【作者】 刘鹏；

【导师】 秦伟；

【作者基本信息】 西安电子科技大学 ， 生物信息科学与技术， 2019， 博士

【摘要】 冲突解决是认知控制最重要的能力之一,其常与健康人群的社会成就,自我控制以及大脑神经和精神疾病患者认知功能损伤相关。Flanker任务是实验室考察冲突解决的常用范式,在Flanker任务中冲突解决很大程度上依赖于视觉空间注意力系统对空间注意力的调控,而偏头痛患者前庭感觉缺失导致了负责处理空间信息的大脑区域的功能变化。因而,深入研究冲突解决的神经机制,除了对于基础心理学与神经科学理论至关重要外,还能帮助我们更好的理解健康群体的认知控制个体差异,发现偏头痛患者群体认知功能紊乱的精准生物学靶点。随着大数据和人工智能技术的发展,影像组学在个体化认知训练选择和疾病精准诊疗的作用日益凸显。一方面,研究冲突解决的学习效应以及建立基于影像组学的预测模型不仅可以帮助特殊场景下的人才选拔,还可作为认知训练辅助选择工具。另一方面,偏头痛作为20大失能疾病之一,不管对个体还是社会都是沉重的负担。在临床上,偏头痛患者常表现出脑梗高发性,但临床影像检查时,偏头痛患者却很少呈现肉眼可见的脑组织结构改变。影像组学纹理特征通过考察局部异构性可以检测大脑组织微小的变化。以其为特征向量是否可以将偏头痛患者从健康患者中诊断出来,对于预防偏头痛样脑梗意义重大。具体的讲,本文的研究内容主要有以下几个方面:近年来,关于冲突解决的主流冲突监测理论受到特征整合和偶然学习理论的质疑,研究者认为冲突解决的神经机制需要重新定义。因此,本论文首先在排除了特征整合和偶然学习的“纯净的”认知控制框架下,考察冲突解决的神经机制。第一个研究利用FCz和P3/4电极之间的EEG theta相位同步考察内侧额叶皮层(Medial Frontal Cortex,MFC)和后顶叶皮层(Posterior Parietal Cortex,PPC)之间的功能连接。结果显示FCz和P4间theta相位同步在冲突状态目标刺激出现后的200到400 ms之间显著增强,提示MFC和右侧PPC之间反应冲突诱发的功能连接,该结果证明了MFC在冲突监测中的重要作用。接下来,将EEG和弥散张量成像技术结合,利用分层回归分析FCz-P4冲突诱发的theta相位同步与白质纤维束骨架统计方法(tract-based spatial statistics,TBSS)分析中显著相关脑区白质完整性之间的关系。分层回归分析显示,44%的FCz-P4冲突诱发的theta相位同步个体变量可以用胼胝体膝部的轴向扩散率(Axial Diffusivity,AD)的差异来解释,表明胼胝体膝部的白质结构完整性可以预测MFC和右侧PPC之间冲突诱发的功能连接。这些发现仅限于右半球,这可能与双侧PPC在冲突处理中不对称的作用有关。不同于第一个工作中的单电极对功能连接分析,本文第二个研究从脑网络层面分析了冲突解决的神经机制。利用图论的理论知识,考察全脑26个电极之间的delta,theta,alpha和beta频段的相位同步,设定阈值,定义高连通度的节点为功能“hub”。结果表明目标刺激出现后200 ms到400 ms之间,theta相位同步在FCz和其周围电极的连接度显著增加,且FCz电极的连接度与试次内冲突解决能力负相关。冲突解决的特质性(个体内稳定,个体间差异大)是研究冲突解决个体差异的基础,因此,本论文第三个工作考察了冲突解决的特质性和学习效应。参与者间隔2周参加相同的Flank任务。研究结果表明尽管试次内冲突解决呈现出较强的特质性,依然可以通过练习得以提升,且提升程度具有较大个体差异。试次间冲突解决并没有显著特质性,也不能通过练习提升,更多是状态依赖。上述研究发现冲突解决的学习效应具有较大的个体差异,因此,本文第四个工作基于影像组学建立了试次内冲突解决学习效应的预测模型。首先,TBSS分析结果表明试次内冲突解决提升与胼胝体体部,压部,右侧上放射冠,右侧后放射冠以及右侧上众束区域的AD值相关。随后,利用最小绝对收缩和选择算子(Least Absolute Shrinkage and Selection Operator,LASSO)逻辑回归和留一法交叉验证来评估显著相关脑区白质微观结构对于学习效应的预测能力。结果表明,冲突解决学习效应的个体差异量61%可以由四个显著脑区的AD值和基线能力来解释。这些结果表明,连接冲突处理相关脑区的白质纤维的轴突形态可以预测试次内冲突解决的学习效应。本文第五个工作分析了无先兆偏头痛患者(Migraine without Aura,MOA)冲突解决能力,建立了基于影像组学的偏头痛诊断模型。偏头痛患者试次内冲突解决显著弱于健康对照组,表明MOA高级认知功能受损。利用临床表观扩散系数(Apparent Diffusion Coefficient,ADC)和T1加权成像(T1-weighted MRI,T1-w),提取脑干区域基于灰度共生矩阵和灰度游程矩阵的纹理特征,建立影像组学标签和偏头痛诊断模型。结果表明,影像组学标签和患者病程正相关。模型在独立验证集的准确率,敏感性和特异性为别为83%,92%,以及75%,均优于单模态模型,这表明T1-w和ADC脑干纹理特征可能反映了差异的,相互补充的偏头痛病理学机制。综上所述,本文探索了冲突解决的神经机制,为theta相位同步和MFC在冲突解决中的“hub”作用提供了新的理论依据。建立了基于影像组学的冲突解决学习效应预测模型及偏头痛诊断模型,为个体特异的认知功能训练和偏头痛样脑梗预测性诊断提供了新的思路。更多还原

【Abstract】 Conflict processing is one of the most important abilities of cognitive control.It is often associated with the social achievement and self-control in healthy people,and cognitive impairment in patients with neurological and psychiatric disorders.Flanker task is a common paradigm for studying conflict processing in the laboratory.Conflict processing in the Flanker task largely relies on the regulation of spatial attention by the visual spatial attention system.Researches have shown that the absence of vestibular sensation in migraine patients leads to functional changes in the brain region responsible for processing spatial information.Thus,in-depth study of the neural mechanisms of conflict resolution is essential for basic psychology and neuroscience theory.It can also help us better understand the individual differences in cognitive control in healthy groups and find precise biological targets of cognitive dysfunction in patient populations.With the development of big data and artificial intelligence technology,the role of radiomics in individualized cognitive training selection and accurate diagnosis and treatment of diseases has become increasingly prominent.The study on the learning effect of conflict resolution and the establishment of the prediction model based on radiomics can not only help the talent selection in special situations,but also serve as an auxiliary selection tool for cognitive training.Migraine,as one of the 20 major disability diseases,is a heavy burden on both individuals and society.Clinically,migraine patients often show a high incidence of cerebral infarction,but in clinical imaging,migraine patients rarely show visible changes in brain structure.Radiomics texture features can detect subtle changes in brain tissue by examining local heterogeneity.Whether the characteristic vector can be used to diagnose migraine patients from healthy patients is of great significance in preventing migraine-liking cerebral infarction.Specifically,the research contents of this paper mainly include the following aspects:In recent years,the dominant conflict monitoring theory of conflict processing has been questioned by feature integration and contingency learning.Researchers believe that the neural mechanism of conflict processing needs to be redefined.Firstly,this study re-examine the neural mechanism of conflict processing in "pure" cognitive control without feature integration or contingency learning confounds.FCz-P3/4 theta phase synchronization was used to measure functional connectivity between the medial frontal cortex(MFC)and posterior parietal cortex(PPC).The results showed that FCz-P4 theta phase synchronization was significantly enhanced during the interval from 200 to 400 ms post-target in conflict condition,suggesting the conflict-induced functional connection between MFC and the right PPC,indicating the important role of MFC in conflict monitoring.And then hierarchical regression analyses were used to assess the relationship between FCz-P4 conflict-induced theta phase synchronization and white matter integrity in significant regions derived from tract-based spatial statistics(TBSS)analysis.It was found that FCz-P4 theta phase synchronization was significantly enhanced during conflict,suggesting a conflict-induced functional connectivity between MFC and right PPC.However,these findings were only found in the right hemisphere,which may be related to the asymmetrical role of the bilateral PPC in response conflict processing.Furthermore,hierarchical regression analyses revealed that 44% of individual variability in FCz-P4 conflict-induced theta phase synchronization could be explained by variations in axial diffusivity(AD)in the genu of the corpus callosum(g CC).These results demonstrated that structural integrity in the g CC predicts conflict-induced functional connectivity between the MFC and right PPC.Different from the single electrode pair functional connectivity analysis in the first study,the second study analyzes the neural mechanism of conflict processing at the brain network level.Based on the theoretical knowledge of graph theory,the phase synchronization of delta,theta,alpha and beta bands between 26 electrodes in the whole brain was investigated,and the threshold was set.The node with high connectivity degree was defined as the functional "hub".The results showed that the theta phase synchronization connection degree of FCz and its surrounding electrodes were significantly increases between 200 ms and 400 ms after the occurrence of target stimulus.More importantly,the connectivity degree of FCz electrode is negatively correlated with the intra-trial conflict processing ability.The trait characteristic of conflict processing(stable within individuals,large differences between individuals)is the basis of studying individual differences in conflict processing.The third part examined the trait characteristics and learning effects of conflict processing.Participants performed the same flanker task on two occasions 2 weeks apart.Despite the trait-like characteristics of conflict processing,within-trial conflict processing in the second behavioral session was significantly shorter than that in the first session,indicating a practice-based improvement in conflict processing.Furthermore,changes in within-trial conflict processing across the two sessions exhibited significant individual differences.We found no trait-like characteristics or learning effect for CSE,indicating that conflict processing at the trial-by-trial level might be more state-dependent.The above study found significant individual differences in the learning effect of conflict processing.Therefore,the fourth work established a prediction model of learning effect of within-trial conflict processing based on radiomics.Tract-based spatial statistics revealed that the improvement across two sessions was related to the axial diffusivity values in white matter regions,including the body and splenium of the corpus callosum,right superior and posterior corona radiate,and right superior longitudinal fasciculus.Subsequently,lasso regression with leave-one-out cross validation was used to assess the predictive ability of white matter microstructural characteristics in significant regions.The results showed that 61% of individual variability in the improvement in the within-trial conflict processing could be explained by variations in the axial diffusivity values in the four significant regions and the within-trial conflict processing in the first session.These results suggest that axonal morphology in the white tracts connecting conflict-related regions predicts the degree of within-trial conflict processing improvement across two sessions.The fifth work analyzed the conflict processing ability of migraineur without aura(MOA)and established a MOA diagnosis model based on radiomics.Intra-trial conflict processing in MOA was significantly weaker than that in the healthy control group,indicating impaired advanced cognitive function in MOA.Clinical ADC and t1-w images were used to extract the texture features of brainstem region.The texture features were computed from gray co-occurrence matrix and gray run-length matrix.Radiomics signature and MOA diagnosis model were established.The results showed that the radiomics signature was significantly positively related to disease.The accuracy,sensitivity,and specificity of the model in the independent validation set were 83%,92%,and 75%,respectively,which were all superior to the single-mode model,suggesting that t1-w and ADC brainstem texture characteristics may reflect different,complementary migraine pathology mechanisms.In summary,this study explores the neural mechanism of conflict processing,providing new theoretical evidences for the important role of theta phase synchronization in conflict resolution and the "hub" role of MFC.Prediction model for conflict processing learning effect and diagnosis model for MOA were established based on radiomics.The study provided new ideas for individualized cognitive function training and predictive diagnosis of migraine-like cerebral infarction.更多还原