果时，基本问题的多样性激励我们提出一个集成的机器学习框架，以共同探索空间，时间和功能信息，以分析FMRI信号，从而允许对传统分析方法的假设和开发的测试和开发。我们假设：1）差异空间大脑模式将表明对药物成瘾的诊断和一种或另一个级别的OL MAOA基因型的成员资格。来自静态3D对比图的空间信息将输入基于PCA的基于PCA和基于体素的方法，使用侧面信息进行ADABOOSOSTING和学习。2）在FMRI范式中，对抑制性控制挑战条件的临时响应模式将在两个数据集中揭示小组成员。时间FMRL信息枯萎可用于隐藏的马尔可夫模型，有条件的随机场等。3）3）与大脑电路相对应的连通性图在功能下抑制性控制的枯萎病，可以通过用动态的贝叶斯网络分析动态贝叶斯网络（动态多连接的HMM等）来揭示大脑区域之间影响方向的指示。