The Research Of Finger -Movement-Related MEG Signal Processing

Human hands are the most dexterous locomotors through long-term evolution. Due to the large hand projection in motor cortex, it can deduce that the hand movement relies on complicated neural mechanism. As a non-invasive technique applied for the function mapping of human brain, magnetoencephalography (MEG) detects precise neural electrophysiological activities both spatially and temporally, which is undoubted the quite tool to investigate central process mechanisms of brain. The research of cortical control mechanisms for hand movements with MEG detection is not only a challenge in engineering, its results can also be applied to clinical diagnose, rehabilitation program, prosthesis control, and so on.An effective processing of MEG signal can be the premise of this mechanism research. Accordingly, this present work is aimed to systematically explore the MEG signal processing method, and a preliminary knowledge to characterize the cortical neural activities associated with finger movements is conducted as well.In the investigation of MEG signal analyzing, preprocessing was discussed first, with the emphasis on artifacts removing. The traditional artifact rejecting method based on independent component analysis (ICA) was improved by adding an automatic independent component (IC) detection module to the ICA model. IC detection module detected artifactual components mainly based on the analysis of statistical and spectral characteristics of each IC, and it helped realizing an automatic method for MEG artifacts identification eventually. Time frequency characters of power were analyzed after MEG preprocessing. Window length variable time-frequency power spectrum estimation with single taper for signals in low-frequency band and with multitapers for signals in high-frequency band were realized here based on the studies of time frequency representations of power (TFRs). Specially, characters of TFRs in three specific frequency bands, the mu (8~14Hz), the beta (15~30Hz) and the low gamma (30~60Hz) bands were analyzed in this paper under a self-paced button press task. Finally, a functional source separation (FSS) method was achieved to extract source activities associated with button pressing from MEG signals, which was obtained by adding a functional constraint to the cost function of a basic ICA model, defined according to prior information we had acquired by TFRs.Neuromagnetic activity was recorded from one healthy adult during a self-paced button press task. TFRs were calculated from this MEG dataset and the results showed that, (1) MEG power evoked by button pressing task was mainly concentrated in the energy distribution below 35Hz; 40Hz to 70Hz band also showed neural activities, but much weaker compared with low frequency bands. (2) Low frequency bands (both mu and beta rhythms) showed strong suppression in power (ERD) during movement preparation and execution; for the beta rhythm, ERD lasted shortly after the termination of movement, and was immediately followed by a large increase in power (ERS), which lasted about 1 second; mu rhythm still showed ERD after movement termination and little to no post-movement ERS was observed; (3) High frequency bands (low gamma bands) showed no notable activities during movement preparation, but increased power was observed following movement onset (4) For the location of each rhythm, mu band activities was observed in bilateral central and parietal channels, corresponding to cortical sensorimotor cortex, with contralateral activities greater; beta rhythm activities concentrated in contralateral central and parietal channels, and active area after button press located anterior to area before button press, which is corresponding to primary motor cortex; low gamma activities were observed in bilateral central channels, and activities were greater contralateral to the side of movement.Discussion about physiological significance of ERD and ERS in each frequency band suggested that mu band ERD in sensorimotor area could be a symbol of motor cortex activation; beta band ERD and ERS described the whole process of finger movement preparation, execution and recovery after movement; low gamma band ERS could be related to the integration of a wide range of information processing.Correlation coefficient was calculated between finger movement related functional source (FS) extracted by the current study and magnetic source extracted by SAM software in CTF MEG system, and the result showed they were significantly correlated, which proved the validity of functional source separation (FSS) method. FS activities mainly concentrated in the posterior part of contralateral central channels, roughly corresponding to the motor cortex in precentral gyrus. Analysis of FS in time domain didn't show any characteristic wave as appeared in sensory evoked magnetic fields in the current self-paced button press task, but the temporal information provided by FSS method could be a basis of exploring cortical control timing mechanisms associated with finger movements and extracting time frequency characteristics of the functional source.