Coupled Intrinsic Connectivity Distribution Analysis: A Method for Exploratory Connectivity Analysis of Paired fMRI Data

We present a novel voxel-based connectivity approach for paired functional magnetic resonance imaging (fMRI) data collected under two different conditions labeled the Coupled Intrinsic Connectivity Distribution (coupled-ICD). Our proposed method jointly models both conditions to incorporate additional paired information into the connectivity metric. Voxel-based connectivity holds promise as a clinical tool to characterize a wide range of neurological and psychiatric diseases, and monitor their treatment. As such, examining paired connectivity data such as scans acquired pre- and post-intervention is an important application for connectivity methodologically. When presented with data from paired conditions, conventional voxel-based methods analyze each condition separately. However, summarizing each connection separately can misrepresent patterns of changes in connectivity. We show that commonly used methods can underestimate functional changes and subsequently introduce and evaluate our solution to this problem, the coupled-ICD metric, using two studies: 1) healthy controls scanned awake and under anesthesia, and 2) cocaine-dependent subjects and healthy controls scanned while being presented with relaxing or drug-related imagery cues. The coupled-ICD approach detected differences between paired conditions in similar brain regions as the conventional approaches while also revealing additional changes in regions not identified using conventional voxel-based connectivity analyses. Follow-up seed-based analyses on data independent from the voxel-based results also showed connectivity differences between conditions in regions detected by coupled-ICD. This approach of jointly analyzing paired resting-state scans provides a new and important tool with many applications for clinical and basic neuroscience research.本研究提出了一种新颖的基于体素的连接性分析方法，用于处理两种不同实验条件下采集的配对功能磁共振成像（functional magnetic resonance imaging，fMRI）数据，该方法被命名为耦合固有连接分布（Coupled Intrinsic Connectivity Distribution，coupled-ICD）。我们所提出的方法可对两种实验条件进行联合建模，将额外的配对信息融入连接性指标之中。基于体素的连接性分析有望成为一种临床工具，用于表征多种神经及精神疾病，并监测疾病治疗进程。因此，对配对连接性数据（如干预前后采集的扫描图像）开展分析，是连接性分析方法的重要应用方向之一。当面对配对条件下的数据集时，传统基于体素的分析方法会分别对每种条件进行独立处理。然而，单独汇总每个连接的信息可能会误判连接性的变化模式。本研究表明，常用的传统方法可能会低估功能连接的变化，为此我们提出并评估了针对该问题的解决方案——coupled-ICD指标，并通过两项研究对其进行验证：1）对健康对照组分别在清醒状态与麻醉状态下进行扫描；2）对可卡因依赖受试者与健康对照组在观看放松或与药物相关的意象线索时进行扫描。coupled-ICD方法在与传统方法相似的脑区中检测到了配对条件间的差异，同时还揭示了传统基于体素的连接性分析未能识别的脑区变化。针对独立于基于体素分析结果的数据集进行的后续基于种子点的分析，同样在coupled-ICD检测到的脑区中发现了条件间的连接性差异。这种联合分析配对静息态扫描数据的方法，为临床与基础神经科学研究提供了一种全新且重要的工具，拥有广泛的应用场景。