Wen Shipeng, Wang Jingru, Liu Wenjie, Meng Xianglian, Jiao Zhuqing
Wangzheng School of Microelectronics, Changzhou University, Changzhou, China.
School of Computer Information and Engineering, Changzhou Institute of Technology, Changzhou, China.
Front Neurosci. 2025 Jun 13;19:1597777. doi: 10.3389/fnins.2025.1597777. eCollection 2025.
Alzheimer's Disease (AD) is a progressive neurodegenerative disorder, with Mild Cognitive Impairment (MCI) often serving as a prodromal stage. Early detection of MCI is critical for timely intervention.
Dynamic Functional Connectivity analysis reveals temporal dynamics obscured by static functional connectivity, making it valuable for analyzing and classifying psychiatric disorders. This study proposes a novel spatio-temporal approach for analyzing dynamic brain networks using resting-state fMRI. The method was evaluated on data from 85 subjects (33 healthy controls, 29 Early Mild Cognitive Impairment (EMCI), 23 AD) from the ADNI dataset.
Our model outperformed existing techniques, achieving 83.9% accuracy and 83.1% AUC in distinguishing AD from healthy controls.
In addition to improved classification performance, key affected regions such as left hippocampus, the right amygdala, the left inferior parietal lobe, the left olfactory cortex, the right precuneus, and the insula, were identified-areas known to be associated with memory function and early Alzheimer's pathology. These findings suggest that dynamic connectivity analysis holds promise for non-invasive and interpretable early-stage diagnosis of AD.
阿尔茨海默病（AD）是一种进行性神经退行性疾病，轻度认知障碍（MCI）常作为其前驱阶段。早期检测MCI对于及时干预至关重要。
动态功能连接分析揭示了静态功能连接所掩盖的时间动态，这使其在分析和分类精神疾病方面具有价值。本研究提出了一种使用静息态功能磁共振成像（fMRI）分析动态脑网络的新颖时空方法。该方法在来自阿尔茨海默病神经影像学倡议（ADNI）数据集的85名受试者（33名健康对照、29名早期轻度认知障碍（EMCI）、23名AD患者）的数据上进行了评估。
我们的模型优于现有技术，在区分AD与健康对照方面，准确率达到83.9%，曲线下面积（AUC）为83.1%。
除了提高分类性能外，还识别出了关键受影响区域，如左侧海马体、右侧杏仁核、左侧顶下小叶、左侧嗅觉皮层、右侧楔前叶和脑岛，这些区域已知与记忆功能和早期阿尔茨海默病病理学相关。这些发现表明，动态连接分析有望用于AD的非侵入性和可解释的早期诊断。
