2001 Experimental Brain Research

Exp Brain Res (2001) 140:290–300 DOI 10.1007/s002210100815

R E S E A R C H A RT I C L E

Te H. Dai · Jing Z. Liu · Vinod Sahgal Robert W. Brown · Guang H. Yue

Relationship between muscle output and functional MRI-measured brain activation Received: 17 October 2000 / Accepted: 25 May 2001 / Published online: 31 July 2001 © Springer-Verlag 2001

Abstract The relationship between functional MRI (fMRI)-measured brain signal and muscle force and or electromyogram (EMG) is critical in interpreting fMRI data and understanding the control mechanisms of voluntary motor actions. We designed a system that could record joint force and surface EMG online with fMRI data. High-quality force and EMG data were obtained while maintaining the quality of the fMRI brain images. Using this system, we determined the relationship between fMRI-measured brain activation and handgrip force and between fMRI-measured brain signal and EMG of extrinsic finger muscles. Ten volunteers participated in the experiments (only seven subjects’ data were analyzed due to excessive noise in the fMRI data of three subjects). The participants exerted 20%, 35%, 50%, 65%, and 80% of the maximal force. During each contraction period, handgrip force, surface EMG of the finger flexor and extensor muscles, and fMRI brain images were acquired. The degree of muscle activation (force and EMG) was directly proportional to the amplitude of the brain signal determined by fMRI in the entire brain and in a number of motor function-related cortical fields, including primary motor, sensory regions, supplementary motor area, premotor, prefrontal, parietal and cingulate cortices, and cerebellum. All the examined brain areas demonstrated a similar relationship between the fMRI signal and force. A stronger fMRI signal during higher force indicates that more cortical output neurons and/or interneurons may participate in generating T.H. Dai · J.Z. Liu · G.H. Yue (✉) Department of Biomedical Engineering/ND20, The Lerner Research Institute, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA e-mail: [email protected] Tel.: +1-216-4459336, Fax: +1-216-4449198 V. Sahgal · G.H. Yue Department of Physical Medicine and Rehabilitation, The Cleveland Clinic Foundation, Cleveland, OH 44195, USA T.H. Dai · J.Z. Liu · R.W. Brown Department of Physics, Case Western Reserve University, Cleveland, OH 44106, USA

descending commands and/or processing additional sensory information. The similarity in the relationship between muscle output and fMRI signal in the cortical regions suggests that correlated or networked activation among a number of cortical fields may be necessary for controlling precise static force of finger muscles. Keywords Finger flexor muscles · Functional magnetic resonance imaging · Handgrip force · Surface EMG · Voluntary contraction

Introduction Because of its noninvasive nature and good spatial resolution, functional magnetic resonance imaging (fMRI) has been increasingly used in studying human brain function since its emergence in 1992 (Bandettini et al. 1992; Kwong et al. 1992; Ogawa et al. 1992). In the field of human motor control, recent fMRI data of brain activation have confirmed and extended the findings of previous electrophysiological studies in animals (Evarts 1968; Hepp-Reymond et al. 1978) and in patients who undergo neurological surgery (Penfield and Rasmussen 1950). For example, a single fMRI experiment involving a simple voluntary motor action can result in activation of the primary sensorimotor cortex and of many other cortical regions (Bandettini et al. 1992; Boecker et al. 1994; Ehrsson et al. 2000; Kim et al. 1993a, 1993b; Liu et al. 1999; Rao et al. 1993, 1996; Sanes et al. 1995; Tyszka et al. 1994; Yue et al. 2000). These fMRI data not only suggest an important role of the primary sensorimotor cortex in controlling motor actions, but also indicate that the participation of multiple cortical areas may be essential for planning and executing a voluntary motor action (König and Engel 1995). Despite the increased use of fMRI in human motor control investigation, only a small number of studies have addressed the relation between muscle output (force) and fMRI-measured brain activity, and the results are contradictory (Dettmers et al. 1996; Ludman et al.

291

1996; Thickbroom et al. 1998). Furthermore, electromyograms (EMG) have not been recorded concurrently with force and fMRI in any of these studies. Without EMG data, the muscle activation level cannot be assessed with high confidence based on force information alone, as joint force is determined by both the agonist and the antagonist muscles. Thus, a low level of force can mean a high level of muscle activation if both the agonist and antagonist muscles are coactivated. The primary difficulty with recording force and EMG signals during an fMRI experiment is that the presence of electrical equipment and signals can disturb the homogeneity of the magnetic field of the imaging scanner and introduce noise to the brain images. Similarly, high-voltage signals generated by running the imaging acquisition sequence can impair signals recorded from muscle. Recently, we designed and built a system that can record joint force and muscle surface EMG online with fMRI data without compromising data quality of any type (force, EMG, or fMRI; Liu et al. 2000). Based on single-cell recordings in the primary motor cortex (MI) in monkeys, a direct relationship between the discharge rate and exerted force has been documented (Evarts 1968; Hepp-Reymond et al. 1978, 1999). A positron emission tomography (PET) study (Dettmers et al. 1995) has reported that there is a direct relationship between the index finger flexion force and increases in cerebral blood flow in four motor-related brain regions: contralateral sensorimotor cortex, supplementary motor area (SMA), cingulate cortex, and cerebellum. One study (Siemionow et al. 2000) has demonstrated that surface electroencephalogram (EEG)-derived motor activity-related cortical potential (MRCP) measured by electrodes located on the scalp overlying the contralateral sensorimotor cortex and SMA is linearly related to isometric joint force and muscle EMG. These findings suggest that cortical activity measured by single-cell discharge, cerebral blood flow, and EEG-derived MRCP is directly related to muscle output. Based on these results and the conclusion that fMRI signal reflects primarily the synaptic activities of cortical neurons (Jueptner and Weiller 1995), we hypothesized that the magnitude of fMRI-measured brain activity would be proportional to muscle force and EMG. Thus, the purpose of this study was to determine the relationship between fMRI-measured human cortical activation and handgrip force and surface EMG of finger flexor muscles.

Methods

Measurement systems The measurement systems included a 1.5-T Siemens Vision scanner and a force/EMG recording system specially built for use in an MRI environment (Liu et al. 2000). In addition, a visualfeedback system was used to provide force information to the subject in the scanner so that he/she could exert a force that matched the target. The major challenge for the measurement systems was that the operation of the force/EMG data acquisition equipment and/or the force/EMG signals should not affect the quality of the MRI data; conversely, running of the MRI scanner should not generate significant noise that would impair the signals of force and EMG. To minimize the risks of interference among these systems, the equipment inside the MRI room is either metalfree or well shielded. We used nonmetallic material (polycarbonate) to build the handgrip device and the connecting tube (nylon). The EMG electrodes were silver-silver chloride (8-mm recording diameter) with double-shielded wires. Devices that were considered to have potential risks of introducing noise were positioned outside of the MRI room. These included the pressure transducer (part of the transducer is made of stainless steel), EMG and force amplifiers, the associated power supply, and the data acquisition unit (laptop computer and its docking station). The electrode wires were formed into a flat cable running from the subject in the scanner to the amplifier outside the MRI room, through the thin gap between the lower edge of the shielding door and the floor along with the nylon tube connecting the handgrip device and the pressure transducer, with the door still tightly shut. Thus, the shielding of the MRI room was not compromised by these connections. An additional layer of shielding was built in the flat cable that covered the electrode wires. Force data acquisition Handgrip force was measured by a pressure transducer (EPX-N1 250 PSIG; Entran Devices, Fairfield, N.J.). The metal-free handgrip device was held in the subject’s right hand in the MRI room and was connected to the pressure transducer located outside the MRI room through a nylon tube filled with distilled water. When force was applied to the handgrip device, it pushed the piston built in the device, and the piston pressed the water against the transducer. The output of the transducer was connected to an amplifier whose output was directed to a data acquisition system (Spike 2; Cambridge Electronic Design, Cambridge, UK) and recorded online on the hard disk of a laptop computer. The transducer output is highly linear (r=0.9998; Liu et al. 2000). The force signal was digitized at 200 samples/s. EMG data acquisition Surface EMG was measured from the flexor digitorum profundus (FDP), flexor digitorum superficialis (FDS), and extensor digitorum (ED) muscles. Bipolar electrodes (8-mm recording diameter; In Vivo Metric, Healdsburg, Calif.) were attached to the skin overlying the belly of each muscle. The electrodes were connected with the custom-built differential amplifiers (Liu et al. 2000) located outside the MRI room. The output of the EMG amplifiers (frequency bandwidth, 10–2,000 Hz) was connected to the Spike 2 data acquisition system and saved on the hard disk of the laptop computer. The EMG signal was digitized at a rate of 2,000 samples/s.

Subjects fMRI data acquisition Ten right-handed volunteers (eight men and two women, aged 31.3±6.5 years) participated in the study. All subjects were healthy and had no known neuromuscular disorders at the time of the study. The experimental procedures were approved by the Institutional Review Board at the Cleveland Clinic Foundation. All subjects gave informed consent prior to participation in the study.

fMRI images were collected on a Siemens 1.5-T Vision system using a circular polarized head coil and an interleaved multislice gradient echo EPI pulse sequence (TR/TE=115/22 ms). The subject was positioned in the MRI chamber (supine) and was told to remain as still as possible. The subject’s head was stabilized by padded restraints and by taping the forehead to the frame of the

292 head coil. Both T1-weighted (anatomic) images and functional images were collected in the transverse plane. At each force level, fMRI images were collected during a rest (baseline) period before the handgrip contraction and during the contraction period. Brain activation was detected by comparing the signal intensity of the active images with that of rest images based on the change of local blood oxygenation level (DeYoe et al. 1994; Kwong et al. 1992; Ogawa et al. 1990, 1992, 1993). In each rest or active period, 20 brain slices (from the top of the head) were imaged (6-mm slice thickness). We refer to these 20 slices that covered the entire brain as one scan. Eleven scans (220 images) were collected in each period. The field of view for an image was 256 mm × 256 mm, and the matrix size was 128×128, thus, yielding a spatial resolution of 2 mm × 2 mm. Visual feedback system This study required subjects to repeatedly exert a predetermined handgrip force; thus, it was essential for the subject in the MRI scanner to know exactly how much force they were producing. This task was accomplished by a visual feedback system. This system included a Silent Vision system (SV-2200; Avotec, Jensen Beach, Fla.), a video camera, and an oscilloscope. During the experiments, a target (e.g., 20% maximal handgrip force) was placed on the oscilloscope. The video camera was pointed to the oscilloscope screen to transmit the image to the video interface/ monitor unit located outside the MRI room. The video interface/ monitor unit was connected to a color LCD projector by a long fiber-optic cable. The LCD projector was located near the scanner inside the MRI room. The output of the color LCD projector was directed to a pair of adjustable biocular glasses via a fiber-optic guide. The biocular glasses were fixed to the top-opening of the head coil, directly above the subject’s eyes. Through the adjustable glasses, the subject could clearly see the target and match the target with the exerted force. The fiber-optic cable and other cables of the visual feedback system were permanently installed into the MRI room. The cables ran to the ceiling together with the MRI system cables and went through the penetration panel to the operation station outside the MRI room. Experiment procedures One pair of electrodes was attached to each of the three muscles (FDP, FDS, and ED). The FDP and FDS are the agonist of handgrip, and the ED is the antagonist. The muscles were identified by palpating the skin when the subject moved the fingers. We located the FDP from the posterior surface of the forearm near the elbow joint (near the medial side of the ulna. The FDS was located from the anterior surface of the forearm between the flexor carpi radialis and flexor carpi ulnaris. The skin overlying the identified muscles was cleaned by alcohol pads prior to electrode placement. A common reference electrode for the three recording pairs was fixed on the skin overlying the lateral epicondyle near the elbow joint. Correct electrode placement was confirmed by asking the subject to make appropriate finger movements while watching the EMG activity. Movements of the distal phalangial joint are primarily contributed by the FDP, whereas those of the intermediate phanlangial joint are accomplished mainly by the FDS. Although the two muscles were accessed from the two different locations on the forearm, the signals recorded from each muscle might have been contributed by both flexors and other forearm muscles. Finger extension is a result of activation of the ED muscle. Before the subject was positioned in the MRI chamber, the force and EMG signals were tested by asking them to squeeze the handgrip device. All data channels were displayed on the monitor of the laptop computer. The placement of the EMG electrodes was adjusted at this time if it was not at the correct location. While standing upright with the right arm in a vertical position on the right side of the body, the subject performed a maximal voluntary contraction (MVC) of the handgrip. Based on the MVC force,

Fig. 1 Examples of force and EMG recordings during fMRI data acquisition. The upper panel shows the force exerted by a subject at 80% MVC. The middle panel displays the EMG signals of the entire trial recorded from the flexor digitorum profundus (FDP) muscle; note that high-voltage noises associated with the pulse sequence were recorded during most of the trial except the first several seconds of the recording. The lower panel exhibits the EMG signal within the 200-ms gap between the two adjacent scans of image acquisition. The EMG signal in the 200-ms period was not significantly affected by the fMRI pulse sequence

20%, 35%, 50%, 65%, and 80% handgrip force was calculated. Each level was displayed on the oscilloscope as the target, and the subject was asked to match the target with the exerted force. One trial was performed at each level, and the time during which the force was on the target was about 5 s for each contraction. At least a 30-s rest was provided between contractions. Five subjects performed the five levels in ascending order and the other five subjects in descending order. The purpose of recording the EMG data at the five levels of force outside the MRI room was to compare the EMG signal recorded without the influence of operating the MRI system to that obtained inside the MRI bore when the scanner was running. If the EMG signal recorded from inside the scanner was contaminated by the magnetic field and image acquisition pulse sequences, then the signal should be significantly different from that recorded outside. The subject was then positioned (supine) within the MRI chamber. The biocular glasses of the feedback system were adjusted to a position at which the subject could see the oscilloscope screen clearly with both eyes and the cursor lines were well focused. The right arm rested comfortably on a pile of hospital sheets placed on the sliding board of the scanner. The handgrip device was held in the right hand, but the subject was told not to apply any force until they were told to do so. One trial of handgrip MVC was tested after the subject was placed at the center of the scanner. Based on this MVC force, the five target levels (20%, 35%, 50%, 65%, and 80%) in the scanner were calculated. Any communication between the experimenter and the subject was through intercom. Functional images were acquired in the transverse plane across the brain while the subject performed a handgrip contraction (active images) and while they rested (rest images) with the eyes looking at the oscilloscope screen on which two cursors were shown. The lower cursor was the force signal (zero or baseline force at rest) and the upper one indicated the target. Twenty contiguous brain slices (one scan) were imaged. T1-weighted anatomical images of the same number of slices at the same positions were collected before acquiring the functional images. Eleven scans were acquired during each fMRI (active or rest) period. The duration of each period was 22 s. During the active

293 Fig. 2 Examples of handgrip force and EMG data at five intensity levels (percentage MVC) from a representative subject. The EMG signals shown here were those recorded before the image acquisition began (FDP flexor digitorum profundus, FDS flexor digitorum superficialis)

period, the subject was given a “go” command, and they squeezed the handgrip device to match the force cursor with the target. About 5 s after the two cursors were matched, the fMRI sequence began. Thus, the actual duration of each contraction was ~29 s. Subjects were told not to move the arm and wrist during the entire data collection period (~20 min), since alteration of muscle length would affect force and EMG output. Five subjects performed the five levels of contractions in descending order and the other five subjects, ascending order. After collecting 11 scans of images at each force level, the system needed about 3 min to save the data before image collections at the next force level. This 3-min period served as a rest for the subject after completing each contraction. Not all EMG data were noise free. As indicated in Fig. 1, “clean” EMG signals are shown only during the time period before the fMRI data collection and during the sequence gaps between adjacent fMRI scans. Each gap was 200 ms. Because of substantial noise created by the pulse sequence, the EMG signal was not readable during each scan period (Fig. 1). However, the available “clean” EMG before or between the fMRI scans was adequate to reflect the level of muscle activation for the whole contraction (Figs. 2, 3; see also Liu et al. 2000). Force/EMG data analysis

Fig. 3A–C Relationship between EMG and force shown by group data. A The EMG signals used for the analysis were recorded from outside the MRI room. B The EMG recordings were obtained prior to the fMRI data acquisition. C The EMG data used were those recorded between the fMRI scans. There was no significant difference in the relationship between the two variables regardless of which set of EMG data was used (ED extensor digitorum)

The force signal was converted from voltage to newtons according to the calibration equation developed for the pressure transducer (Liu et al. 2000). At each level, the force was averaged across the entire contraction, from the time when it reached the target and was kept steady to the time when the force began to decline from the target after the subject was told to relax. For the EMG data recorded inside the scanner, the signal was analyzed during the first 5 s of each contraction, during which no fMRI pulse sequence was executed. In addition, the EMG was also analyzed during each 200-ms period between each two consecutive scans. The EMG signal was first rectified and then averaged over the beginning 5-s period and each of the 200-ms periods using the Spike 2 data-analysis software. A grand average was then calculated over the ten 200-ms periods. These values of average EMG were then normalized to the average MVC EMG performed inside the scanner. For the EMG data obtained outside the MRI room, the signal was rectified and averaged over the period (~5 s) during which the force was on the target and stable. The averaged EMG was normalized to the averaged MVC EMG acquired outside the MRI room.

294 fMRI data analysis fMRI data analysis was performed by using the MEDx 3.2 software package (Sensor Systems, Sterling, Va.) specially designed for functional imaging analysis. During postexperiment image processing, image motion detection and correction were performed before the statistical comparisons. The first set of 20 rest images preceding the images during each force-level contraction was used as the reference images. Detected displacement on the x (ear-toear), y (backhead-to-forehead), and z (head-to-feet) axes was 1.41±1.28 mm, 2.47±0.77 mm, and 0.35±0.20 mm, respectively. Images obtained during force performance were realigned with the reference images (motion correction). After motion correction, the displacement between the reference and experimental images was reduced, on average, to less than 0.5 mm on all axes, which was substantially smaller than the pixel size (2 mm). When motion correction was performed, we considered not only the x, y, z translation, but also the rotation. Detected displacement on x-, y-, and z-axes was the total result of both the translation motion correction and the rotational motion correction. Images acquired during the force production period were compared with the same (location) images acquired during the rest period on a pixel-to-pixel basis with Student t-tests. (The first of the 11 image scans in each period was excluded from the data analysis to allow a same T1 weighting for all the images.) Pixels with a z-score of 2.5 or more (corresponding P-value ≤0.006) were included in the functional map. The statistical functional map (z map) was overlaid onto a T1-weighted anatomical image to determine locations of activation. For measurements of activation in the individual cortical fields, activated pixels were circled by hand at each location and the number calculated. Up to this stage, we found that fMRI data of three of the ten subjects could not be used due to substantial noise, probably caused by head motion during the experiment. In these three subjects, no activation pattern could be recognized. In one subject, an extremely large number of activated pixels were almost evenly spread in every image. In the other two, activated pixels were concentrated around the edge of the brain to form a “circular color band,” a typical noise pattern caused by head movement. Thus, the data reported in the Results section are based on the results of seven subjects who had “good” fMRI data. The MI is defined as the region between the precentral sulcus and central sulcus. The primary sensory cortex (SI) was identified from the central sulcus to the postcentral sulcus. The SMA was classified as the medial wall area above the cingulate sulcus, posterior to the anterior paracentral sulcus, and anterior to the precentral sulcus. The premotor cortex (PM) was an area lateral to the SMA and anterior to the precentral sulcus or primary motor cortex. The cingulate gyrus (CG) was identified between the corpus callosum and cingulate sulcus. Prefrontal cortex (PFC) activity was measured from the superior, middle, inferior, and orbital frontal gyri anterior to the premotor cortex, frontal eye area, and Broca’s area. Activation in the parietal lobe (PL) was measured in the area between the postcentral sulcus and occipital lobe, not including any areas in the temporal lobe. We measured cerebellum (CBL) activation globally without further quantifying the activity in single hemispheres or individual cerebellar areas. The identification of the brain areas was assisted by consulting a standard brain atlas, high-resolution MRI brain atlas, and experienced neurologists (see Acknowledgements). Activation in the entire brain and in each of the identified cortical fields was quantified by the number of pixels that passed the z-score threshold and by the average z-score (average intensity). The average intensity was defined as the total z-score divided by the total number of activated pixels in the entire brain or in each of the cortical fields defined above. Our earlier analyses have shown that changes in the z-score-indicated average intensity did not differ substantially from the changes in the average intensity calculated from the original image signals (unpublished results). Statistical analysis The signal difference between baseline and performance conditions was determined by pixel-to-pixel comparisons using Student

t-tests with Bonferroni adjustment for the significance level. A significant increase in signal was determined by a z-threshold (unadjusted z-score, 2.5 or more) in each comparison (corresponding P-value for z≥2.5 is 0.006). Linear regression analyses were performed between EMG and force, fMRI and force, and fMRI and EMG to determine the relationship between the two variables. All data in the text are presented as mean ± standard deviation (SD).

Results Force and EMG In Fig. 2, data of handgrip force (left column) at five different levels (20%, 35%, 50%, 65%, and 80%) from a subject are displayed with the corresponding surface EMG signals from the FDP (middle column) and FDS (right column). These data show that higher force is associated with a greater EMG signal and that the quality of force and EMG signals was not affected by the presence of the strong magnetic field. The group data of the force and EMG are shown in Fig. 3. EMG data recorded outside the MRI room are shown in Fig. 3A. EMG signals analyzed from the first 5 s of each contraction inside the scanner are presented in Fig. 3B, and those averaged from the 200-ms periods between the fMRI scans are displayed in Fig. 3C. The EMG values between the three data sets (outside, inside before the scans, and inside between the scans) were similar, indicating that: (1) the signals were not significantly affected by the magnetic field; and (2) the signals recorded during a brief interruption (200 ms) between two fMRI scans at each force level were adequate to represent EMG activity of the contraction. For the EMG recorded outside the MRI room, the values of the normalized EMG signal corresponding to the five force levels were 16%, 33%, 44%, 64%, and 78% MVC for the FDS muscle. The same five values were 19%, 30%, 50%, 58%, and 79% for the FDP muscle (Fig. 3A). For the first 5-s EMG recorded inside the scanner, the five values were 14%, 33%, 41%, 62%, and 73% MVC for the FDS muscle. For the FDP muscle, the five EMG values were 14%, 30%, 51%, 65%, and 78%, respectively (Fig. 3B). For the normalized average EMG recorded during the 200-ms periods between the fMRI scans, the values at the five force levels for the FDP muscle were 12%, 24%, 39%, 58%, and 74%; and those for the FDS were 13%, 28%, 41%, 66%, and 81% (Fig. 3C). Linear regression analysis resulted in correlation values of 0.98, 0.99, and 0.94 between force and EMG for the FDS, FDP, and ED muscles, respectively, for the EMG data collected outside (Fig. 3A). The three correlation values were 0.98, 0.99, and 0.99, respectively, for the EMG data of the first 5 s in the scanner (Fig. 3B). The three correlation values for the EMG data of the 200-ms periods between the scans were 0.99, 0.99, 0.98, respectively (Fig. 3C). Note that values from the ED muscle, the antagonist, also correlated strongly with handgrip force.

295

Fig. 4A, B Functional images (at the transverse plane) taken at the same brain locations at the five force levels. A The cortical level, clearly showing activation of the primary motor, sensory, and premotor cortices. Primary motor and sensory cortices are separated by the central sulcus, with the motor cortex located above the central sulcus. The premotor cortex is located above the precentral sulcus. B Activities of the three association cortices: prefrontal (PFC), cingulate (CG), and parietal (PL). It is apparent that at these particular cortical locations both the number of activated pixels and average intensity increased with handgrip force. A and B were taken from two different subjects. The z-score threshold for Fig. 6 was set at 3.0 for greater image clarity. Because the image is viewed from feet to head orientation, the right side of the image represents the left side of the brain

Correlation between fMRI and force Cortical activation was quantified by calculating the number of pixels passed the statistical threshold (z≥2.5) and average activation intensity (summed z-score per number of activated pixels) in the entire brain and individual cortical regions. Examples are given in Fig. 4, showing brain activation pattern during five levels of handgrip contraction; each image was collected from one of the five force levels. The five images in Fig. 4A show activities of the MI, SI, and PM in the hemisphere contralateral to the performing hand (right-side image means left-side brain). The five images in Fig. 4B illustrate activation patterns of the PFC, CG, and PL. Clearly, as the force increased from low to high, both the activated pixel number and average intensity increased. Figure 5A shows the relationship between force and number of activated pixels (filled diamonds) and between force and average intensity (filled circles), calculated based on the data of the entire brain. The correlation (r) value for the force and pixel number was 0.97, and that for the force and average intensity was 0.95. The correlation data between force and activated pixel number and average intensity in each individual cortical region is presented in Fig. 6. PFC (r=0.96), SMA (r=0.93), CG (r=0.95), and CBL (r=0.94) were measured

Fig. 5A, B Relationship between number of activated pixels and force (filled diamonds) and between average intensity and force (filled circles). The two fMRI measurements (number of pixels and average intensity) were made from the data of the entire brain

bilaterally; and MI (r=0.92), SI (r=0.89), PM (r=0.99), and PL (r=0.90) were measured contralaterally in the left hemisphere. The correlation values between force and average intensity in the PFC, SMA, CG, and CBL were 0.98, 0.97, 0.95, and 0.98, respectively. Those for the MI, SI, PM, and PL were 0.96, 0.98, 0.77, and 0.92, respectively. The actual values of the pixel number and average intensity at each force level are listed in Table 1.

296

activated pixels (filled diamonds) and between EMG and average intensity (filled circles), calculated based on the data of the entire brain. The r-value for the EMG and pixel number was 0.97, and that for the EMG and average intensity was 0.96. The correlation data between EMG and activated pixel number and average intensity in each individual cortical region were also determined. PFC (r=0.98), SMA (r=0.97), CG (r=0.96), and CBL (r=0.98) were measured bilaterally; and MI (r=0.94), SI (r=0.99), PM (r=0.78), and PL (r=0.93) were measured contralaterally in the left hemisphere. The correlation values between EMG and average intensity in the PFC, SMA, CG, and CBL were 0.99, 0.99, 0.97, and 0.97, respectively. Those for the MI, SI, PM, and PL were 0.95, 0.93, 0.98, and 0.93, respectively. Activation of the ipsilateral cortical fields

Fig. 6A–D Relationship between number of activated pixels and force (filled diamonds) and between average intensity and force (filled circles). The pixel number and average intensity were measured from individual cortical areas: left side of primary motor cortex (MI_L), left side of primary sensory cortex (SI_L), left side of premotor cortex (PM_L), left side of parietal lobe (PL_L), cingulate gyrus (CG), supplementary motor area (SMA), prefrontal cortex (PFC), and cerebellum (CBL). The measurements were made bilaterally in the CG, SMA, PFC, and CBL

Correlation between fMRI and EMG

The number of pixels passing the statistical threshold and average intensity were also quantified in the MI, SI, PM, and PL of the ipsilateral (right) hemisphere. These values are presented in Table 1. The ipsilateral activation in these cortical areas seems quite strong as compared to the fMRI data of the same regions of the contralateral side. The high ipsilateral activity may be a result of precision control of force during the contractions (subjects were required to precisely match the target force). It has been reported that power grip contractions requiring little precision control were associated predominantly with contralateral primary sensorimotor area activation, whereas the precision grip involved extensive activation in a number of motor areas in both hemispheres (Ehrsson et al. 2000). It is particularly interesting that the number of activated pixels in the ipsilateral PL (PLR) was consistently greater than that in the contralateral PL (PLL; Table 1).

Because the three sets of the EMG data were similar (Fig. 3), we used only the EMG data in Fig. 3B to perform correlation analyses with the fMRI data. Figure 5B shows the relationship between EMG and number of

Discussion

Table 1 Number of “activated” pixels and average intensity (z-score) of these pixels measured in the entire brain (Global) and individual cortical regions. (PFC prefrontal cortex, CG cingulate

gyrus, SMA supplementary motor area, CBL cerebellum, L and R left and right hemisphere, PM premotor cortex, MI primary motor cortex, SI primary sensory cortex, PL parietal lobe)

Force(%) Global Pixels (n) 20 35 50 65 80 Average 20 intensity 35 50 65 80

PFC

CG

SMA

CBL

The purpose of this study was to determine the relationship between muscle output and fMRI-measured brain activity

PML

PMR

MIL

MIR

1,655.00 334.00 40.00 25.00 90.00 39.00 27.00 29.00 29.00 3,683.00 478.00 116.00 86.00 216.00 77.00 67.00 87.00 78.00 6,626.00 808.00 240.00 176.00 427.00 116.00 128.00 148.00 165.00 7,503.00 1,041.00 257.00 232.00 461.00 153.00 196.00 197.00 175.00 8,429.00 1,220.00 321.00 237.00 541.00 198.00 184.00 192.00 159.00 3.01 3.27 3.38 3.45 3.60

3.03 3.18 3.29 3.41 3.62

2.82 2.97 3.10 3.12 3.31

2.88 2.99 3.12 3.16 3.24

2.99 3.23 3.33 3.48 3.63

2.91 3.17 3.24 3.17 3.41

2.90 3.29 3.38 3.44 3.50

3.01 3.25 3.36 3.43 3.53

2.84 3.19 3.42 3.33 3.44

SIL

SIR

PLL

PLR

35.00 29.00 50.00 99.00 104.00 84.00 170.00 222.00 165.00 154.00 258.00 354.00 206.00 161.00 278.00 335.00 199.00 161.00 315.00 320.00 2.92 3.13 3.26 3.37 3.54

2.91 3.20 3.35 3.35 3.50

2.92 3.17 3.27 3.30 3.51

2.90 3.16 3.34 3.37 3.51

297

in healthy human subjects. The major findings were that: (1) high-quality fMRI data, handgrip force, and surface EMG of the finger flexor and extensor muscles could be acquired at the same time; (2) there was a direct relationship between handgrip force and fMRI signal and between muscle EMG and fMRI signal in a number of motor function-related cortical fields; and (3) these cortical fields acted in a correlated way in controlling magnitude of the static force. Recording force and EMG information in an MRI environment A major feature of this study was that we were able to concurrently record fMRI, joint force, and muscle EMG data without compromising data quality. Precise recordings of these data are critical in assessing the relation between central nervous system activity and muscle output. Comparisons between fMRI images acquired with and without operation of the force/EMG recording system indicated that the quality of the fMRI data was the same between the two conditions (Liu et al. 2000). Similarly, there was no difference in the force and EMG signals recorded inside the operating MRI scanner and those signals recorded outside the MRI room (Fig. 3). These data suggest that the force/EMG recording device is a well-shielded system that may be used in a wide range of studies for investigating neural control mechanisms of voluntary motor actions. In addition, the visual feedback system incorporated in the data acquisition system enabled the subjects to precisely control the exerted force, thus improving the data quality. The fMRI signal The fMRI signal is not a direct measure of synaptic activities or action potentials of cortical neurons; instead, it results from the so-called blood-oxygen-level-dependent (BOLD) effect (reviews: DeYoe et al. 1994; Kim and Ugurbil 1997; Ogawa et al. 1998). An increase in neural activity in a cortical region increases local blood flow. On the other hand, the consumption of oxygen in the region does not increase or increases only slightly (Fox and Raichle 1986; Fox et al. 1988). Consequently, the relative content of venous deoxyhemoglobin in the affected brain area is reduced. Because deoxyhemoglobin is paramagnetic, it produces microscopic magnetic inhomogeneities that increase the dephasing of spinning hydrogen protons. A decrease in the quantity of deoxyhemoglobin reduces the rate of dephasing and causes the magnetic resonance signal to decay at a slower rate. As a result, the signal is increased in the area where the uncoupling of the changes in blood flow and oxygen consumption occurs. Recent reports have shown that human fMRI signal was directly proportional to the firing rate of single neurons recorded in the same cortical region of monkey

(Heeger et al. 2000; Rees et al. 2000; however, see Disbrow et al. 2000). A study measured human surface EEG-derived MRCP from the electrodes overlying the contralateral sensorimotor cortex and the SMA during five levels of elbow flexion contractions. The magnitude of the MRCP from both electrode locations was directly proportional to the elbow flexion force (r=0.95) and surface EMG (r=0.85) of the elbow flexor muscles (Siemionow et al. 2000). These results suggest that the BOLD-based fMRI signal probably reflects synaptic activities of neural cells in the brain (Jueptner and Weiller 1995). A major limitation of fMRI measurement is that it cannot distinguish the signal of the output neurons from that of other neurons (e.g., interneurons). The fMRI measurement is an integrated signal probably contributed by synaptic activities occurring on all types of cortical neurons. In addition, because the BOLD effect-induced signal change lags the synaptic activity, it cannot capture the dynamic temporal activation pattern among the cortical fields that show fMRI signal changes. Relationship between muscle output and fMRI signal Primary sensorimotor cortex An interesting observation of this study was that in the primary motor and sensory cortices, the signal intensity increased linearly across the entire range of force levels tested. The number of activated pixels, however, only showed a linear increase from 20% to 65% force level. From 65% to 80% level, the number of activated pixels did not increase (Fig. 6A). Perhaps an increase in the number of pixels represented primarily an increase in the number of neurons participating in controlling force, and an increase in fMRI signal intensity represented mainly an increase in the discharge rate of these neurons. If these assumptions are true, then the results indicate that the nervous system cannot recruit additional neurons in the primary motor and sensory cortices at a force higher than 65% MVC level, but the discharge rate of the neurons continues to increase. It is worth noting that as the rate of increase in the number of pixels flattened from the 65% to 80% level (Fig. 6A), the rate of increase in the signal intensity from 65% to 80% force became higher (compare the rate from 65% to 80% to that from 50% to 65%). Perhaps the higher increase rate in signal intensity (neuron discharge rate) was intended to compensate for the inability to recruit additional neurons (pixels). Electrophysiological results have long indicated that the discharge rate of output neurons in the primary motor cortex of primates is directly related to force exerted by upper limbs (Cheney and Fetz 1980; Evarts 1968; Evarts et al. 1983; Hepp-Reymond et al. 1978, 1999; Werner et al. 1991). These results support our findings of increases in fMRI signal intensity at higher force levels, although the range of force that could be studied in primates was usually small and the upper level (amplitude) of force that could be examined in the animals was usually low.

298

Thus, we do not know whether the discharge rate of motor cortex cells in primates continues to increase at very high force levels (e.g., more than 65%). It is also difficult for single-cell studies to determine whether the number of participating cells increases with force.

that the CBL is important in controlling posture and balance, and it participates in motor learning (review: Ghez and Thach 2000). The finding of a linear involvement of the CBL in controlling human muscle force stresses the importance of the CBL in sensorimotor integration.

SMA and premotor cortex

Association cortices and correlated activation

We found that both the SMA and PM are proportionally activated with handgrip force. These two regions consist of Brodmann’s area 6, but each contributes to a different aspect of motor planning (Krakauer and Ghez 2000). The SMA has been shown to be important in programming sequential finger movements (Roland et al. 1980), storing information necessary for the orderly performance of multiple movements of the arm (Tanji and Shima 1994), and bimanual coordination (Brinkman 1981, 1984). The SMA is also associated with controlling muscle force. Smith (1979) has reported that the discharge rate of the SMA neurons is proportional to the finger grip force in monkeys. Dettmers et al. (1995) have found a positive relationship between dynamic force of human index flexion and cerebral blood flow in the SMA. Our fMRI data provide further evidence that the SMA is involved in scaling static force of finger flexor muscles. Similar to the MI and SI, the SMA showed a linear relationship between force and fMRI signal intensity. The number of activated pixels increased steeply from 20% to 65% force but leveled off from 65% to 80% force (Fig. 6). This similarity in the force and fMRI signal relation between the primary sensorimotor cortex and SMA suggests that a population of SMA neurons may share a similar function. The PM is considered to contribute to the selection of motor actions on the basis of visual cues (Halsband and Passingham 1985; Passingham 1986), and the activity of PM neurons is linked to motor set, i.e., intention to make a movement in response to the cue (Weinrich and Wise 1982; Wise et al. 1983). However, the discharge rate of PM neurons was also found to be related to movement velocity and acceleration (Kubota and Hamada 1978; Weinrich et al. 1984), and amplitude and direction (Kurata 1993). More recent studies have reported direct relationships between PM neuron firing rate and static force of finger pinch in monkeys (Hepp-Reymond et al. 1994, 1999). Our study further supports the conclusion that the PM neurons are involved in controlling static force of the finger muscles.

Three association cortices (PFC, CG, and PL) demonstrated a direct relationship between their fMRI signals and force. These association areas are all associated with sensorimotor integration. For example, the PFC is involved in the selection of upper limb movements (Frith et al. 1991), and in the CG a proportional relationship between cerebral blood flow and human finger flexion force has been reported (Dettmers et al. 1995). In the parietal association cortex (areas 5 and 7), neurons are involved in passive and active joint movements (area 5) and discharge in high rates (area 7) when the monkey purposefully projects his arm or manipulates with his hand (Mountcastle et al. 1975). Our results show, for the first time, that the activation level of prefrontal and parietal cortices is linearly related to muscle force. It is surprising to see that the slopes of the curves of the fMRI signal in the association cortices were similar to or even steeper than the slope for the primary motor cortex (Fig. 6). In particular, the number of activated pixels in the PFC (more than 1,200 at 80% force) was substantially larger than that in the MI and other cortical regions (~300 at 80% force) except the CBL. In a simple hierarchical control model, one would expect that the association cortices only give abstractive commands, which should not need a tremendous level of activation and linearity in force and activation relation. The secondary (SMA, PM) and primary (MI, SI) sensorimotor areas are expected to formulate and carry out details of the planning and execution for the task, which should demand a greater level of activation and activation linearity. The novel finding of linear involvement of multiassociation cortices in a relatively simple static force task deserves further investigation. A prominent feature of this study was that many cortical regions (MI, SI, SMA, PM, PL, CG, PFC, and CBL) showed a similar proportional relationship between muscle output and fMRI signal. This type of correlated or networked activation among various cortical fields raises an intriguing question of whether these brain regions activate in a parallel or a hierarchical manner. It seems unlikely that they were activated in a strictly hierarchical order, since it is not efficient to make six relays of signal transduction from a high-order association cortex to a primary functional area such as motor cortex. Increasing evidence has suggested that the central nervous system acts in a correlated or networked manner and the brain stores and processes information only when millions of neurons work together, perhaps with their electrical potentials correlated or synchronized in patterns at various frequencies (König and Engel 1995).

Cerebellum The fMRI signals recorded from the entire CBL showed a linear correlation with force. Dettmers et al. (1995) have reported a significant correlation between cerebral blood flow in cerebellar vermis and finger flexion force; due to technical limitations, these authors could not image the entire CBL. Numerous studies have shown

299

Multiple cortical-region activation during simple motor actions such as finger tapping has been reported in many neuroimaging studies (Colebatch et al. 1991; Ehrsson et al. 2000; Roland et al. 1980; Stephan et al. 1995; Yue et al. 2000). Many motor function-related brain regions are activated in a similar pattern (Dettmers et al. 1995). In animal experiments, correlated firing of cells between somatosensory and motor cortex (Murthy and Fetz 1992), between thalamus and sensory cortex (Johnson and Alloway 1994), and among intra- and interhemispheric regions (Nowak et al. 1995) has been observed. Synchronous or correlated activity among cortical regions has been suggested to have an important function in sensory-motor integration and memory (König and Engel 1995). Our data argue that generating precision static force of human upper limb muscles requires correlated activities of primary, secondary, and association cortical regions. Conclusions High-quality joint force, surface EMG, and fMRI data can be measured at the same time. The degree of muscle activation (force and EMG) is directly proportional to the amplitude of the brain signal determined by fMRI in the entire brain and individual primary, secondary, and association motor function-related cortical fields. The stronger fMRI signals recorded during exertion of higher force levels probably indicate that more cortical output neurons and interneurons (indicated by a larger number of activated pixels) participate in generating descending commands and processing additional sensory information. The activity (discharge rate) of individual neurons may also be higher (indicated by higher signal intensity) during stronger contractions. The similarity in the relationship between muscle output and fMRI signal in various cortical regions suggests that they participate in controlling finger force in a correlated manner. Acknowledgements This work was supported by NIH grants R01-NS35130, R01-NS37400, and R01-HD36725 to G.H. Yue, and by departmental research funds of Physical Medicine and Rehabilitation at the Cleveland Clinic Foundation. The Silent Vision (visual feedback) system was purchased by an Infrastructure Supplement Award, R01-NS37400. The authors thank Drs. I.M. Najm and E.P. Pioro, Department of Neurology, the Cleveland Clinic Foundation, for their assistance in outlining functional brain regions on the magnetic resonance images. Finally, we thank the anonymous reviewers whose comments improved the manuscript.

References Bandettini PA, Wong EC, Hinks RS, Tikofsky RS, Hyde JS (1992) Time course EPI of human brain function during task activation. Magn Reson Med 25:390–397 Boecker H, Kleinschmidt A, Requardt M, Hanicke W, Merboldt KD, Frahm J (1994) Functional cooperativity of human cortical motor areas during self-paced simple finger movements: a high-resolution MRI study. Brain 117:1231–1239

Brinkman C (1981) Lesions in supplementary motor area interfere with a monkey’s performance of a bimanual coordination task. Neurosci Lett 27:267–270 Brinkman C (1984) Supplementary motor area of the monkey’s cerebral cortex: short- and long-term deficits after unilateral ablation and the effects of subsequent callosal section. J Neurosci 4:918–929 Cheney PD, Fetz EE (1980) Functional classes of primate corticomotoneuronal cells and their relation to active force. J Neurophysiol 44:773–791 Colebatch JG, Deiber MP, Passingham RE, Friston KJ, Frackowiak RS (1991) Regional cerebral blood flow during voluntary arm and hand movements in human subjects. J Neurophysiol 65:1392–1401 Dettmers C, Fink GR, Lemon RN, Stephan KM, Passingham RE, Silbersweig D, Holmes A, Ridding MC, Brooks DJ, Frackowiak RS (1995) Relation between cerebral activity and force in the motor areas of the human brain. J Neurophysiol 74:802–815 Dettmers C, Connelly A, Stephan KM, Turner R, Friston KJ, Frackowiak RS, Gadian DG (1996) Quantitative comparison of functional magnetic resonance imaging with positron emission tomography using a force-related paradigm. Neuroimage 4:201–209 DeYoe EA, Bandettini P, Neitz J, Miller D, Winans P (1994) Functional magnetic resonance imaging (FMRI) of the human brain. J Neurosci Methods 54:171–187 Disbrow EA, Slutsky DA, Roberts TP, Krubitzer LA (2000) Functional MRI at 1.5 tesla: a comparison of the blood oxygenation level-dependent signal and electrophysiology. Proc Natl Acad Sci USA 97:9718–9723 Ehrsson HH, Fagergren A, Jonsson T, Westling G, Johansson RS, Forssberg H (2000) Cortical activity in precision- versus power-grip tasks: an fMRI study. J Neurophysiol 83:528–536 Evarts EV (1968) Relation of pyramidal tract activity to force exerted during voluntary movements. J Neurophysiol 31:14–27 Evarts EV, Fromm C, Kroller J, Jennings VA (1983) Motor cortex control of finely graded forces. J Neurophysiol 49:1199–1215 Fox PT, Raichle ME (1986) Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. Proc Natl Acad Sci USA 83:1140–1144 Fox PT, Raichle ME, Mintun MA, Dence C (1988) Nonoxidative glucose consumption during focal physiologic neural activity. Science 241:462–464 Frith CD, Friston K, Liddle PF, Frackowiak RS (1991) Willed action and the prefrontal cortex in man: a study with PET. Proc R Soc Lond B Biol Sci 244:241–246 Ghez C, Thach WT (2000) The cerebellum. In: Kandel ER, Schwartz JH, Jessell TM (eds) Principles of neural science. Elsevier, New York, pp 832–850 Halsband U, Passingham RE (1985) Premotor cortex and the conditions for movement in monkeys (Macaca fascicularis). Behav Brain Res 18:269–77 Heeger DJ, Huk AC, Geisler WS, Albrecht DG (2000) Spikes versus BOLD: what does neuroimaging tell us about neuronal activity? Nat Neurosci 3:631–633 Hepp-Reymond MC, Wyss UR, Anner R (1978) Neuronal coding of static force in primate motor cortex. J Physiol (Paris) 74: 287–91 Hepp-Reymond MC, Husler EJ, Maier MA, Qi HX (1994) Forcerelated neuronal activity in two regions of the primate ventral premotor cortex. Can J Physiol Pharmacol 72:571–579 Hepp-Reymond MC, Kirkpatrick-Tanner M, Gabernet L, Qi HX, Weber B (1999) Context-dependent force coding in motor and premotor cortical areas. Exp Brain Res 128:123–133 Johnson MJ, Alloway KD (1994) Sensory modulation of synchronous thalamocortical interactions in the somatosensory system of the cat. Exp Brain Res 102:181–197 Jueptner M, Weiller C (1995) Review: Does measurement of regional cerebral blood flow reflect synaptic activity? Implications for PET and fMRI. Neuroimage 2:148–156

300 Kim SG, Ugurbil K (1997) Functional magnetic resonance imaging of the human brain. J Neurosci Methods 74:229–243 Kim SG, Ashe J, Georgopoulos AP, Merkle H, Ellermann JM, Menon RS, Ogawa S, Ugurbil K (1993a) Functional imaging of human motor cortex at high magnetic field. J Neurophysiol 69:297–302 Kim SG, Ashe J, Hendrich K, Ellermann JM, Merkle H, Ugurbil K, Georgopoulos AP (1993b) Functional magnetic resonance imaging of motor cortex: hemispheric asymmetry and handedness. Science 261:615–617 König P, Engel AK (1995) Correlated firing in sensory-motor systems. Curr Opin Neurobiol 5:511–519 Krakauer J, Ghez C (2000) Voluntary movement. In: Kandel ER, Schwartz JH, Jessell TM (eds) Principles of neural science. Elsevier, New York, pp 756–781 Kubota K, Hamada I (1978) Visual tracking and neuron activity in the post-arcuate area in monkeys. J Physiol (Paris) 74: 297–312 Kurata K (1993) Premotor cortex of monkeys: set- and movementrelated activity reflecting amplitude and direction of wrist movements. J Neurophysiol 69:187–200 Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM, Poncelet BP, Kennedy DN, Hoppel BE, Cohen MS, Turner R, Cheng HM, Brandy TJ, Rosen BR (1992) Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci USA 89: 5675–5679 Liu JZ, Dai TH, Siemionow V, Sahgal V, Ranganathan VK, Yue GH (1999) Brain activation during muscle fatigue. Soc Neurosci Abstr 25:1915 Liu JZ, Dai TH, Elster TH, Sahgal V, Brown RW, Yue GH (2000) Simultaneous measurements of human joint force, surface EMG, and functional MRI-measured brain activation. J Neurosci Methods 101:49–57 Ludman CN, Cooper TG, Ploutz-Synder LL, Potchen EJ, Meyer RA (1996) Force of voluntary exercise does not affect sensorimotor cortex activation as detected by functional MRI at 1.5 T. NMR Biomed 9:228–232 Mountcastle VB, Lynch JC, Georgopoulos A, Sakata H, Acuna C (1975) Posterior parietal association cortex of the monkey: command functions for operations within extrapersonal space. J Neurophysiol 38:871–908 Murthy VN, Fetz EE (1992) Coherent 25- to 35-Hz oscillations in the sensorimotor cortex of awake behaving monkeys. Proc Natl Acad Sci USA 89:5670–5674 Nowak LG, Munk MH, Nelson JI, James AC, Bullier J (1995) Structural basis of cortical synchronization. I. Three types of interhemispheric coupling. J Neurophysiol 74:2379–2400 Ogawa S, Lee TM, Nayak AS, Glynn P (1990) Oxygenationsensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn Reson Med 14:68-78 Ogawa S, Tank DW, Menon R, Ellermann JM, Kim SG, Merkle H, Ugurbil K (1992) Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 89:5951–5955 Ogawa S, Menon RS, Tank DW, Kim S-G, Merkle H, Ellermann JM, Ugurbil K (1993) Functional brain mapping by blood oxygenation level-dependent contrast magnetic resonance imaging: a comparison of signal characteristics with a biophysical model. Biophys J 64:803-812

Ogawa S, Menon RS, Kim SG, Ugurbil K (1998) On the characteristics of functional magnetic resonance imaging of the brain. Annu Rev Biophys Biomol Struct 27:447–474 Passingham RE (1986) Cues for movement in monkeys (Macaca mulatta) with lesions in premotor cortex. Behav Neurosci 100: 695–703 Penfield W, Rasmussen T (1950) The cerebral cortex of man: a clinical study of localization of function. Macmillan, New York Rao SM, Binder JR, Bandettini PA, Hammeke TA, Yetkin FZ, Jesmanowicz A, Lisk LM, Morris GL, Mueller WM, Estkowski LD, Wong EC, Haughton VM, Hyde JS (1993) Functional magnetic resonance imaging of complex human movements. Neurology 43:2311–2318 Rao SM, Bandettini PA, Binder JR, Bobholz JA, Hammeke TA, Stein EA, Hyde JS (1996) Relationship between finger movement rate and functional magnetic resonance signal change in human primary motor cortex. J Cereb Blood Flow Metab 16: 1250–1254 Rees G, Friston K, Koch C (2000) A direct quantitative relationship between the functional properties of human and macaque V5. Nat Neurosci 3:716–723 Roland PE, Larsen B, Lassen NA, Skinhoj E (1980) Supplementary motor area and other cortical areas in organization of voluntary movements in man. J Neurophysiol 43:118–136 Sanes JN, Donoghue JP, Thangaraj V, Edelman RR, Warach S (1995) Shared neural substrates controlling hand movements in human motor cortex. Science 268:1775–1777 Siemionow V, Yue GH, Ranganathan VK, Liu JZ, Sahgal V (2000) Relationship between motor activity-related cortical potential and voluntary muscle activation. Exp Brain Res 133: 303–311 Smith AM (1979) The activity of supplementary motor area neurons during a maintained precision grip. Brain Res 172:315–327 Stephan KM, Fink GR, Passingham RE, Silbersweig D, CeballosBaumann AO, Frith CD, Frackowiak RS (1995) Functional anatomy of the mental representation of upper extremity movements in healthy subjects. J Neurophysiol 73:373–386 Tanji J, Shima K (1994) Role for supplementary motor area cells in planning several movements ahead. Nature 371:413–416 Thickbroom GW, Phillips BA, Morris I, Byrnes ML, Mastaglia FL (1998) Isometric force-related activity in sensorimotor cortex measured with functional MRI. Exp Brain Res 121:59–64 Tyszka JM, Grafton ST, Chew W, Woods RP, Colletti PM (1994) Parceling of mesial frontal motor areas during ideation and movement using functional magnetic resonance imaging at 1.5 tesla. Ann Neurol 35:746–749 Weinrich M, Wise SP (1982) The premotor cortex of the monkey. J Neurosci 2:1329–1345 Weinrich M, Wise SP, Mauritz KH (1984) A neurophysiological study of the premotor cortex in the rhesus monkey. Brain 107: 385–414 Werner W, Bauswein E, Fromm C (1991) Static firing rates of premotor and primary motor cortical neurons associated with torque and joint position. Exp Brain Res 86:293–302 Wise SP, Weinrich M, Mauritz KH (1983) Motor aspects of cue-related neuronal activity in premotor cortex of the rhesus monkey. Brain Res 260:301–305 Yue GH, Liu JZ, Siemionow V, Ranganathan VK, Ng TC, Sahgal V (2000) Brain activation during human finger extension and flexion movements. Brain Res 856:291–300