重性抑郁症（Major depressive disorder, MDD）是一种非常普遍的心理疾病。根据世界卫生组织的研究报告和中国卫生部的有关资料,我国约7%的人患有MDD，给社会和家庭带来了巨大的负担。MDD具备发病原因复杂、病情表现多样的特点，给诊断带来了极大的挑战。近年来，大量研究报道抑郁症大脑存在大范围的功能异常。然而，多数研究受限于样本数量与研究方法，所得到的结果无法相互印证。鉴于此，本研究利用多中心静息态功能磁共振影像数据，结合脑网络理论，分别从静态和动态脑网络的角度挖掘大样本条件下抑郁症影像组学的异常特点，具体而言：

    在静态脑网络抑郁症组学异常特点研究中，本课题结合不同的滤波策略探索与验证不同频带条件下体素级多中心静态脑功能网络功能连接强度的空间分布特点，从更精细的空间尺度搜索具备稳定空间分布差异的频带。研究发现重度MDD患者大脑功能连接强度在左侧顶下皮层、颞下皮层、中央前皮层、梭状回皮层以及双侧的楔前叶等区域存在显著的频率依赖性异常，并且低频滤波(0.01-0.06 Hz)下的异常最为明显。通过对功能连接强度具备显著频率依赖的区域进行细化的功能连接分析，研究发现这些依赖于频率的连接强度改变主要由中长距离连接的出现异常导致，并且这些连接的改变是依赖于脑网络的。值得注意的是，研究发现在高频滤波(0.16-0.24 Hz)条件下，患者左侧楔前叶的连接强度异常降低与病程显著负相关。

    在动态脑网络抑郁症组学异常特点研究中，本课题利用隐马尔可夫模型获取多中心可重复的动态脑状态特征，从时间尺度更精细的发掘抑郁症动态脑网络具备4种不同的脑状态。并结合常模建模方法，对动态脑状态特征进行进一步划分，获取具备区分度的动态脑状态抑郁症亚型，发现动态脑状态可以按照激活强弱划分为较强激活及较弱激活两种亚型。两种动态脑状态亚型间存在汉密尔顿评分及病程长度的差异。通过与临床变量的相关性分析，研究进一步发现大脑的动态激活较弱的抑郁症亚型患者的状态占比及切换特征与抑郁病程显著负相关。    

    综上，本研究通过多中心大样本静息态功能磁共振数据，结合静态与动态脑网络算法，为抑郁症影像组学的异常特点提供了结果支持。

    Depression is a very common mental disorder. According to research reports from world health organization and relevant data from the Chinese Ministry of Health, about 7% of people in China have MDD, which has brought a huge burden to society and families. MDD is characterized by complex etiology and various manifestations, which poses a great challenge for diagnosis. In recent years, a large number of studies have reported widespread functional abnormalities in the brain of depression. However, most studies are limited by the number of samples, and the results obtained cannot be mutually verified. At present, there is still a lack of research on brain abnormalities in depression under large-sample conditions. In view of this, this study uses multi-center resting-state functional magnetic resonance imaging data combined with brain network theory to explore the abnormal imaging characteristics of depression under multi-center large-sample conditions from the perspective of static and dynamic brain networks, specifically：

    In the study of abnormal features of depression in static brain networks, this project combined different filtering strategies to explore and verify the spatial distribution characteristics of functional connectivity strength in voxel-level multi-center static brain networks under different frequency bands. We reported significant frequency-dependent connectome alterations in MDD in left inferior parietal, inferior temporal, precentral, and fusiform cortices and bilateral precuneus. These frequency-dependent connectome alterations are mainly derived by abnormalities of medium- and long-distance connections and are brain network dependent. Moreover, the connectome alteration of left precuneus in high frequency band (0.16–0.24 Hz) is significantly associated with illness duration.

    In the study of abnormal characteristics in dynamic brain networks of depression, this project utilizes hidden Markov model to acquire repeatable dynamic brain states features from a finer time scale. Four different activation states of dynamic brain networks of depression were identified across all imaging centers. Combined with the normative modeling method, two dynamic brain states subtypes were further identified according to the strength of activation: strong activation and weak activation. There were differences in Hamilton depression rating score and illness duration between the two dynamic brain states subtypes. Through correlation analysis with clinical variables, the study found that the proportion of states and transition features of the dynamic activation weaker subtype of depression patients were significantly negatively correlated with the depression duration.

    In summary, this study provides results support to the abnormal features of depression imaging genetics by multi-center large-sample resting-state functional magnetic resonance imaging data combined with static and dynamic brain network algorithms.