Dynamic functional coupling of high resolution EEG potentials related to unilateral internally triggered one-digit movements

A. Urbano, C. Babiloni, P. Onorati, F. Babiloni*

* Institute of Human Physiology, Division of High Resolution EEG (CIMS),

University of Rome ‘La Sapienza,’ P.le Aldo Moro 5, 00185 Rome, Italy

Electroencephalography and clinical Neurophysiology 106 (1998) 477–487

Abstract

Between-electrode cross-covariances of delta (0–3 Hz)- and theta (4–7 Hz)-filtered high resolution EEG potentials related to

preparation, initiation, and execution of human unilateral internally triggered one-digit movements were computed to

investigate statistical dynamic coupling between these potentials. Significant (P , 0.05, Bonferroni-corrected) cross-

covariances were calculated between electrodes of lateral and median scalp regions. For both delta- and theta-bandpassed

potentials, covariance modeling indicated a shifting functional coupling between contralateral and ipsilateral frontal-

central-parietal scalp regions and between these two regions and the median frontalcentral scalp region from the

reparation to the execution of the movement (P , 0.05). A maximum inward functional coupling of the contralateral with the

ipsilateral frontal-central-parietal scalp region was modeled during the preparation and initiation of the movement,

and a maximum outward functional coupling during the movement execution. Furthermore, for theta-bandpassed

potentials, rapidly oscillating inward and outward relationships were modeled between the contralateral frontal-central-

parietal scalp region and the median frontal-central scalp region across the preparation, initiation, and execution of the

movement. We speculate that these cross-covariance relationships might reflect an oscillating dynamic functional coupling

of primary sensorimotor and supplementary motor areas during the planning, starting, and performance of unilateral

movement. The involvement of these cortical areas is supported by the observation that averaged spatially enhanced delta-

and theta-bandpassed potentials were computed from the scalp regions where task-related electrical activation of primary

sensorimotor areas and supplementary motor area was roughly represented.

High resolution EEG data used for the present crosscross covariance analysis were obtained

from a previous functional topography study of spatially enhanced scalp potentials related to

unilateral voluntary movements (Urbano et al., 1996 The EEG recording (0.1–100 Hz bandpass)

was performed with a 128 gold-plated electrode cup from 4 right-handed informed male

volunteers (28–36 years old) executing self-initiated right middle finger extensions at irregular

intervals of 6–20 s. The electrode positions andreference landmarks were digitized for

subsequent integration with magnetic resonance images. Electrooculogram

(0.1–100 Hz bandpass) and surface electromyograms (1– 100 Hz bandpass) from bilateral

extensor digitorum muscles were also collected. All data were digitized at 400 Hz sampling rate.

Approximately 200 artifact-free single trials were averaged for each subject. The selected single trials presented neither eye movements and blinking nor concomitant mirror movements. Averaged potentials were spatially enhanced with the surface Laplacian estimate of the potential over a realistic magnetic resonance constructed subject’s scalp surface model (2–3 cm vs. the 6–9 cm spatial resolution of the traditional EEG; Babiloni et al., 1995, 1996). The veraged spatially enhanced potentials presented components related to the preparation (readiness potential, RP), initiation (pre-motion positivity and motor potential, PMP-MP), and execution (first and second movementrelated responses, MRR1 and MRR2) of the movement (Urbano et al., 1996). The early RP presented no consistent peak.

These potentials were filtered in the delta (0–3 Hz) and theta (4–7 Hz) frequency bands with a gaussian zero phase finite impulse response-type technique that introduced no time shift in the filtered EEG wave forms. Delta- and theta-filtering was carried out because functional coupling of neuronal populations involved in motor control is reflected well in the synchronization of their averaged macropotentials in the low frequency (0–7 Hz) band (Gevins and Cutillo, 1993; Gevins et al., 1994). Also, the strongest variations of instantaneous coherence between pseudoinverse-derived dipole timeseries of movementrelated cortical activity have been recognized in the low frequency bands (Thatcher et al., 1994; Thatcher, 1995).

Lagged covariance functions were computed for the averaged spatially enhanced low-bandpassed movement-related potentials following the procedure described by Gevins et al. (1989). Filtered EEG data were sub-sampled from 400 Hz to 12 Hz for the delta band and to 28 Hz for the theta band to avoid over-sampling effects. The time interval for the computation of each covariance function had the width of one period of the band center frequency. The width of this interval for the delta band was about 700 ms (one period at 1.5 Hz), while that for the theta band was about 180 ms (one period at 5.5 Hz). Covariance functions were computed up to a lag time of one-half period of the highest frequency for each band.

Patterns of statistically significant (P , 0.05) between-electrode covariances of averaged spatially enhanced delta- and theta-filtered movementrelated EEG potentials (RP, PMP-MP, MRR1, and MRR2) estimated from lateral frontal-central-parietal and median frontal-central scalp regions of subject 1. Arrows point from leading to lagging scalp electrode sites.

Across-subject mean zero and non-zero lag covariance strength of averaged, spatially enhanced delta- and theta-filtered movement-related EEG potentials (RP, PMP-MP, MRR1, and MRR2) computed between contralateral and ipsilateral frontal-central-parietal scalp regions (C-I), contralateral frontalcentral-parietal and median frontal-central scalp regions (C-M), and median frontal-central and ipsilateral frontal-central-parietal scalp regions (M-I).

Across-subject mean inward and outward (arrows) of the non-zero lag covariances previously plotted

Inward minus outward across-subject mean total covariancestrength of the averaged spatially enhanced delta- and theta-filtered movement-related EEG potentials (RP, PMP-MP, MRR1, and MRR2) computed from contralateral (C) and ipsilateral (I) frontal-central-parietal scalp regions and median frontal-central scalp region (M). Positive and negative values indicate prevailing inward and outward total covariance strength, respectively.

Patterns of maximum across-subject mean inward-outward non-zero lag covariance strength between averaged, spatially enhanced delta- and thetafiltered movement-related EEG potentials (RP, PMP-MP, MRR1, and MRR2) for the 3 scalp regions of interest (C, M, and I). The thickness of the arrows is proportional to the covariance strength. Arrow thickness is normalized for delta- and theta-filtered EEG potentials.

Further studies should be carried out to compare the functional coupling patterns computed from high-frequency EEG rhythms and averaged movement-related EEG potentials. These studies might provide useful information on the cortical networks subserving the preparation, initiation, and execution of movement. Recently, it has been suggested that averaged movementrelated EEG potentials would represent increased excitability of the cortical motor areas, whereas variations of highfrequency oscillatory EEG responses would signal a participation of these cortical areas independent of their excitatory or inhibitory nature