Power Spectral Density Analysis of Long

Background. Prognostic measures of long-term motor recovery are important in patients with stroke presenting with severe hemiplegia. Objective. We aimed to investigate whether initial power spectral density (PSD) analysis of resting-state functional magnetic resonance (fMRI) data can provide a sensitive prognostic predictor in patients with subacute stroke with severe hand disability. Methods. Twelve patients with good recovery, 14 patients with poor recovery, and 12 healthy subjects were included. PSD analysis was performed using resting-state fMRI data. Contralesional and ipsilesional PSD in the motor cortex were measured. Pearson correlation analysis was performed to assess a possible association between the difference in ipsilesional versus contralesional PSD and motor outcomes. A receiver operating characteristic (ROC) curve was constructed to estimate the discriminative value of the difference between the ipsilesional PSD and the contralesional PSD for good versus poor recovery. Results. There were no differences in PSD between the contralesional and ipsilesional hemispheres in the good recovery group (P = .77). In contrast, there were significant differences in PSD between the 2 hemispheres in the poor recovery group (P = .07). The difference in PSD between the 2 hemispheres had a positive correlation with post Brunnstrom stage scores. ROC analysis showed that the difference in PSD between the 2 hemispheres was sensitive in discriminating good versus poor recovery. Conclusion. The present study suggests that PSD in the motor cortex may be a sensitive predictor of late-onset motor recovery following stroke