Abstract
[Objective] To evaluate our proposed method, which incorporate results of sensor signal analyses as weighting of recorded data prior to execution of source estimation, with simulated data by comparing with conventional method. The simulated data were generated by assuming hippocampal-neocortical circuits, which had a functional coupling between hippocampal theta activity and neocortical gamma activity.
[Method] First, we formulated analysis procedure to apply our proposed method to cross-frequency correlation analysis to detect brain portions connected each other with different frequency range. In this procedure, Spearman's rank correlation was used to evaluate cross-frequency correlation so as to execute statistical evaluation of recorded data which can not be limited to normal distribution. Sensor signal analyses with cross-frequency correlation maps and correlation topography were utilized to identify significant characteristics (e.g. frequency pair at which significant cross-frequency correlation were observed.). And the results of the sensor signal analysis were used to weight recorded signals. Then source estimation was executed by standardized low-resolution brain electromagnetic tomography (sLORETA) in frequency domain and sLORETA modified for quantifiable method (sLORETA-qm) was employed to obtain averaged source image. Second, we simulated connection between hippocampal theta activity and correlated neo-cortical gamma activity. Three sources (s1: hippocampal theta activity, s2: neocortical gamma activity correlated to phase of s1, s3: non-correlated occipital theta activity) were placed by referring MR image of normal subject and sensor signals were generated by four layers spherical model and additional white noise of 16 or 4 of signal to noise ratio. Then, the formulated procedure were applied to the simulated data to detect and estimate the correlated sources.
[Results] Our proposed method could estimate only correlated sources corresponded to the correlated sources (s1 and s2), while conventional method estimated non-correlated source corresponded non-correlated source (s3) as if correlated source.
[Conclusion] We demonstrated that our proposed method could extract the sources of specific neural activities in the brain such as cross-frequency correlation. Our proposed method will become a useful and helpful tool for obtaining detailed information on neural activities.
