Brain Connectivity of MCI Based on Dynamic High-Order Correlation

To address the limitations of traditional functional connectivity and existing high-order dynamic functional connectivity in neglecting temporal information, this paper introduces a novel method called Dynamic High-Order Time Correlation (DHTC). This method utilizes a dynamic high-order time correlation matrix to analyze subjects with Mild Cognitive Impairment (MCI) and Normal Control (NC) subjects through classification tests and difference analysis. Initially, the paper proposes and defines the time correlation method. Then, the optimal parameter values are determined using the objective function auxiliary method. Subsequently, the sliding window method is employed to extract the low-order time correlation matrix. Finally, all low-order time correlation matrices are further processed to derive the high-order time correlation matrix. Experimental results demonstrate that compared to traditional functional connectivity matrices, the classification accuracy of the dynamic high-order time correlation matrix proposed in this paper is enhanced by approximately 12%, reaching 75.25%. The high values of the high-order temporal correlation matrix intensity of the MCI group and the NC group were mainly concentrated in four regions, and significant differences were observed in the DHTC matrix between the two groups of subjects.