Estimating multiple sources in somatosensory area to SEF of finger stimulus using SVD and time-frequency analysis

In order to discriminate multiple sources in somatosensory area of human cortex, we measured SEF (somatosensory evoked field) elicited by electrical stimulation to three different fingers (the thumb, fourth finger, and little finger), respectively. First, we applied singular value decomposition (SVD) method to the measured SEF data to discriminate second somatosensory (SII) activity from primary somatosensory (SI) activity overlapping in time. Time-frequency analysis (wavelet transforms, WT) was also applied to the calculated SEF data to examine frequency component of the SEF data. It was shown that the SI activity of the measured SEF data was represented with reconstructed SEF data with first singular value. And the SII activity was detected by the reconstructed SEF data with the fourth and fifth singular value in the thumb and fourth finger stimulation, respectively. Second, in the result of time-frequency analysis, distribution of dominant power spectrum of the SEF to the thumb (median nerve) stimulation was different to that of the little finger (ulna nerve) stimulation (thumb: 16-18Hz, little finger: 13-15Hz). In the result of fourth finger stimulation, the frequency of dominant power spectrum existed between that of the thumb and little finger stimulation. Source localization was done by the SEF data to the fourth finger stimulation, having different frequency ranges (13-15Hz and 16-18Hz band-pass filter). We conclude that the proposed method, combining the SVD method and the time-frequency analysis, is useful for discriminating multiple source activities of the somatosensory area overlapping in time.