【作者】 高原；

【导师】 相洁； 谭淑平；

【作者基本信息】 太原理工大学 ， 计算机技术（专业学位）， 2020， 硕士

【摘要】 阿尔茨海默病(Alzheimer’s disease,AD)是多见的人类神经退行性疾病之一,并且常受多种基因和遗传变异的直接影响。目前这种不可逆的人类脑部神经损伤疾病被卫生部列为第四位最常见的疾病和死亡危险原因,对于社会、家庭都会造成严重的经济负担。一些研究人员认为,轻度的认知障碍(Mild cognitive impairment,MCI)是属于从正常人(Normal Control,NC)过渡发展到AD的一个中间状态,且超过50%的MCI患者会转换成AD,也有一些MCI患者能够保持稳定甚至能恢复成NC。因此,发现AD早期异常特征,研究计算机辅助的自动、高效、准确的方法,实现痴呆早期诊断和干预极其必要。目前,国内外对于AD的发病机制研究尚未达到统一,因此寻找AD早诊的可靠指标仍然是当前研究的热点。近年来,随着以功能磁共振成像(functional Magnetic Resonance Imaging,f MRI)为基础的图论分析和复杂网络研究不断深入,为AD谱系人群中异常脑神经活动量化分析及生物标志物寻找提供了技术保证。同时,国内外的学者们发现人脑在婴儿期间就存在不同的功能模块组织,其拓扑属性更精确地反映了脑模块间的信息交流模式和组织规律的变化。近年来,遗传因素已经被医学界广泛认为是能够最早准确预测和判断AD发病风险的一个重要指标,也有大量的学者通过双生子研究结果证实了患有AD的人群在临床上表现出极高遗传性。在对AD早期遗传风险因素的筛查中,全基因组关联分析(Genome-Wide Association Studies,GWAS)结合生物表型,进而发现在AD疾病发展中具有致病风险的基因,其基本原理是采取对所有单核苷酸多态性(Single Nucleotide Polymorphism,SNP)进行早期遗传疾病的风险评估,在筛选具有普通风险覆盖率和疾病遗传效应微弱的风险变异基因方面具有强大的优势。因此,本研究充分利用了GWAS研究的优点,结合生物表型分析发现AD的致病风险变异基因。因此,本研究根据诊断组将被试人群分为NC,MCI以及AD,基于全基因组结合图论理论的关联分析,寻找阿尔茨海默病谱系被试人群的全基因组中与脑功能网络活动相关联的变异性基因,分析在变异基因影响下的脑神经活动呈现出的异常发展规律及对AD疾病发展的影响。为之后AD谱系疾病的辅助诊断及治疗的研究提供了依据。本次研究主要的内容和研究成果主要包括:(1)构建功能脑网络及定量分析以图论分析为技术基础的复杂大脑功能网络定量分析方法能够迅速地广泛应用在对脑部功能网络和组织的研究当中。其拓扑结构特征分析具有重要的科学研究的价值,能够有效直接地反应复杂大脑功能网络连接的异常。因此,研究基于解剖自动化的标记模板(Automated Anatomical Labeling,AAL)对大脑功能网络拓扑结构分析,利用MATLAB对二值邻接矩阵进行脑网络拓扑结构属性的计算,进而达到对脑网络定量分析的目的。(2)基于图论全基因与脑网络指标关联分析针对AD患者人群具有极高的基因遗传性,且目前的研究在对AD早期致病风险基因的分析中常常因忽略了潜在的风险变异基因与表型间的相互作用,会导致出现丢失了AD等神经疾病的潜在致病风险基因位点的异常情况,因此本研究采取GWAS作为进一步研究AD早期发病机理的重要分析手段。另一方面由于功能磁共振影像能够及时呈现脑疾病的神经活动变化,基于图论发展构建功能脑网络能够实现对脑神经活动的定量分析。因此,研究首次提出利用GWAS结合功能脑网络指标寻找发现AD早期风险变异基因。(3)分析不同基因型对脑网络与模块网络的指标影响针对当前研究中缺乏AD谱系人群脑连接模块网络下信息交流的变化在致病位点基因影响下的分析研究,本课题进一步研究在考虑基因分型的情况下,提出了将AD发展人群的脑功能连接网络模块化,全面地考虑各功能模块脑网络下功能连接间信息交流变化。研究者首先根据磷酸二酯酶(Phosphodiesterases 4D,PDE4D)不同的基因分型,分别对两两被试间的脑网络和其模块下的拓扑网络属性进行了统计分析检验,分析了PDE4D野生型和变异型对早期AD谱系人群全脑连接网络及其模块子网络拓扑属性的直接影响。(4)分析PDE4D的野生型与变异型分型对NC,MCI和AD分类准确率的影响为了分析全脑与模块网络拓扑属性在目标基因PDE4D野生型和变异型影响下对AD早期诊断的差异,研究者提取了不同基因型下诊断组间存在显著差异的脑网络属性特征,使用支持向量机(Support Vector Machine,SVM)进行分类,进行分类测试,加入基因变量后网络指标分类准确率对比无基因变量的网络指标分类准确率,进而分析在PDE4D不同目标基因型的影响下被试组间的网络指标分类准确率的差异。这一步研究结果表明,将被试人群按PDE4D的基因型细化分组后,分类准确率都有明显的提升。综上所述,本研究通过两两比较AD谱系人群诊断组,对比分析携带PDE4D变异基因型人群与携带PDE4D野生型人群的功能脑网络属性,提取诊断组间存在有显著性差异的特征加入到疾病辅助诊断中。研究结果一致表明,PDE4D基因变异型直接影响早期AD的功能网络神经的正常活动,若之后临床研究能够充分考虑到的PDE4D变异基因分型对早期AD人群疾病诊断的重要性影响,分类的准确性将有可能会进一步提高。更多还原

【Abstract】 Alzheimer’s disease(AD)is one of the most common human neurodegenerative diseases and is often directly affected by a variety of genes and genetic variations.Currently,this neuroprogressive and irreversible degenerative disease of nerve damage in the human brain is listed by the ministry of health as the fourth most common cause of disease and risk of death.Some researchers believe that Mild cognitive impairment(MCI)is an intermediate state from Normal Control(NC)to AD,and more than 50% of MCI patients will convert to AD,and some MCI patients can remain stable or even recover to NC.Therefore,it is extremely necessary to discover abnormal features of early AD and study computer-aided automatic,efficient and accurate methods to achieve early diagnosis and intervention of dementia.At present,the research on the pathogenesis of AD has not been unified at home and abroad,so it is still a hot topic to search for reliable indicators of early diagnosis of AD.In recent years,with the deepening of the graph theory analysis and complex network research based on functional Magnetic Resonance Imaging(f MRI),the quantitative analysis of abnormal brain neural activity and the search for biomarkers in AD spectrum population have been provided with technical guarantee.At the same time,scholars at home and abroad have found that there are different functional modules in human brain during infancy,and its topological properties more accurately reflect the changes of information communication patterns and organizational rules between brain modules.In recent years,genetic factors have been widely recognized by the medical community as an important indicator for the earliest accurate prediction and judgment of the risk of AD,and a large number of scholars have confirmed the high clinical heritability of people with AD through the results of twin studies.In the early screening of genetic risk factors for AD,genome-wide Association analysis(Genome-Wide Association Studies,GWAS)combined with pathogenic biological phenotype analysis,and then found in the development of AD disease with the mutated gene disease risk,its basic principle is to take all SNPs for risk assessment of genetic disease,early in the filter with ordinary genetic effect weak coverage and disease risk gene variant has strong advantage.Thus,this study made full use of the advantages of the GWAS study,and combined with the biological phenotype analysis,found the pathogenic risk mutation gene of AD.Therefore,this study divided the subjects into NC,MCI and AD according to the diagnostic group.Based on the association analysis of whole genome combined with graph theory,this study searched for the variant genes associated with the brain functional network activity in the whole genome of the subjects of alzheimer’s disease pedigree,and analyzed the abnormal development rule of the brain neural activity under the influence of the mutation gene and its influence on the development of AD disease.It provides a basis for the research of adjunctive diagnosis and treatment of AD spectrum diseases.The main contents and results of this study mainly include:(1)Construction of functional brain network and quantitative analysisThe quantitative analysis method of complex brain functional network based on graph theory analysis can be widely used in the study of brain functional network and organization.The analysis of its topological structure features is of great scientific value and can effectively and directly reflect the abnormality of complex brain functional network connection.Therefore,this paper studies the topological structure analysis of brain functional network based on automated evaluation labeling(AAL),and USES MATLAB to calculate the topological structure attribute of brain network based on binary adjacency matrix,so as to achieve the purpose of quantitative analysis of brain network.(2)Correlation analysis between whole gene and brain network index based on graph theoryTargeted on the AD patients have high genetic heritage,and the current study pathogenic genes of risk analysis in the early of AD were often ignore the potential risk of interaction between genetic and phenotypic variation,will result in a lost and the potential risk of neurological diseases pathogenic gene loci of abnormal situation,so this research take the GWAS as further study of the pathogenesis of AD early important analysis method.On the other hand,functional magnetic resonance imaging(fmri)can present the changes of neural activity of brain diseases in a timely manner.Therefore,this study proposed for the first time that GWAS was combined with functional brain network index to find the risk mutation gene of early AD.(3)Analyze the influence of different genotypes on indexes of brain network and module networkIn view of the current lack of AD brain connection module spectrum crowd in the study of information communication under the network analysis and study of the impact of pathogenic gene loci,this topic research under the condition of considering the genotyping,AD is put forward to develop the crowd of brain function to connect to the Internet modular,fully consider each function module brain connections between information exchange network function changes.Firstly,according to the different genotypes of PDE4 D,the researchers conducted statistical analysis and test on the brain network between two subjects and the topological network properties under their modules respectively,and analyzed the direct influence of the wild type and variant type of PDE4 D on the whole brain connection network and the topological properties of the module sub-network in early AD pedigree population.(4)To analyze the influence of target genotyping of wild type and variant type on the classification accuracy of NC,MCI and ADIn order to analyze the whole brain and module network topological properties in the target gene PDE4 D under the influence of the wild-type and variant for early diagnosis of AD differences,the researchers extracted the diagnosis of significant differences between groups under different genotypes of brain network attribute characteristics,using the SVM classification,classification test: after joining genetically variable network index classification accuracy vs.no genetically variable network indicators classification accuracy is analyzed under the influence of different target PDE4 D genotype subjects network index classification accuracy of the differences between groups.The results of this step showed that the classification accuracy was significantly improved after the target gene of PDE4 D and the subject group with variant gene were refined.To sum up,this study compared the diagnosis group of people of AD spectrum in pairs,compared and analyzed the functional brain network properties of people with PDE4 D variant genotype and people with PDE4 D wild type,and extracted the characteristics of significant differences between the diagnosis groups and added them into the auxiliary diagnosis of diseases.Consistent research results show that the PDE4 D genotype directly affects the normal activity of the functional network nerves in early AD.If the importance of the PDE4 D genotype on disease diagnosis in early AD population can be fully considered in subsequent clinical studies,the accuracy of classification may be further improved.更多还原