Frequency-specific Effective Connectivity in Subjects with Cerebral Infarction as Revealed by NIRS Method.

A connectivity-based approach can highlight the network reorganization in the chronic phases after stroke and contributes to the development of therapeutic interventions. Using dynamic Bayesian inference, this study aimed to assess the effective connectivity (EC) in various frequency bands through the near-infrared spectroscopy (NIRS) method in subjects with cerebral infarction (CI). A phase-coupling model was established based on phase information extracted using the wavelet transform of NIRS signals. Coupling strength and the main coupling direction were estimated using dynamic Bayesian inference. Wilcoxon test and chi-square test were used to determine the significant difference in EC between two groups. Results showed that the coupling strength of the EC in the CI group significantly decreased relative to that in the healthy group. The decrease was most significant in the frequency intervals IV (0.021 Hz-0.052 Hz; p = 0.0006) and VI (0.005 Hz-0.095 Hz; p = 0.0028). The main coupling direction changed from the right prefrontal cortex to the right motor cortex and left motor cortex in the frequency intervals IV (p = 0.041, p = 0.047) and II (p = 0.0017, p = 0.0036), respectively. The EC decreased or was even lost significantly in the EC map of the CI group. Experimental results indicated that information propagation was blocked in the CI group than in the healthy group and resulted in the decreased coupling strength and connectivity loss. The main coupling direction of the motor section changed from driving into being driven because of the degradation of limb movement function.