- Email: [email protected]
Statistical probability mapping reveals high-frequency magnetoencephalographic activity in supplementary motor area during self-paced finger movements
Neuroscience Letters 283 (2000) 81±84 www.elsevier.com/locate/neulet
Statistical probability mapping reveals high-frequency magnetoencephalographic activity in supplementary motor area during self-paced ®nger movements Jochen Kaiser a,*, Werner Lutzenberger a, Hubert Preissl a, Dirk Mosshammer a, Niels Birbaumer a, b a
MEG-Center, Institute of Medical Psychology and Behavioral Neurobiology, University of TuÈbingen, Gartenstrasse 29, 72074 TuÈbingen, Germany b Department of General Psychology, University of Padova, Padova, Italy Received 9 December 1999; received in revised form 17 February 2000; accepted 17 February 2000
Abstract Investigations of both haemodynamic and electroencephalographic measures of brain activity have demonstrated supplementary motor area (SMA) involvement in self-paced ®nger movements. In contrast, analysis of magnetoencephalographic (MEG) signals in the time domain has usually failed to detect SMA activity in healthy individuals. We investigated oscillatory MEG activity in 12 normal adults during (a) a self-paced, complex sequence of ®nger movements and (b) a simple ®nger opposition task paced externally by tactile stimuli presented to the contralateral thumb. Statistical probability mapping revealed enhanced non-phase-locked spectral amplitudes in the 22±28 Hz range over bilateral frontal cortex during self-paced as compared to externally cued ®nger movements. This activity may re¯ect recruitment of cell assemblies in SMA during self-paced, complex movements. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Magnetoencephalography; Oscillatory activity; Statistical probability mapping; Finger movements; Self-paced; Supplementary motor area; Human subjects
During movements, contralateral primary sensorimotor areas exhibit increased regional cerebral blood ¯ow as assessed with positron emission tomography [17] and functional magnetic resonance imaging [16]. In addition, certain kinds of motor responses are accompanied by activations of more anterior regions such as the supplementary motor area (SMA) in the medial and dorsal parts of Brodmann area 6. There is some debate on whether SMA is involved only in the generation of complex movements [16,17] or whether it is also activated when simple movements are performed [1,2,5,20]. More importantly, most studies have observed SMA activation only when movements were self-paced or internally cued [4,10,16,19], although there have been contradictory ®ndings of SMA involvement also in externally paced motor responses [5]. Electroencephalographic (EEG) recordings have * Corresponding author. Tel.: 149-7071-297 4224; fax: 1497071-29 5956. E-mail address: [email protected] (J. Kaiser)
suggested that the earlier component of the readiness potential preceding self-initiated movements may be generated in bilateral SMA and the later part in contralateral primary motor cortex [7,8]; however, there was no such time lag in intracranial recordings [1]. In contrast, most magnetoencephalographic (MEG) studies analyzing signals in the time domain have failed to detect SMA activity [8,11]. This has been attributed to SMA anatomy: while radial sources in the dorsal parts of Brodmann area 6 may not generate detectable magnetic ®elds, opposed tangential sources in the bilateral medial parts of this area may cancel each other out [12]. The present study investigated oscillatory MEG responses to externally paced versus self-paced ®nger movements. Such responses have been recorded previously from primary sensorimotor areas in the alpha, beta and gamma ranges [6,9,13]. Here we used a statistical probability mapping approach, enabling the disclosure of signi®cant differences in high-frequency signals of low amplitude that would not be detectable in the analysis of evoked magnetic ®elds. Twelve right-handed subjects were tested in a magneti-
0304-3940/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S03 04 - 394 0( 0 0) 00 92 1- 6
82
J. Kaiser et al. / Neuroscience Letters 283 (2000) 81±84
cally shielded room (Vakuum-Schmelze, Hanau, Germany). MEG was recorded while tactile stimuli were delivered to the right thumb with a pneumatic stimulator (BTI Inc., San Diego, USA) containing a membrane driven by air pressure changes (stimulus duration: 100 ms). 89% of the stimuli were presented at randomized inter-stimulus intervals (ISI) varying between 400 and 500 ms, whereas the remaining 11% were presented at randomized ISI between 1450 and 1500 ms to avoid rapid habituation. 400 stimuli were presented in each of the following two conditions: (a) paced ®nger movements: subjects were instructed to respond as quickly as possible to the tactile stimuli by touching the left thumb with the left index ®nger, and (b) self-paced ®nger movements: while instructed to ignore the tactile stimuli, subjects continuously performed the following task: they had to oppose sequentially the left thumb to the left index ®nger once, to the middle ®nger twice, to the ring ®nger three times and to the little ®nger four times, and then repeat this sequence in reverse order. Both ®nger movement tasks were trained prior to MEG recording. The sequence of these two conditions was counterbalanced across subjects. At the end of the experiment, subjects were asked to rate (a) how dif®cult they found the tasks on scales ranging from 1 very easy to 10 very dif®cult, and (b) how accurately they thought they had performed the tasks on scales from 1 very poorly to 10 very accurately. MEG was collected with a whole head magnetometer (CTF Inc., Vancouver, Canada) comprising 143 hardware ®rst order magnetic gradiometers (5 cm baseline, average distance between sensors: 2.5 cm, sampling rate: 312.5 Hz, high-frequency cutoff: 100 Hz). The subject's head position was determined with localization coils ®xed at the nasion and the preauricular points both at the beginning and end of each experimental condition. Recordings with head movements exceeding 0.5 cm were discarded. The threshold for rejecting trials with eye movement artifacts was 1.3 pT in frontal channels. Epoch length was 450 ms including a 50ms prestimulus baseline. Frequency analysis of MEG was performed for each single epoch over 225-ms intervals selected randomly from the period of 0±400 ms post stimulus onset. This was done to account for the fact that for the paced task, movement onset was time-locked to the stimulus, whereas this was not the case for the self-paced task. Choosing random time windows reduced the potential error due to task-dependent differences in the amount of pre- and postmovement periods included in the analysis. Additional analyses were also conducted both for the full 450-ms epoch and for the interval of 300±400 ms post-stimulus. Selecting a 225-ms window resulted in records of 70 points which were zero-padded to obtain 128 points. To reduce the frequency leakage for the different frequency bins, the records were multiplied with Welch windows (recommended in Press et al. [15]). Subsequent Fast Fourier Transform yielded 40 spectral amplitude values for each MEG sensor between 0±100 Hz (frequency resolution: 2.44 Hz).
These values were averaged across epochs to obtain measures of non-phase-locked activity. Differences in spectral amplitude between self-paced and externally cued ®nger movement tasks were assessed with paired, two-sided t-tests for each spectral amplitude and MEG sensor in the whole subject sample, resulting in 40 £ 143 5720 tests. T-values were converted to P-values. Pvalues from three adjacent frequencies had to meet the criterion of P # u0.001u to be considered signi®cant. This was an approximation to the problem of obtaining false positives due to multiple comparisons. The number of three consecutive P-values was chosen as a tradeoff between the aim (a) to avoid chance ®ndings and (b) to keep the frequency resolution as high as possible. In addition, a con®rmatory statistical analysis was conducted based on randomization tests suggested by Blair and Karniski [3] which were extended to multichannel data. The task-related spectral amplitude differences in the 18±30 Hz range were swapped between tasks for each recording channel and all 2 12 possible permutations across subjects were tested. The observed maximum t-value was then compared with the distribution of maximum t-values across all tests. To assess the topographical localization of signi®cant spectral amplitude values for the whole group, we projected the sensor locations onto a map with the major anatomical landmarks based on the magnetic resonance image from one representative subject (common coil system). To estimate the error which is introduced by this procedure, we used a single dipole localization of the ®rst somatosensory component generated by the stimulation to the thumb during the externally cued task. The spatial coordinates of this dipole were determined both individually and for the representative head model, enabling the estimation of (a) natural variation across subjects and (b) error introduced by using the common coil system. Individual localization of the dipole source showed no difference in variation between individual data (range: 2.1 cm in anterior-posterior direction and 1.6 cm in left-right direction) and common coil system (2.8 and 1.7 cm, respectively). The usage of the statistical probability method implies that the resolution is determined by the sensor spacing (2.5 cm). The results above justify the application of a common coil system. Fig. 1a shows the results of spectral amplitude analysis for all MEG channels. Applying the signi®cance criterion of P u0:001u for three consecutive P-values disclosed that self-paced movements were distinguished from externally cued movements by enhanced spectral amplitudes in a frequency range between 22±28 Hz (Fig. 1b). Randomization tests con®rmed this effect (P 0:03). In this frequency band, the mean amplitude difference between tasks across the two MEG sensors at the center of activation was 3.4 fT (SD 0:7 fT), t(11) 4.76, P , 0:001. Similar results were obtained for the whole recording interval and for the window of 300±400 ms post-stimulus. For the whole interval the difference between the two tasks was 6.8 fT at the
J. Kaiser et al. / Neuroscience Letters 283 (2000) 81±84
Fig. 1. Statistical mapping of spectral amplitude differences between self-paced and externally paced tasks. (a) Distribution of log-transformed P-values for each frequency between 0±80 Hz overlaid for each MEG sensor. The P-value curves were smoothed with a sliding average across three log-transformed P-values. (b) When the criterion of three consecutive P-values # u0.001u is applied, a signi®cant spectral amplitude increase in the range of 22±28 Hz is disclosed for the self-paced compared to the externally paced task. (c) Isocontour plot of the complete P-value distributions between 22±28 Hz projected onto a map of MEG sensors (seen from above, nose up). (d) Isocontour plot of the signi®cant effects between 22±28 Hz projected onto a map of MEG sensors (seen from above, nose up). (e) Major anatomical landmarks derived from the magnetic resonance image of one representative subject aligned with the MEG sensor map (e.g. sulcus centralis (s.ce), sulcus parieto-occipitalis (s.po) and sulcus calcarinus (s.ca)). (f) Area of signi®cant task-related spectral amplitude differences projected onto the anatomical map. In addition, the three dipoles explaining the stimulus-related magnetic ®elds (residual variance ,3%) in the externally paced task are depicted as numbered black squares. Dipoles 1 and 2 explain the earlier, sensory evoked components contralateral to the stimulation side, whereas dipole 3 re¯ects the later, motorrelated component contralateral to the response side. Note that these dipoles are located closely to the central sulcus, thus supporting the validity of the present mapping approach. Considering the relative distances between landmarks, the observed spectral amplitude effect is likely to be localized in Brodmann areas 6 and 8, probably including the SMA.
same signi®cance level. This could have been caused by the fact that the probability for a self-paced movement in the random interval decreased by a factor of two. There were
83
neither signi®cant differences in any other frequency band between 0±100 Hz nor were there frequencies where the externally paced task elicited larger spectral amplitudes than the self-paced task. The topography of the 22±28 Hz spectral amplitude increase is shown for the common coil system (Fig. 1d) and mapped onto the head model (Fig. 1f). In an additional analysis we used a three-dipole model for the externally cued task to calculate the location of the ®rst and second somatosensory evoked responses and the motor-related response which were also mapped onto the head model (Fig. 1f). This demonstrates that the area of signi®cant spectral amplitude increase was in frontal cortex anterior to primary motor areas. In summary, complex, self-paced but not simple, externally cued ®nger movements were accompanied by increases in MEG spectral amplitudes in the 22±28 Hz range. Although the present approach did not allow an exact assignment of the area of activation to an anatomical structure, mapping onto a head model and comparison with magnetic ®eld component sources whose topography is well known showed that the activation was located in premotor areas. Considering that self-paced movements are known to involve SMA [2,4,10,16,19,20], it is very likely that the observed effects were at least in part generated in SMA. However, it cannot be ruled out that more anterior (`preSMA' [1]) areas were also involved. The present results suggest that a statistical mapping approach allows the detection of activity that is otherwise not seen in MEG [8,11]. They are consistent with ®ndings of EEG alpha (9±11 Hz) power desynchronization over medial premotor areas and increased EEG coherence between primary and supplementary motor cortex in the 20±22 Hz range to self-paced compared with externally paced simple movements [9]. Gerloff et al. [9] suggested that internal pacing of movement poses higher demands on the motor system than external pacing, leading both to increased regional oscillatory activity of neuronal assemblies in the premotor system and to enhanced inter-regional coupling between premotor and sensorimotor areas. Unexpectedly subjective ratings indicated that the externally paced tapping task was perceived as more dif®cult than the self-paced task (paced: 5.8 (SD 0:5), selfpaced: 3.9 (SD 0:3), F 1; 11 13:0, P 0:004), and subjects felt they had performed the former task more accurately than the latter (self-paced: 7.1 (SD 0:4), paced: 5.7 (SD 0:3), F 1; 11 10:9, P 0:007). While we had no objective measure of task complexity or performance accuracy, these reports suggested that internal versus external triggering was more relevant in eliciting SMA activation than task complexity. The fact that no MEG spectral amplitude enhancements were found for the paced compared to the self-paced task indicates that subjective effort was not re¯ected by cortical oscillatory activity. In contrast to the premotor regions, oscillatory activity over primary sensorimotor areas did not differentiate
84
J. Kaiser et al. / Neuroscience Letters 283 (2000) 81±84
between the self-paced and the paced tasks. Pfurtscheller and colleagues [13] have described a desynchronization of EEG mu and fast beta (20±24 Hz) activity before, a synchronization of gamma band (40 Hz) activity during [14] and increased slow beta (14±18 Hz) oscillations following the completion of movements over the primary sensorimotor areas. Our tasks which led to comparable rates of ®nger movements, seem to have elicited equally strong oscillatory activity in primary sensorimotor areas to both types of movement. No comparison was made with a pre-movement baseline. Taken together with earlier ®ndings of enhanced MEG gamma band activity over visual areas to attended moving bars [18], the present results suggest that statistical mapping of oscillatory MEG activity provides an estimate of cortical activation that may allow the mapping of higher brain functions. Supported by (SFB550/C1).
Deutsche
Forschungsgemeinschaft
[1] Allison, T., McCarthy, G., Luby, M., Puce, A. and Spencer, D.D., Localization of functional regions of human mesial cortex by somatosensory evoked potential recording and by cortical stimulation. Electroenceph. clin. Neurophysiol., 100 (1996) 126±140. [2] Ball, T., Schreiber, A., Feige, B., Wagner, M., Lucking, C.H. and Kristeva-Feige, R., The role of higher-order motor areas in voluntary movement as revealed by high-resolution EEG and fMRI. Neuroimage, 10 (1999) 682±694. [3] Blair, B.C. and Karniski, W., An alternative method for signi®cance testing of waveform difference potentials. Psychophysiol., 30 (1993) 518±524. [4] Boecker, H., Kleinschmidt, A., Requardt, M., Hanicke, W., Merboldt, K.D. and Frahm, J., Functional cooperativity of human cortical motor areas during self-paced simple ®nger movements A high-resolution MRI study. Brain, 117 (1994) 1231±1239. [5] Colebatch, J.G., Deiber, M.P., Passingham, R.E., Friston, K.J. and Frackowiak, R.S., Regional cerebral blood ¯ow during voluntary arm and hand movements in human subjects. J. Neurophysiol., 65 (1991) 1392±1401. [6] Crone, N.E., Miglioretti, D.L., Gordon, B. and Lesser, R.P., Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis II. Event-related synchronization in the gamma band. Brain, 121 (1998) 2301±2315. [7] Cui, R.Q., Huter, D., Lang, W. and Deecke, L., Neuroimage of voluntary movement: topography of the Bereitschaftspotential, a 64-channel DC current source density study. Neuroimage, 9 (1999) 124±134.
[8] Deecke, L., Electrophysiological correlates of movement initiation. Rev. Neurol. Paris, 146 (1990) 612±619. [9] Gerloff, C., Richard, J., Hadley, J., Schulman, A.E., Honda, M. and Hallett, M., Functional coupling and regional activation of human cortical motor areas during simple, internally paced and externally paced ®nger movements. Brain, 121 (1998) 1513±1531. [10] Jenkins, I.H., Brooks, D.J., Nixon, P.D., Frackowiak, R.S. and Passingham, R.E., Motor sequence learning: a study with positron emission tomography. J. Neurosci., 14 (1994) 3775±3790. [11] Joliot, M., Crivello, F., Badier, J.M., Diallo, B., Tzourio, N. and Mazoyer, B., Anatomical congruence of metabolic and electromagnetic activation signals during a self-paced motor task: a combined PET-MEG study. Neuroimage, 7 (1998) 337±351. [12] Lang, W., Cheyne, D., Kristeva, R., Beisteiner, R., Lindinger, G. and Deecke, L., Three-dimensional localization of SMA activity preceding voluntary movement. A study of electric and magnetic ®elds in a patient with infarction of the right supplementary motor area. Exp. Brain Res., 87 (1991) 688± 695. [13] Pfurtscheller, G. and Lopes da Silva, F.H., Event-related EEG/MEG synchronization and desynchronization: basic principles. Clin. Neurophysiol., 110 (1999) 1842±1857. [14] Pfurtscheller, G., Neuper, C. and Kalcher, J.S.O., 40-Hz oscillations during motor behavior in man. Neurosci. Lett., 164 (1993) 179±182. [15] Press, W.H., Teukolsky, S.A., Vetterling, W.T. and Flannery, B.P., Numerical recipes, Cambridge University Press, Cambridge, 1992, pp. 5±47. [16] Rao, S.M., Binder, J.R., Bandettini, P.A., Hammeke, T.A., Yetkin, F.Z., Jesmanowicz, A., Lisk, L.M., Morris, G.L., Mueller, W.M., Estkowski, L.D., Wong, E.C., Haughton, V.M. and Hyde, J.S., Functional magnetic resonance imaging of complex human movements. Neurol., 43 (1993) 2311± 2318. [17] Roland, P.E., Larsen, B., Lassen, N.A. and Skinhoj, E., Supplementary motor area and other cortical areas in organization of voluntary movements in man. J. Neurophysiol., 43 (1980) 118±136. [18] Sokolov, A., Lutzenberger, W., Pavlova, M., Preissl, H., Braun, C. and Birbaumer, N., Gamma-band MEG activity to coherent motion depends on task-driven attention. NeuroReport, 10 (1999) 1997±2000. [19] Wessel, K., Zef®ro, T., Toro, C. and Hallett, M., Self-paced versus metronome-paced ®nger movements. A positron emission tomography study. J. Neuroimaging, 7 (1997) 145±151. [20] Wildgruber, D., Erb, M., Klose, U. and Grodd, W., Sequential activation of supplementary motor area and primary motor cortex during self-paced ®nger movement in human evaluated by functional MRI. Neurosci. Lett., 227 (1997) 161±164.
Statistical probability mapping reveals high-frequency magnetoencephalographic activity in supplementary motor area during self-paced ®nger movements Jochen Kaiser a,*, Werner Lutzenberger a, Hubert Preissl a, Dirk Mosshammer a, Niels Birbaumer a, b a
MEG-Center, Institute of Medical Psychology and Behavioral Neurobiology, University of TuÈbingen, Gartenstrasse 29, 72074 TuÈbingen, Germany b Department of General Psychology, University of Padova, Padova, Italy Received 9 December 1999; received in revised form 17 February 2000; accepted 17 February 2000
Abstract Investigations of both haemodynamic and electroencephalographic measures of brain activity have demonstrated supplementary motor area (SMA) involvement in self-paced ®nger movements. In contrast, analysis of magnetoencephalographic (MEG) signals in the time domain has usually failed to detect SMA activity in healthy individuals. We investigated oscillatory MEG activity in 12 normal adults during (a) a self-paced, complex sequence of ®nger movements and (b) a simple ®nger opposition task paced externally by tactile stimuli presented to the contralateral thumb. Statistical probability mapping revealed enhanced non-phase-locked spectral amplitudes in the 22±28 Hz range over bilateral frontal cortex during self-paced as compared to externally cued ®nger movements. This activity may re¯ect recruitment of cell assemblies in SMA during self-paced, complex movements. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Magnetoencephalography; Oscillatory activity; Statistical probability mapping; Finger movements; Self-paced; Supplementary motor area; Human subjects
During movements, contralateral primary sensorimotor areas exhibit increased regional cerebral blood ¯ow as assessed with positron emission tomography [17] and functional magnetic resonance imaging [16]. In addition, certain kinds of motor responses are accompanied by activations of more anterior regions such as the supplementary motor area (SMA) in the medial and dorsal parts of Brodmann area 6. There is some debate on whether SMA is involved only in the generation of complex movements [16,17] or whether it is also activated when simple movements are performed [1,2,5,20]. More importantly, most studies have observed SMA activation only when movements were self-paced or internally cued [4,10,16,19], although there have been contradictory ®ndings of SMA involvement also in externally paced motor responses [5]. Electroencephalographic (EEG) recordings have * Corresponding author. Tel.: 149-7071-297 4224; fax: 1497071-29 5956. E-mail address: [email protected] (J. Kaiser)
suggested that the earlier component of the readiness potential preceding self-initiated movements may be generated in bilateral SMA and the later part in contralateral primary motor cortex [7,8]; however, there was no such time lag in intracranial recordings [1]. In contrast, most magnetoencephalographic (MEG) studies analyzing signals in the time domain have failed to detect SMA activity [8,11]. This has been attributed to SMA anatomy: while radial sources in the dorsal parts of Brodmann area 6 may not generate detectable magnetic ®elds, opposed tangential sources in the bilateral medial parts of this area may cancel each other out [12]. The present study investigated oscillatory MEG responses to externally paced versus self-paced ®nger movements. Such responses have been recorded previously from primary sensorimotor areas in the alpha, beta and gamma ranges [6,9,13]. Here we used a statistical probability mapping approach, enabling the disclosure of signi®cant differences in high-frequency signals of low amplitude that would not be detectable in the analysis of evoked magnetic ®elds. Twelve right-handed subjects were tested in a magneti-
0304-3940/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S03 04 - 394 0( 0 0) 00 92 1- 6
82
J. Kaiser et al. / Neuroscience Letters 283 (2000) 81±84
cally shielded room (Vakuum-Schmelze, Hanau, Germany). MEG was recorded while tactile stimuli were delivered to the right thumb with a pneumatic stimulator (BTI Inc., San Diego, USA) containing a membrane driven by air pressure changes (stimulus duration: 100 ms). 89% of the stimuli were presented at randomized inter-stimulus intervals (ISI) varying between 400 and 500 ms, whereas the remaining 11% were presented at randomized ISI between 1450 and 1500 ms to avoid rapid habituation. 400 stimuli were presented in each of the following two conditions: (a) paced ®nger movements: subjects were instructed to respond as quickly as possible to the tactile stimuli by touching the left thumb with the left index ®nger, and (b) self-paced ®nger movements: while instructed to ignore the tactile stimuli, subjects continuously performed the following task: they had to oppose sequentially the left thumb to the left index ®nger once, to the middle ®nger twice, to the ring ®nger three times and to the little ®nger four times, and then repeat this sequence in reverse order. Both ®nger movement tasks were trained prior to MEG recording. The sequence of these two conditions was counterbalanced across subjects. At the end of the experiment, subjects were asked to rate (a) how dif®cult they found the tasks on scales ranging from 1 very easy to 10 very dif®cult, and (b) how accurately they thought they had performed the tasks on scales from 1 very poorly to 10 very accurately. MEG was collected with a whole head magnetometer (CTF Inc., Vancouver, Canada) comprising 143 hardware ®rst order magnetic gradiometers (5 cm baseline, average distance between sensors: 2.5 cm, sampling rate: 312.5 Hz, high-frequency cutoff: 100 Hz). The subject's head position was determined with localization coils ®xed at the nasion and the preauricular points both at the beginning and end of each experimental condition. Recordings with head movements exceeding 0.5 cm were discarded. The threshold for rejecting trials with eye movement artifacts was 1.3 pT in frontal channels. Epoch length was 450 ms including a 50ms prestimulus baseline. Frequency analysis of MEG was performed for each single epoch over 225-ms intervals selected randomly from the period of 0±400 ms post stimulus onset. This was done to account for the fact that for the paced task, movement onset was time-locked to the stimulus, whereas this was not the case for the self-paced task. Choosing random time windows reduced the potential error due to task-dependent differences in the amount of pre- and postmovement periods included in the analysis. Additional analyses were also conducted both for the full 450-ms epoch and for the interval of 300±400 ms post-stimulus. Selecting a 225-ms window resulted in records of 70 points which were zero-padded to obtain 128 points. To reduce the frequency leakage for the different frequency bins, the records were multiplied with Welch windows (recommended in Press et al. [15]). Subsequent Fast Fourier Transform yielded 40 spectral amplitude values for each MEG sensor between 0±100 Hz (frequency resolution: 2.44 Hz).
These values were averaged across epochs to obtain measures of non-phase-locked activity. Differences in spectral amplitude between self-paced and externally cued ®nger movement tasks were assessed with paired, two-sided t-tests for each spectral amplitude and MEG sensor in the whole subject sample, resulting in 40 £ 143 5720 tests. T-values were converted to P-values. Pvalues from three adjacent frequencies had to meet the criterion of P # u0.001u to be considered signi®cant. This was an approximation to the problem of obtaining false positives due to multiple comparisons. The number of three consecutive P-values was chosen as a tradeoff between the aim (a) to avoid chance ®ndings and (b) to keep the frequency resolution as high as possible. In addition, a con®rmatory statistical analysis was conducted based on randomization tests suggested by Blair and Karniski [3] which were extended to multichannel data. The task-related spectral amplitude differences in the 18±30 Hz range were swapped between tasks for each recording channel and all 2 12 possible permutations across subjects were tested. The observed maximum t-value was then compared with the distribution of maximum t-values across all tests. To assess the topographical localization of signi®cant spectral amplitude values for the whole group, we projected the sensor locations onto a map with the major anatomical landmarks based on the magnetic resonance image from one representative subject (common coil system). To estimate the error which is introduced by this procedure, we used a single dipole localization of the ®rst somatosensory component generated by the stimulation to the thumb during the externally cued task. The spatial coordinates of this dipole were determined both individually and for the representative head model, enabling the estimation of (a) natural variation across subjects and (b) error introduced by using the common coil system. Individual localization of the dipole source showed no difference in variation between individual data (range: 2.1 cm in anterior-posterior direction and 1.6 cm in left-right direction) and common coil system (2.8 and 1.7 cm, respectively). The usage of the statistical probability method implies that the resolution is determined by the sensor spacing (2.5 cm). The results above justify the application of a common coil system. Fig. 1a shows the results of spectral amplitude analysis for all MEG channels. Applying the signi®cance criterion of P u0:001u for three consecutive P-values disclosed that self-paced movements were distinguished from externally cued movements by enhanced spectral amplitudes in a frequency range between 22±28 Hz (Fig. 1b). Randomization tests con®rmed this effect (P 0:03). In this frequency band, the mean amplitude difference between tasks across the two MEG sensors at the center of activation was 3.4 fT (SD 0:7 fT), t(11) 4.76, P , 0:001. Similar results were obtained for the whole recording interval and for the window of 300±400 ms post-stimulus. For the whole interval the difference between the two tasks was 6.8 fT at the
J. Kaiser et al. / Neuroscience Letters 283 (2000) 81±84
Fig. 1. Statistical mapping of spectral amplitude differences between self-paced and externally paced tasks. (a) Distribution of log-transformed P-values for each frequency between 0±80 Hz overlaid for each MEG sensor. The P-value curves were smoothed with a sliding average across three log-transformed P-values. (b) When the criterion of three consecutive P-values # u0.001u is applied, a signi®cant spectral amplitude increase in the range of 22±28 Hz is disclosed for the self-paced compared to the externally paced task. (c) Isocontour plot of the complete P-value distributions between 22±28 Hz projected onto a map of MEG sensors (seen from above, nose up). (d) Isocontour plot of the signi®cant effects between 22±28 Hz projected onto a map of MEG sensors (seen from above, nose up). (e) Major anatomical landmarks derived from the magnetic resonance image of one representative subject aligned with the MEG sensor map (e.g. sulcus centralis (s.ce), sulcus parieto-occipitalis (s.po) and sulcus calcarinus (s.ca)). (f) Area of signi®cant task-related spectral amplitude differences projected onto the anatomical map. In addition, the three dipoles explaining the stimulus-related magnetic ®elds (residual variance ,3%) in the externally paced task are depicted as numbered black squares. Dipoles 1 and 2 explain the earlier, sensory evoked components contralateral to the stimulation side, whereas dipole 3 re¯ects the later, motorrelated component contralateral to the response side. Note that these dipoles are located closely to the central sulcus, thus supporting the validity of the present mapping approach. Considering the relative distances between landmarks, the observed spectral amplitude effect is likely to be localized in Brodmann areas 6 and 8, probably including the SMA.
same signi®cance level. This could have been caused by the fact that the probability for a self-paced movement in the random interval decreased by a factor of two. There were
83
neither signi®cant differences in any other frequency band between 0±100 Hz nor were there frequencies where the externally paced task elicited larger spectral amplitudes than the self-paced task. The topography of the 22±28 Hz spectral amplitude increase is shown for the common coil system (Fig. 1d) and mapped onto the head model (Fig. 1f). In an additional analysis we used a three-dipole model for the externally cued task to calculate the location of the ®rst and second somatosensory evoked responses and the motor-related response which were also mapped onto the head model (Fig. 1f). This demonstrates that the area of signi®cant spectral amplitude increase was in frontal cortex anterior to primary motor areas. In summary, complex, self-paced but not simple, externally cued ®nger movements were accompanied by increases in MEG spectral amplitudes in the 22±28 Hz range. Although the present approach did not allow an exact assignment of the area of activation to an anatomical structure, mapping onto a head model and comparison with magnetic ®eld component sources whose topography is well known showed that the activation was located in premotor areas. Considering that self-paced movements are known to involve SMA [2,4,10,16,19,20], it is very likely that the observed effects were at least in part generated in SMA. However, it cannot be ruled out that more anterior (`preSMA' [1]) areas were also involved. The present results suggest that a statistical mapping approach allows the detection of activity that is otherwise not seen in MEG [8,11]. They are consistent with ®ndings of EEG alpha (9±11 Hz) power desynchronization over medial premotor areas and increased EEG coherence between primary and supplementary motor cortex in the 20±22 Hz range to self-paced compared with externally paced simple movements [9]. Gerloff et al. [9] suggested that internal pacing of movement poses higher demands on the motor system than external pacing, leading both to increased regional oscillatory activity of neuronal assemblies in the premotor system and to enhanced inter-regional coupling between premotor and sensorimotor areas. Unexpectedly subjective ratings indicated that the externally paced tapping task was perceived as more dif®cult than the self-paced task (paced: 5.8 (SD 0:5), selfpaced: 3.9 (SD 0:3), F 1; 11 13:0, P 0:004), and subjects felt they had performed the former task more accurately than the latter (self-paced: 7.1 (SD 0:4), paced: 5.7 (SD 0:3), F 1; 11 10:9, P 0:007). While we had no objective measure of task complexity or performance accuracy, these reports suggested that internal versus external triggering was more relevant in eliciting SMA activation than task complexity. The fact that no MEG spectral amplitude enhancements were found for the paced compared to the self-paced task indicates that subjective effort was not re¯ected by cortical oscillatory activity. In contrast to the premotor regions, oscillatory activity over primary sensorimotor areas did not differentiate
84
J. Kaiser et al. / Neuroscience Letters 283 (2000) 81±84
between the self-paced and the paced tasks. Pfurtscheller and colleagues [13] have described a desynchronization of EEG mu and fast beta (20±24 Hz) activity before, a synchronization of gamma band (40 Hz) activity during [14] and increased slow beta (14±18 Hz) oscillations following the completion of movements over the primary sensorimotor areas. Our tasks which led to comparable rates of ®nger movements, seem to have elicited equally strong oscillatory activity in primary sensorimotor areas to both types of movement. No comparison was made with a pre-movement baseline. Taken together with earlier ®ndings of enhanced MEG gamma band activity over visual areas to attended moving bars [18], the present results suggest that statistical mapping of oscillatory MEG activity provides an estimate of cortical activation that may allow the mapping of higher brain functions. Supported by (SFB550/C1).
Deutsche
Forschungsgemeinschaft
[1] Allison, T., McCarthy, G., Luby, M., Puce, A. and Spencer, D.D., Localization of functional regions of human mesial cortex by somatosensory evoked potential recording and by cortical stimulation. Electroenceph. clin. Neurophysiol., 100 (1996) 126±140. [2] Ball, T., Schreiber, A., Feige, B., Wagner, M., Lucking, C.H. and Kristeva-Feige, R., The role of higher-order motor areas in voluntary movement as revealed by high-resolution EEG and fMRI. Neuroimage, 10 (1999) 682±694. [3] Blair, B.C. and Karniski, W., An alternative method for signi®cance testing of waveform difference potentials. Psychophysiol., 30 (1993) 518±524. [4] Boecker, H., Kleinschmidt, A., Requardt, M., Hanicke, W., Merboldt, K.D. and Frahm, J., Functional cooperativity of human cortical motor areas during self-paced simple ®nger movements A high-resolution MRI study. Brain, 117 (1994) 1231±1239. [5] Colebatch, J.G., Deiber, M.P., Passingham, R.E., Friston, K.J. and Frackowiak, R.S., Regional cerebral blood ¯ow during voluntary arm and hand movements in human subjects. J. Neurophysiol., 65 (1991) 1392±1401. [6] Crone, N.E., Miglioretti, D.L., Gordon, B. and Lesser, R.P., Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis II. Event-related synchronization in the gamma band. Brain, 121 (1998) 2301±2315. [7] Cui, R.Q., Huter, D., Lang, W. and Deecke, L., Neuroimage of voluntary movement: topography of the Bereitschaftspotential, a 64-channel DC current source density study. Neuroimage, 9 (1999) 124±134.
[8] Deecke, L., Electrophysiological correlates of movement initiation. Rev. Neurol. Paris, 146 (1990) 612±619. [9] Gerloff, C., Richard, J., Hadley, J., Schulman, A.E., Honda, M. and Hallett, M., Functional coupling and regional activation of human cortical motor areas during simple, internally paced and externally paced ®nger movements. Brain, 121 (1998) 1513±1531. [10] Jenkins, I.H., Brooks, D.J., Nixon, P.D., Frackowiak, R.S. and Passingham, R.E., Motor sequence learning: a study with positron emission tomography. J. Neurosci., 14 (1994) 3775±3790. [11] Joliot, M., Crivello, F., Badier, J.M., Diallo, B., Tzourio, N. and Mazoyer, B., Anatomical congruence of metabolic and electromagnetic activation signals during a self-paced motor task: a combined PET-MEG study. Neuroimage, 7 (1998) 337±351. [12] Lang, W., Cheyne, D., Kristeva, R., Beisteiner, R., Lindinger, G. and Deecke, L., Three-dimensional localization of SMA activity preceding voluntary movement. A study of electric and magnetic ®elds in a patient with infarction of the right supplementary motor area. Exp. Brain Res., 87 (1991) 688± 695. [13] Pfurtscheller, G. and Lopes da Silva, F.H., Event-related EEG/MEG synchronization and desynchronization: basic principles. Clin. Neurophysiol., 110 (1999) 1842±1857. [14] Pfurtscheller, G., Neuper, C. and Kalcher, J.S.O., 40-Hz oscillations during motor behavior in man. Neurosci. Lett., 164 (1993) 179±182. [15] Press, W.H., Teukolsky, S.A., Vetterling, W.T. and Flannery, B.P., Numerical recipes, Cambridge University Press, Cambridge, 1992, pp. 5±47. [16] Rao, S.M., Binder, J.R., Bandettini, P.A., Hammeke, T.A., Yetkin, F.Z., Jesmanowicz, A., Lisk, L.M., Morris, G.L., Mueller, W.M., Estkowski, L.D., Wong, E.C., Haughton, V.M. and Hyde, J.S., Functional magnetic resonance imaging of complex human movements. Neurol., 43 (1993) 2311± 2318. [17] Roland, P.E., Larsen, B., Lassen, N.A. and Skinhoj, E., Supplementary motor area and other cortical areas in organization of voluntary movements in man. J. Neurophysiol., 43 (1980) 118±136. [18] Sokolov, A., Lutzenberger, W., Pavlova, M., Preissl, H., Braun, C. and Birbaumer, N., Gamma-band MEG activity to coherent motion depends on task-driven attention. NeuroReport, 10 (1999) 1997±2000. [19] Wessel, K., Zef®ro, T., Toro, C. and Hallett, M., Self-paced versus metronome-paced ®nger movements. A positron emission tomography study. J. Neuroimaging, 7 (1997) 145±151. [20] Wildgruber, D., Erb, M., Klose, U. and Grodd, W., Sequential activation of supplementary motor area and primary motor cortex during self-paced ®nger movement in human evaluated by functional MRI. Neurosci. Lett., 227 (1997) 161±164.