Synchrosqueezing algorithm application in TMS-EEG analysis

Objective: In this study, we attempted to develop an accurate and effective rhythm analysis tool which gives spectrum of Transcranial magnetic stimulation (TMS) evoked Electroencephalography (EEG) oscillation with both high resolution in time and frequency domain for understanding the TMS induced EEG and relative neural mechanisms. Methods: We investigated the possibility of applying the synchrosqueezing transform spectrum for TMS evoked EEG analysis. Synchrosqueezing is a special case of reallocation method which aims to give a high resolution and readable time-frequency representation. In order to evaluate the performance of it, we compared it with the Hilbert-Huang transform (HHT) and Wavelet transform (WT). Simulated data based tests were applied to compare time and frequency resolution of spectrums from these methods. Then we used a modified neural mass model to simulate oscillations and the propagation evoked by TMS. The model could generate natural frequency of Brodmann Area 19, 7 and 6 evoked by TMS. Spectrums measured by the three algorithms were compared. Results: It showed that the synchrosqueezing outperforms than HHT and WT in detecting transient activity. Synchrosqueezing offers clearer instantaneous frequency information and it was more sensitive in detecting the TMS evoked oscillations. Conclusions: It suggests that synchroqueezing is a reliable analytical tool and it outperforms than HHT and WT in exploring the time-frequency features of TMS-evoked oscillations. Significance: It will contribute to investigating the TMS evoked oscillations and giving more details of the evoked activity.