卒中后抑郁患者脑网络拓扑改变的研究

发布时间：2018-05-08 05:06

本文选题：弥散张量成像 + 小世界属性　；参考：《广西医科大学》2017年硕士论文

【摘要】：目的:应用DTI技术及图论方法探讨卒中后抑郁症(Post stroke depression,PSD)患者全脑白质结构网络拓扑属性可能变化特征,寻找PSD脑网络影像标志物,为早期PSD诊断和疗效评价提供客观的诊断指标。方法:对48例发病在2周以内的急性缺血性脑卒中患者按卒中后抑郁的诊断DSM-IV标准,采用24项汉密尔顿抑郁量表评分,将患者分为PSD组和Non-PSD组,其中HAMD抑郁量表评分〉8分为PSD组,HAMD抑郁量表评分8分为Non-PSD组,两组病人均进行3.0T西门子MRI常规及弥散张量成像扫描(DTI)。首先,使用PANDA软件对所有患者的DTI数据进行头动和涡流效正、标准化等预处理,然后,利用解剖学自动标记模板(Automated anatomical labeling,AAL)将整个大脑划分为90个区域,使用PANDA对全脑白质进行确定性纤维追踪,最后使用gretna软件分析两组被试脑白质网络属性,并对其进行两样本t检验。本研究主要关注以下几个问题:(1)基于脑网络全局性属性角度研究,包括PSD组和Non-PSD组脑白质网网络的小世界属性(sigma)、聚类系数(Cp)、标准特征路径长度(Lambda)、特征路径长度(Lp)进行双样本t检验。(2)基于网络效率角度研究,包括PSD组和Non-PSD组全局效率(Eglob)、局部效率(Eloc)进行双样本t检验。(3)基于网络局部效率角度研究,包括对PSD组和Non-PSD组每个患者的90个脑区分别进行聚类系数和局部效率属值进行双样本t检验。结果:(1)PSD组及Non-PSD组均具有小世界属性,但PSD组小世界属性小于Non-PSD组,两组差异有统计学意义(P0.05),同时,PSD聚类系数(Cp)也升高,两种比较差异有统计学意义(P0.05),另外,PSD组的归一化的聚类系数(Gamma);标准特征路径长度(Lambda)、特征路径长度Lp较Non-PSD组低,差异有统计学意义(P0.05)。(2)PSD组网络效率的全局效率(Eglob)、局部效率(Eloc)较Non-PSD组升高,两组差异有统计学意义(P0.05)。(3)节点效率改变的脑区:两侧眶部额下回、左内侧额上回、两侧前扣带和旁扣带脑回、左后扣带回、右海马、右丘脑、右颞中回节点效率增高。节点聚类系数(Cp)改变脑区:右侧嗅皮质节点效率升高,两侧背外侧额上回、右眶部额上回、左额中回、左岛盖部额下回、左三角部额下回、左内侧和旁扣带脑回、右海马、右角回、左尾状核、左豆状苍白球、左颞上回节点效率升高,这些异常节点基本属于默认网络(DMN)结论(1)PSD患者仍存在小世界属性,但小世界属性损坏,趋向随机网络;(2)凸显网络(SN)、默认网络(DMN),中央执行控制网络(CEN)拓扑异常,容易导致PSD发生。
[Abstract]:Objective: to explore the possible changes of the network topological properties of the white matter structure in patients with post stroke depression (PSD) by using DTI technique and graph theory, and to search for the image markers of PSD brain network. To provide an objective diagnostic index for early PSD diagnosis and evaluation of curative effect. Methods: according to DSM-IV criteria for diagnosis of post-stroke depression, 48 patients with acute ischemic stroke within 2 weeks were divided into PSD group and Non-PSD group by 24 items of Hamilton depression scale. The HAMD depression scale score > 8 was divided into PSD group and Hamd depression scale score. The patients in both groups were divided into Non-PSD group and 3. 0T Siemens MRI routine and diffused Zhang Liang imaging scanning. First, the DTI data of all patients were preprocessed with PANDA software, such as head movement, eddy current effect, standardization, etc. Then, the whole brain was divided into 90 regions using automated anatomical labeling template. PANDA was used to track the white matter of the whole brain, and gretna software was used to analyze the properties of the white matter network in the two groups of subjects. This study focuses on the following questions: 1) based on the global attributes of brain networks, The small world properties of white matter network including PSD group and Non-PSD group were studied from the point of view of network efficiency, such as sigma, clustering coefficient, standard characteristic path length, Lambdaa, characteristic path length, and characteristic path length. The global efficiency of PSD group and Non-PSD group is analyzed by double sample t-test. 3) based on the local efficiency of the network. The clustering coefficient and local efficiency of 90 brain regions in PSD group and Non-PSD group were tested by double sample t test. Results both Non-PSD and Non-PSD had small-world attributes, but the small-world attributes in PSD group were smaller than those in Non-PSD group. The difference between the two groups was statistically significant (P 0.05), and the clustering coefficient (CP) of PSD was also higher than that of Non-PSD group. The difference between the two groups was statistically significant (P 0.05), in addition, the normalized clustering coefficients of PSD group and the standard characteristic path length (LP) were lower than those of Non-PSD group. The difference was statistically significant in the global efficiency and local efficiency of the P0.05PSD group compared with the Non-PSD group. The difference between the two groups was statistically significant in the brain areas where the nodal efficiency changed: bilateral orbital inferior frontal gyrus, left medial superior frontal gyrus, and the left medial superior frontal gyrus. The efficiency of bilateral anterior cingulate and paracentric gyrus, left posterior cingulate gyrus, right hippocampus, right thalamus and right middle temporal gyrus were increased. The nodal clustering coefficient (CP) changed the brain region: the efficiency of the right olfactory cortex node increased, bilateral dorsolateral superior frontal gyrus, right orbital superior frontal gyrus, left middle frontal gyrus, left subfrontal gyrus of the tegmental part, subfrontal gyrus of left triangular part, left medial and paracarpal cingulate gyrus, right hippocampus, left lateral superior frontal gyrus, left middle frontal gyrus, left subfrontal gyrus, left medial and accessory cingulate gyrus, right hippocampus. Right corner gyrus, left caudate nucleus, left lentiform globus pallidus, left superior temporal gyrus node efficiency increased, these abnormal nodes basically belong to the default network of DMN. The trend of random network is to highlight the network, default network, central executive control network (CENs) topology anomalies, easy to lead to the occurrence of PSD.
【学位授予单位】：广西医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R743.3;R749.4

【相似文献】