WO2003100450A1

WO2003100450A1 - Simultaneous acquisition of bioelectric signals and magnetic resonance images

Info

Publication number: WO2003100450A1
Application number: PCT/JP2003/006737
Authority: WO
Inventors: Kimitaka Anami
Original assignee: Physio-Tech Co., Ltd.
Priority date: 2002-05-29
Filing date: 2003-05-29
Publication date: 2003-12-04
Also published as: JP2006166929A; AU2003241886A1

Abstract

A method and a system for simultaneous acquisition of bioelectric signals and magnetic resonance images, and a pulse sequence that is used for recording of bioelectric signals during magnetic resonance imaging without imaging artifacts on the bioelectric signals. With this pulse sequence, each readout gradient pulse has a waveform with a ramp up segment during which the amplitude of the readout gradient pulse increases, a null segment during which the amplitude of the readout gradient pulse does not change, and a ramp down segment during which the amplitude of the readout gradient pulse decreases. Each null segment is sandwiched between a ramp up segment and a ramp down segment. The combination of a ramp up segment, null segment, and ramp down segment is repeated a predetermined number of times, and the bioelectric signals are recorded during application of each gradient pulse only within the respective readout null segments.

Description

DESCRIPTION

SIMULTANEOUS ACQUISITION OF BIOELECTRIC SIGNALS AND MAGNETIC RESONANCE IMAGES

Technical Field The present invention relates to the simultaneous acquisition of bioelectric signals and magnetic resonance images . More specifically, the present invention relates to a method and a system for the simultaneous acquisition of bioelectric signals and magnetic resonance images , and a pulse sequence that is used for the digital sampling of bioelectric signals during magnetic resonance imaging.

Background Art

Magnetic resonance imaging (MRI) , which uses magnetic fields and radio waves sent from a scanner to produce images of the internal structures of the human body, is a non-invasive diagnostic imaging modality that provides anatomical and functional information of the human body. MRI is based on the principle of nuclear magnetic resonance (NMR) . According to this principle, if nuclei are placed in a static magnetic field and excited by an oscillatory magnetic field, the nuclei emit useful signals. In MRI, a subject is placed in a strong static magnetic field and hydrogen nuclei in the body emit signals. In order to produce magnetic resonance (MR) images of the body from these signals , the location from which each signal comes must be determined. The MRI modality determines their location in terms of x, y and z coordinates. This is done by the use of varying (gradient) magnetic fields along the x, y, and z axes within a static magnetic field.

The MRI principle requires a subject to remain inside a magnet for as long as 30 minutes or more during the scanning procedure.

During this period, there is the possibility for the magnetic fields and radio waves to adversely affect the subject. Thus, it is sometimes desirable to monitor bioelectric signals, such as electrocardiographic (ECG) signals, from the body of the subject concurrent to the MRI scanning. Such concurrent acquisition of bioelectric signals and MRI provides potential benefits , especially for higher risk patients with, for example, heart disease. In fact, conventional MRI systems are equipped with a monitoring device to measure bioelectric signals , such as ECG signals, when a subject is undergoing MRI scanning. However, the practical measurement of such bioelectric signals has encountered many technical problems which will be described later.

Concurrent or simultaneous acquisition of bioelectric signals and MRI is also advantageous to other fields of study that associate MR images with certain physiological phenomena. A typical example is the simultaneous acquisition of an electroencephalogram (EEG) and a functional MRI (fMRI). Functional MRI is a variation of the traditional MRI and it is becoming the clinical technique of choice for mapping brain functions onto magnetic resonance images as well as for assessing the potential risks for invasive treatment of the brain. Although positron emission topography (PET) and single photon emission computer topography (SPECT) have made it possible to map brain function during various cognitive tasks, these imaging modalities, because of their poor temporal resolution, cannot yet evaluate a discrete activity of the brain related exclusively to task execution.

Electromagnetic imaging, on the other hand, can provide good temporal resolution. Discrete activities of the brain, such as perception, cognition, and memory occur within extremely short instants of time. This means that an imaging technique with a temporal resolution of as short as 1 millisecond is essential to the in-depth investigation of specific brain functions. The electroencephalographic (EEG) and magnetoencephalographic (MEG) modalities are the only two options now available to reliably capture such brain activity in a microsecond. These modalities do allow the monitoring of electrical and magnetic phenomena that occur in the brain within such a short period of time. However, they also present such serious disadvantages as poor spatial resolution and inapplicability to the so-called "subcortical gray and white matters" of the brain. With the EEG and MEG modalities, it is almost impossible to determine how and where each brain phenomenon occurs . To offset these serious limitations , functional MRI was developed as an alternate imaging modality, and now combining EEG with functional MRI is expected to provide newer possibilities for in-depth brain research. This merged modality should offer broad utility ranging from neurophysiological investigations to clinical examinations .

The simultaneous acquisition of bioelectric signals and MRI is greatly complicated by troublesome interference from the static and gradient magnetic fields . Any change in magnetic fields crossing through a loop induces an electromotive force (EMF) as described by Faraday's Law. For example, simultaneous EEG/fMRI involves the placement of EEG electrodes onto a sub ect ' s head prior to its introduction into an RF coil, such as a head coil, a surface coil, and a body coil. This arrangement creates a loop between each EEG electrode and the subject's head. With such a setup, even minute movements , including normal pulsations of the heart , induce electromotive forces (EMFs) in the loop. The changing gradient fields and the radio waves induce much larger electromotive forces .

The electromotive force is but one cause of so-called artifacts that interfere with EEG signals . The practical mapping of brain phenomena onto fMRI images has encountered many technical problems associated with these artifacts. The most problematic artifacts for simultaneous EEG/fMRI are those associated with MRI acquisition and those deriving from cardiac pulse motion in the head (even when no images are being acquired) . The former artifact is referred to herein as "imaging artifact". The latter, cardiac-related artifact is designated herein a "ballistocardiogram" or "ballistocardiographic artifact" . These two terms are used interchangeably for referring to a broadband (electrostatic) noise component resulting from electromagnetic induction caused by cardiac pulses . Although the term "ballistocardiogram" commonly refers to a recording made by a ballistocardiograph, this term also indicates a cardiac-related artifact caused by ballistic head movement due to cardiac pulsation within the fMRI field.

A ballistocardiographic artifact is inseparably related to the pulsation of the heart. Even minute movements of the head, due to pulsation, induce EMF in a loop (Faraday's law) between each EEG electrode and the subject's head. Much larger artifacts, however, are observed during fMRI acquisition. Such artifacts include the ballistocardiographic and imaging artifacts which result from the changing gradient fields and the radio frequency (RF) pulses essential to fMRI.

Some solutions to the ballistocardiographic artifacts have already been proposed. These include the use of: (1) bipolar montage instead of monopolar montage, (2) Hjorth transform, (3) a spatial filter, (4) a bite bar for fixing the subject's head,

(5) twisted pair cables, and (6) an average subtraction method for artifacts . These solutions have been partially successful in suppressing the ballistocardiographic artifacts but some artifact components still obscure the EEG signals . In respect to this nuisance, the present inventors attempted to suppress artifacts by using a vacuum cushion system. Conventionally, vacuum cushion systems are used in the field of therapeutic radiology to immobilize an affected area of a patient's body for the treatment of cancer. Adoption of the vacuum cushion system has resulted in the successful reduction of ballistocardiographic artifacts even with the use of monopolar montage.

After the reduction of ballistocardiographic artifacts, the large imaging artifacts induced during fMRI acquisition still interfere with EEG. As to simultaneous acquisition of EEG and fMRI , imaging artifacts, including broadband "spiky" components with an amplitude of around 2,000 μV, appear on the EEG signals. Attenuating these components has proven challenging.

Interleaved recording of bioelectric signals and MRI provides a partial solution for this artifact-relatedproblem. The interleaved recording enables bioelectric signals to be retrieved during the process of MRI scanning. In other words, a bioelectric signal and a magnetic resonance image are recorded serially and alternately. As to the simultaneous acquisition of EEG and f I, the interleaved method allows recording of EEG signals in an interleaved manner during the non-imaging intervals (1 to 2 seconds) between fMRI scans.

The drawback of this method, however, is the potential for spontaneous or uncontrolled physiological events (e.g., EEG phenomena such as epileptic spikes) to occur during an imaging period, and to be missed by the interleaved recording process . The only solution to this nuisance is to record desired bioelectric signals and MRI truly concurrently while avoiding artifacts .

To cope with this problem, several techniques for suppressing or reducing artifacts have been reported. Kreger and Giordano disclose an adaptive filtering system for reducing the artifact on bio-potential signals as generated by rapidly switched gradient fields in MRI (U.S. Patent No. 5,436,564). Although the method disclosed by Kreger and Giordano is targeted mainly at ECG signals , similar results would be expected for EEG readings. Nevertheless, the extent to which any substantial effect is achieved in the results of bioelectric signal measurement, using this method, is not ascertained from Kreger and Giordano's disclosure. Moreover, as can be seen from the artifacts of Fig. 6 of the '564 patent, the artifact waveforms differ from one another. Each artifact has a much higher frequency than a sampling frequency for bioelectric signals. Therefore, according to the sampling theorem, no precise measurement can be made at a lower sampling frequency. In addition, the artifacts have a significantly large amplitude. As a result, the bioelectric signals are sampled inevitably at different instants of time for each artifact . In other words , some bioelectric signals may be samplednear the beginning of an artifact, others near the center, and still others may be sampled near the end of an artifact. Each artifact has a different waveform after low-pass filtering and the difficulty of artifact removal is intensified. It would seem the possibility for artifacts to be controlled effectively by this method, at least as shown by Figs.

6 and 7 of the '564 patent, is remote even when an adaptive filter is used; and a considerable amount of residual artifact will still be found.

Hoffmann et al. use a fast Fourier transform (FFT) to filter imaging artifacts in a given frequency range in post-hoc processing

(Hoffmann, A., Jager, L., Werhahn, K.J., Jaschke, M. , Noachtar, S., Reiser, M. 2000. Electroencephalography during functional echo-planar imaging: detection of epileptic spikes using postprocessing methods. Magn . Reson . Med. 44(5): 791-798). Such post-hoc processing requires considerable mathematical computation. FFT is helpful in removing artifacts but it also removes some EEG signals that appear in the same frequency range as that used for artifact removal. The removal of some important parts of the EEG spectra is a serious drawback of this method.

Alternately, Allen et al. disclose that imaging artifacts can be reduced by subtracting from the EEG an averaged artifact waveform, followed by adaptive noise cancellation to reduce any residual artifact (Allen, P.J., Josephs, O. , Turner, R. 2000. A method for removing imaging artifact from continuous EEG recorded during functional MRI. Neuroimage 12(2 ): 230-239) . Their achievement is notable in terms of being the world' s first practical approach to obtaining a truly simultaneous acquisition of EEG and f RI , but even this technique also has some significant drawbacks . In this method, an averaged imaging artifact waveform is subtracted from the EEG, and then adaptive noise cancellation is used to attenuate residual artifacts. Although the subtraction followed by adaptive noise cancellation removes most of the artifacts, there do remain some residual artifacts as can often be observed. The amplitudes of these residual artifacts are significantly lower than those of the original artifacts but still are not negligible because the EEG signals are naturally very low in amplitude. Additionally, the algorithm that Allen et al. use to synchronize artifact averaging requires the capturing of a functional MRI slice-timing signal with a resolution of 10 microseconds. It is, however, impossible to exactly determine the starting point of the artifact, a measurement which is essential to the algorithm of Allen et al. This variation in the starting point results in significant fluctuation of artifact waveforms. In short, the averaged subtraction is inadequate for removing artifacts . To obtain EEG signals at a satisfactory level, it is necessary to use an adaptive filter involving substantial mathematical processing. Additionally, only EEG signals having spectra of up to about 50 Hz can be captured. The removal of some important parts of the EEG spectra is another drawback of this method.

Furthermore, according to experiments made by the present inventors, the shift of a sampling point by 1 μs sometimes results in a difference in underlying imaging artifact by as much as 70 μV during the readout period. It is extremely difficult to precisely capture the starting point of each artifact . Allen et al. solved this problem by increasing the sampling frequency. However, increasing the sampling frequency is not a complete solution because a sampling frequency higher than 1 MHz would be required to keep the error below 70 μV. Such a high sampling frequency is impractical in EEG measurement where typical sampling frequencies are around 1,000 Hz. In consideration of the above-mentioned drawbacks, an object of the present invention is to provide a truly simultaneous acquisition of bioelectric signals and magnetic resonance images and to provide a pulse sequence and a system to achieve such simultaneous acquisition.

Disclosure of Invention According to a first aspect, the present invention provides a method performed in a combined system capable of acquiring a magnetic resonance image and recording bioelectric signals at predetermined intervals , the bioelectric signal being an electric signal from the body of a subject (e.g., a mammalian subject, especially a human subject). This method comprises an image acquisition step for acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis , a phase encoding axis, and a readout axis; and a bioelectric signal recording step for recording a bioelectric signal. Each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired has a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment , the combination of a readout ramp up segment, readout null segment , and readout ramp down segment being repeated a predetermined number of times . The combined system is configured to record the bioelectric signals within the readout null segment , the recording thereby being performed without imaging artifacts on the bioelectric signal.

In this method, the time durations of all readout ramp up segments may be generally equal, the time durations of all readout ramp down segments may be generally equal, the time durations of all readout null segments may be generally equal, and the time durations of all conjoined readout ramp down/up segments may be generally equal. The recording during the readout phase may be performed at constant intervals within predetermined readout null segments. As a practical example, recording during the readout phase may be performed within all readout null segments.

Alternately, in the method according to the first aspect of the present invention, the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, and the time durations of all readout null segments are generally equal. The recording during the readout phase may be performedwithin every other readout null segment .

In the present invention, the readout ramp up and readout ramp down periods may have "generally equal" time durations. This ensures that "readout null segments (readout null periods)" will be generated at constant intervals, thus allowing for constant recording of bioelectric signals. It should be noted that while, in principle, all readout ramp up segments or periods have exactly the same time durations, in practice, it may actually be difficult to equalize the time durations of all readout ramp up segments due to various reasons, such as, for example, limitations in the performance of an amplifier. Therefore, the present invention encompasses "all readout ramp up segments" being generally identical in time duration. This being the case, the conditions where bioelectric signals can be recorded (sampled) at certain constant intervals is satisfied by the term "generally identical" or "generally the same" (or "generally equal"). The same applies to the readout ramp down segments and the readout null segments. In the method according to the first aspect of the present invention, the bioelectric signal may be selected from the group consisting of an electroencephalographic signal, an electrocardiographic signal, an electrooculographic signal, an electromyographic signal, a spirographic signal, a galvanic skin response signal, an electrogastrographic signal, and a pupillary reflex signal. In addition, the magnetic resonance imaging may be functional magnetic resonance imaging. Recording may be performed at constant intervals ranging between 0.1 milliseconds and 10 milliseconds (e.g., at 1-millisecond intervals). In addition, according to a second aspect, the present invention provides a method performed in a magnetic resonance imaging system capable of acquiring a magnetic resonance image which is used in combination with a predetermined device capable of recording bioelectric signals at predetermined intervals, the bioelectric signal being an electric signal from the body of a subject, the method comprising an image acquisition step for acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis, a phase encoding axis, and a readout axis, each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases, each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment, the combination of a readout ramp up segment, readout null segment , and readout ramp down segment being repeated a predetermined number of times , the recording during the readout phase being performed by the predetermined device within the readout null segment , the recording thereby being performed without imaging artifacts on the bioelectric signal.

According to a third aspect, the present invention provides a method performed in a combined system having a magnetic resonance imaging system capable of acquiring a magnetic resonance image and a predetermined device capable of recording bioelectric signals at predetermined intervals , the bioelectric signal being an electric signal from the body of a subject, the method comprising an image acquisition step for acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis , a phase encoding axis , and a readout axis ; a clock generation step for generating clock pulses at a first predetermined frequency; a sampling pulse generation step for generating sampling pulses from the clock pulses by means of dividing the first predetermined frequency into a second predetermined frequency; and a bioelectric signal recording step for recording a bioelectric signal according to the sampling pulses , each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment , the combination of a readout ramp up segment , readout null segment , and readout ramp down segment being repeated a predetermined number of times, the recording during the readout phase being performed within the readout null segment , the recording thereby being performed without imaging artifacts on the bioelectric signal.

In addition, the present invention also provides a combined system comprising a magnetic resonance imaging system capable of acquiring a magnetic resonance image, the magnetic resonance imaging system having a magnet coil assembly adapted to apply gradient pulses to a slice selection axis, a phase encoding axis, and a readout axis to acquire a magnetic resonance image; and a predetermined device capable of recording bioelectric signals at predetermined intervals, the bioelectric signal being an electric signal from the body of a subject, each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with readout ramp up segments during which the amplitude of the gradient pulse increases , readout null segments during which the amplitude of the gradient pulse does not change, and readout ramp down segments during which the amplitude of the gradient pulse decreases, each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment , the combination of a readout ramp up segment, readout null segment, and readout ramp down segment being repeated a predetermined number of times , the recording during the readout phase being performed by the predetermined device within the readout null segment , the recording thereby being performed without imaging artifacts on the bioelectric signal. Furthermore, the present invention also provides a pulse sequence program used in a magnetic resonance imaging system capable of acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis , a phase encoding axis , and a readout axis , said program comprising causing the magnetic resonance imaging system to acquire the magnetic resonance image, wherein each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment, the combination of a readout ramp up segment, readout null segment, and readout ramp down segment being repeated a predetermined number of times .

Brief Description of Drawings Fig.1 shows an example of ballistocardiographic and imaging artifacts superimposed on the EEG, occurring when fMRI images are being acquired using conventional simultaneous acquisition of EEG/fMRI;

Fig. 2 shows EEG signals and artifacts which have been subjected to low-pass filtering at a cut-off frequency of 100 Hz; Fig. 3 is a view used to describe a conventional interleaved EEG/fMRI method;

Fig. 4 shows the details of a single spiky component of a waveform (Fig.2) in which the imaging artifacts have been subjected to low-pass filtering at a cut-off frequency of 3,000 Hz;

Figs . 5A to 5C show the relationships among a gradient waveform, an imaging artifact, and EEG sample points, all obtained at a sampling frequency of 1,000 Hz by using a conventional blip type pulse sequence;

Figs . 5D to 5F are similar to Figs . 5A to 5C except that the pulse sequence being used is a stepping stone sequence according to the present invention;

Fig. 6A is the timing diagram for a stepping stone sequence, according to the present invention, along the three gradient axes (Gs , Gr, Gp) ;

Fig. 6B is the corresponding waveform of an imaging artifact obtained when gradient pulses are applied as shown in Fig. 6A;

Fig. 6C is an enlarged view of the initial 30-millisecond portion of the timing diagram shown in Fig. 6A;

Fig.6D is an enlarged view of a readout portion of the timing diagram shown in Fig. 6A;

Fig.7 is a schematic block diagram showing the configuration of a combined system for simultaneous acquisition of EEG/fMRI according to an embodiment of the present invention;

Fig. 8 is a schematic block diagram showing essential components of a clock divider in the system shown in Fig. 7;

Fig. 9 is a time chart for signals in the various components of the clock divider; Fig. 10 is a schematic block diagram showing the configuration of a combined system for simultaneous acquisition of EEG/fMRI according to a second embodiment of the present invention;

Fig. 11 is a schematic block diagram showing the configuration of a combined system for simultaneous acquisition of EEG/fMRI according to a third embodiment of the present invention;

Fig. 12 shows the result of a clinical test made on a human subject using a combined system for simultaneous acquisition of EEG/fMRI according to one of the embodiments of the present invention;

Fig.13 provides a comparison of artifact waveforms obtained with stepping stone sampling and a conventional sampling method; Fig. 14 shows EEG data for a 10-millisecond period after artifact correction;

Fig. 15A shows the power spectrum density obtained by estimation of aliasing contamination on human EEG data without fMRI acquisition; Fig. 15B shows the power spectrum density obtained by estimation of aliasing contamination on human EEG data with fMRI acquisition;

Fig. 16 shows the waveforms obtained through two different estimation methods on EEG data obtained at a sampling frequency of 20,000 Hz;

Fig.17A shows the results of a clinical experiment on a human subject using a visual checker board stimulation paradigm, with a conventional pulse sequence;

Fig.17B shows the results of a clinical experiment on a human subject using a visual checker board stimulation paradigm, with the stepping stone sequence according to the present invention;

Fig.18A shows an MRI slice image on which an epileptic region is mapped using stepping stone sampling;

Fig. 18B shows a conventional magneto encephalogram with an epileptic region;

Figs. 19A and 19B show the characteristic waveform of an averaged event related potential on Cz and Pz electrodes , respectively, obtained during 15% presentation of a target stimulus;

Figs. 19C and 19D show the characteristic waveform of an averaged event related potential on Cz and Pz electrodes , respectively, obtained during 30% presentation of a target stimulus; and Figs. 20A and 20B show mapping results on MRI slice images during 15% and 30% presentations of the target stimuli.

Best Mode for Carrying Out the Invention The conditions leading to the present invention are described first for the purpose of facilitating an understanding of the present invention. The description is made for the case in which a bioelectric signal is an EEG signal and functional MR images are acquired in parallel to EEG sampling. In the following description, the terms "sampling", "digital sampling" and "recording" are used interchangeably unless otherwise specified.

Functional MRI provides a technique for locating an activated region of the brain. Activated regions can be mapped using "blood oxygen level dependent" (BOLD) signals. This fMRI modality is based on the acquisition of nuclear magnetic resonance (NMR) signals that reflect variations in the magnetic field generated according to the deoxyhemoglobin levels in the blood. Functional MR images of the brain are "image slices" through the brain, as in traditional MR images . Typically, the entire brain can be imaged with twenty to thirty slices which together are called a "volume scan". In other words, one "volume scan" consists of twenty to thirty slices .

In fMRI, the acquisition of a single image slice takes about

100 milliseconds. When a subject with EEG electrodes placed on his/her head is introduced into an RF coil for simultaneous acquisition of EEG and fMRI, the changing gradient fields and the radio frequency pulses required for MRI acquisition inevitably induce voltages between the head of the subject and each EEG electrode. These voltages are the source of artifacts which superimpose on the EEG signals. This is shown in Fig. 1 in which EEG signals are completely obscured by ballistocardiographic artifacts 100 and imaging artifacts 200. The imaging artifacts 200 appear as a dense band throughout each imaging period for a volume scan, which is indicated as a black rectangle at the bottom of the waveform in Fig. 2. It should be noted that ballistocardiographic artifacts are sometimes overlapped by imaging artifacts and become hidden in them.

Fig. 2 shows EEG signals and artifacts which have been subjected to low-pass filtering at a cut-off frequency of 100 Hz for the case where one volume scan consists of twenty-four slices. In Fig. 2, one "band" of an imaging artifact 200' corresponds to one volume scan and consists of twenty-four "spiky" components corresponding to twenty-four slices. An EEG signal 300 can be observed only partially due to the imaging artifacts 200'. The above-mentioned interleaved EEG/fMRI was developed for the very reason that it is impossible to extract EEG signals from the artifact/EEG data shown in Fig. 2. The interleaved method allows recording of EEG signals during non-imaging intervals (1-2 seconds each) between fMRI scans. This is shown in Fig.3, in which EEG recording and stimulus presentation windows 310 are alternated with fMRI acquisition windows 210. The interleaved method makes it possible to record EEG signals 300 at an acceptable level, but it also has a significant drawback. There is the potential for spontaneous brain phenomena to occur during the fMRI imaging period when it cannot be captured by the interleaved method. The solution to this problem is the development of a truly simultaneous acquisition of EEG and fMRI data.

Incidentally, since no EEG signals ever approach such a high frequency as characterized by imaging artifacts, a low-pass filter with a cut-off frequency of around 100 Hz is typically used.

Filtering at this low cut-off frequency can efficiently remove the higher waveform components which are useless anyway. Even though an amplitude of around 2,000 μV or more is observed for imaging artifacts, such an amplitude should be a consequence of low-pass filtering. The present inventors speculated that the original imaging artifacts must have a much larger amplitude because the use of low-pass filtering would have significantly reduced the recorded amplitude of the artifacts. Based on this theory, the present inventors made observations of imaging artifacts at a much higher sampling frequency (20 kHz) and a higher cut-off frequency (3,000 Hz) than typically used for conventional EEG sampling. Their observations showed that imaging artifacts obscuring EEG signals during fMRI scanning typically have a waveform profile as shown in Fig. 4. This is the waveform of an imaging artifact after low-pass filtering at a cut-off frequency of 3,000 Hz over a 100-msec period (equivalent to one slice of the brain) . This waveform detail corresponds to one spiky component of a full waveform as shown in Fig. 2. Their observations also revealed that an imaging artifact that appears to have an amplitude of around 2,000 μV, as shown in Fig. 4, actually has an amplitude of around 40,000 μV at a high frequency of up to 800 Hz . The fact is not that the imaging artifact has an amplitude of about 2,000 μV but that because of low-pass filtering, imaging artifacts contain a decreased number of spiky components, and their amplitude is attenuated from 40,000 μV down to about 2,000 μV, as shown in Fig.2. This finding is significant. Typical EEG amplifiers have a dynamic range of up to 40,000 μV. If an imaging artifact has an amplitude of 40,000 μV or more, it passes beyond this range, and cannot be amplified with such an EEG amplifier. When a spiky component of an imaging artifact exceeds the dynamic range, the vertex is truncated (becomes flat) and the signal becomes useless. Mistakenly, the useless signal is seen to be a continuous good signal after being subjected to low-pass filtering. Observations made by the present inventors demonstrate that such a problem can actually occur during clinical examinations and laboratory investigations .

In addition, using a cut-off frequency of 3,000 Hz for filtering clarifies the "real" shape of the imaging artifact. It has been found that the imaging artifact has an extremely high frequency. The higher frequency produces waveform spikes of extremely small width. The relatively wide gaps between artifact spikes make sampling at the baseline much easier even with a sampling frequency of 1, 000 Hz (i.e. , sampling every one millisecond) . This concept is the basis for a unique pulse sequence according to the present invention.

At present, conventional EEG recording takes digital samples at equal intervals with a sampling period of, for example, about 1 millisecond (sampling frequency Fs = 1,000 Hz). During a

100-millisecond imaging period for a single fMRI image slice, one hundred EEG sample points are recorded. In this situation, some

EEG sample points are in the vicinity of the baseline of the imaging artifact and others are near the peak of the artifact . Still other

EEG sample points may register between the peak and the baseline.

It would seem the loci of sample points is determined almost by chance. EEG sample points on or around the baseline can be captured clearly, but remainders are obscured by the imaging artifacts. This is shown graphically in Fig. 5 which also shows the relationships among a gradient waveform, an imaging artifact, and EEG sample points obtained at a sampling frequency of 1,000 Hz by using a conventional blip type pulse sequence (ep2d_fid_60b2080_62_64.ekc available from Siemens AG) . Fig. 5A shows the timing diagram of a gradient field 500. For this simple illustration, the gradient field 500 is the superposed result of three gradient fields which are normally labeled Gs, Gp, and Gr. In MRI systems , gradient pulses are applied to three gradient axes : the slice selection axis (Gs), the phase encoding axis (Gp), and the readout axis (Gr) . Fig. 5B shows the timing diagram of an imaging artifact 510 that occurs when the gradient field 500 shown in Fig.5A is applied. Fig. 5C shows EEG sample points 91 recorded during the same imaging period. As shown in Fig.5C, some EEG sample points are near the peak of an artifact 520. For comparison. Fig. 5D is similar to Fig. 5A except that Fig. 5D represents a pulse sequence according to the present invention. The gradient field 530 is the superposed result of three gradient fields which are normally labeled Gs , Gp, and Gr. In other words, the gradient field 530 corresponds to a spoiler gradient. a slice selection gradient, a dephasing and rephasing gradient, and to readout gradient pulses 603 to 606, which will be described later in conjunction with Fig.6. Fig.5E shows the timing diagram of an imaging artifact 540 that occurs when the gradient field 530 shown in Fig. 5D is applied. Fig. 5F shows EEG sample points 91 recorded during the same imaging period. As shown in Fig. 5F, all

EEG sample points are at the baseline level.

As described above, the present inventors made observations of imaging artifacts at a sampling frequency of 20 kHz and a cut-off frequency of 3,000 Hz. When the present inventors compared the waveform of the 3,000-Hz low-pass filtered imaging artifacts and the pulse sequence used, the results implied that each spiky component of an imaging artifact corresponds precisely to the gradient pulses and RF pulses . Thus , the following conclusions were drawn from the comparison.

(1) Most spiky components of an imaging artifact are precisely associated with gradient pulses that are produced by the gradient coil in an MRI system.

(2) Imaging artifacts resulting from RF pulses are much smaller in amplitude than those resulting from changing gradient fields . The amplitude of an imaging artifact resulting from changing gradient fields is approximately 40,000 μV while the visible amplitude of an imaging artifact resulting from RF pulses is similar to the amplitude of background noise or, at most, several hundred micro volts.

(3) The waveform of an imaging artifact resulting from changing gradient fields depends, in principle, on Faraday's law. This means that, in principle, the artifact from one gradient pulse has the differential waveform of the waveform of the corresponding gradient pulse.

Based on the above observations and analysis , the present inventors found, after great consideration, that previous knowledge about gradient and RF pulses defined by a pulse sequence could be used to estimate the onset, duration, and shape of each spiky component of an imaging artifact . This means that it is possible to predict and control the waveform and time schedule of imaging artifacts. Based on this theory, and in order to attenuate the large amplitude of the imaging artifact, the present inventors provide a strategy in which when EEG sampling and fMRI scanning are completely synchronized, the differential waveforms of gradient pulses periodically make artifact-free gaps that allow

EEG sampling with a high signal-to-artifact ratio.

1. Determining the fMRI Pulse Sequence

An embodiment of a unique pulse sequence, according to the present invention, is described in this section. This exemplary pulse sequence was specifically programmed for the proposed "simultaneous EEG/fMRI acquisition system" by means of modifying the above-mentioned typical blip type conventional pulse sequence. In the following description, the amplitudes are schematic only and are not drawn to scale.

To achieve a condition of simultaneous acquisition of EEG and fMRI , the following parameters were considered in terms of the imaging artifacts that appear when fMRI images are being acquired.

(A) What is being emphasized is the timing of the gradient pulses (produced by the gradient coil in an MRI system) - rather than the RF pulses - because the imaging artifacts are not greatly affected by RF pulses . (B) The waveform of an imaging artifact, which is a differential of the waveform of the gradient fields, should return to the baseline at constant intervals (e.g. , 1 millisecond) . This can be achieved by (1) ensuring the time duration throughout which the gradient fields are changing (i.e., the time duration throughout which the value of the differentiation is not equal to zero) does not exceed the duration of each interval (e.g., 1 millisecond) and by (2) ensuring that the time interval between the starting point of a "gradient plateau" and that of the adjacent gradient plateau is at least equal to the above-mentioned constant interval (e.g., 1 millisecond). The term "gradient plateau" as used herein means the time duration throughout which the gradient fields do not change even when the gradient coil is in an ON state (i.e., the time duration throughout which the value of the differentiation becomes equal to zero) .

(C) The simplest possible pulse sequence is designed with a minimum number of changes in gradient pulses .

On the basis of parameters (A) to (C) above, the present inventors modified and adjusted the blip type pulse sequence and thereby developed a unique pulse sequence . The pulse sequence according to the present invention is designated "stepping stone sequence". As the name implies, the stepping stone sequence produces a line of periodic gradient plateaus . The differentiation of each gradient plateau has a value of zero, thus providing a "sampling shelf" onwhich EEG sampling can be made without artifacts . In other words, the waveform of an imaging artifact inevitably returns to the baseline at a predetermined interval, allowing for artifact-free EEG recording.

A better understanding of the stepping stone sequence can be obtained from Table 1 which provides a comparison between the stepping stone sequence and a conventional blip type pulse sequence

(ep2d_fid_60b2080_62_64.ekc available from Siemens AG) . The left half of the table contains data for a conventional blip type pulse sequence while the right half of the table represents the stepping stone sequence. The bandwidth of the readout gradient pulse is

1,136 Hz in this embodiment.

TABLE 1

t σ

Details of the modifications required to provide a stepping stone sequence are as follows . The number of spoiler pulses that occur in advance of the readout gradient pulses is reduced by removing the first spoiler pulse. In addition, the timing and duration are modified, as shown in Table 1, for the second spoiler pulse (Spoiler 2), the preparation gradient plO, the slice selective gradient, the dephasing and rephasing gradient, the readout gradients, the preparation gradient p09, and the third spoiler pulse (Spoiler 3) . These modifications are made in order to provide appropriate time-gaps between the gradient fields (corresponding to the rows entitled "static period" in Table 1). A time-gap serves as a gradient-invariant window in which the EEG sampling can be made without artifacts. In this way, gradient artifacts are removed from the portion of the sequence that precedes the readout gradient .

A timing diagram of the stepping stone sequence when the values shown in Table 1 are applied is shown in Fig. 6A, with radio frequency and gradient labeling RF, Gs, Gp, and Gr. In Fig. 6A, the reference numerals are: 601 = fat suppression pulses,

602 = a slice selection pulse,

603 = a spoiler gradient pulse,

604 = a slice selection gradient pulse,

605 = a dephasing and rephasing gradient pulse, 606 = the readout gradient pulses, and

607 = a spoiler gradient pulse.

Fig.6B shows the waveform of an imaging artifact 610 obtained when gradient pulses are applied using the stepping stone sequence shown in Table 1. The imaging artifact 610 is the sum of the differentiation of the gradient pulses on the Gs, Gp, and Gr axes.

In Fig.6B, the EEG sample points 91 recorded at a sampling frequency of 1,000 Hz are illustrated over the imaging artifact 610.

Fig. 6C is an enlarged view of a 30-millisecond portion of the timing diagram shown in Fig. 6A, that precedes the readout gradient. This portion is from the row entitled "Fat Suppression" to the row entitled "p08 ramp up" of the stepping stone sequence shown in Table 1. It is apparent from Figs. 6B and 6C that each time-gap serves as a gradient-invariant window in which the EEG sampling can be made without artifact .

The modification of the timing and duration for the preliminary 30-millisecond portion, however, does not work well for the readout portion because the readout gradient induces an imaging artifact with sharp fluctuations in amplitude within a short period of time. Accordingly, the waveform profile of the readout gradient is modified. A conventional readout gradient has a sinusoidal waveform whereas the readout gradient according to the present invention has a modified sinusoidal waveform with its vertices "truncated" horizontally. This is described with reference to Fig.6D which is an enlarged view of the readout portion of the timing diagram shown in Fig. 6A. Each half cycle of the sine wave is modified to have a bell-shaped profile. The readout ramp down/up segment or period 606a consists of a negative (minus range) ramp down (300 μs) and a positive (plus range) ramp up (300 μs) . The readout flat (null) segment or period 606b has a duration of 400 μs . Then, the readout ramp down/up segment or period 606c having a duration of 600 μs consists of a positive ramp down (300 μs) and a negative ramp up (300 μs) . As apparent from Fig. 6D, the flat segment 606b of the readout gradient waveform corresponds to a null period having a predetermined interval. With the stepping stone sequence, the resulting artifact (which is the differentiation of a bell-shaped gradient waveform) includes flat segments 611 along and on the baseline. The flat segment 611 of the artifact waveform is hereinafter referred to as a "sampling shelf" . These sampling shelves 611 correspond exactly to the flat segments 606b of the readout gradient waveform. Such sampling shelves 611 allow easy EEG sampling at the baseline level. In this way, the null period is added at every peak of the sine wave gradient pulses , which provides a near-baseline gap by differential procedure (Faraday's Law) . The sampling shelf 611 is long enough for EEG to be digitally sampled between resultant artifacts .

It is apparent from Fig. 6B that no sample point 91 occurs on a vertex of the spiky component of an imaging artifact 610. The EEG sampling points 91 always drop into baseline gaps on the differential gradient waveform.

2. System Configuration and Operation

The present inventors undertook thorough studies with numerous pilot models in order to obtain a system to which the stepping stone sequence could be applied. For this purpose, the present inventors made the stepping stone sequence to be procedure driven. Conventional fMRI pulse sequence programs are of two categories : event driven programs and procedure driven programs . Event driven programs return control to the operating system to run another task after each volume scan. On the other hand, procedure driven programs work sequentially from the beginning to the end of the pulse sequence process without returning control to the operating system after each volume scan. The stepping stone sequence according to the present invention is designed to be procedure driven in order to achieve strict synchronization with the electroencephalograph. The present inventors succeeded to develop a unique system for the simultaneous acquisition of EEG and fMRI in order to capture EEG and fMRI data in parallel and continuously. This system is in sharp contrast to conventional systems for interleaved EEG/fMRI .

Next, an embodiment of a combined system for simultaneous acquisition of EEG and fMRI according to the present invention is described with reference to Figs . 7 to 9.

Fig. 7 is a schematic diagram showing the configuration of a system for simultaneous acquisition of EEG and fMRI 25 according to an embodiment of the present invention. The system 25 comprises a functional MRI system 30 for acquiring fMRI data, an electroencephalograph 70, and an EEG event analyzer/display 80.

The functional MRI system 30 includes a control unit 31. The control unit 31 has a sequence control 310 and a communication unit which is not shown. The sequence control 310 has a gradient waveform generator 312, a radio frequency (RF) waveform generator 313, and a functional MRI data acquisition unit 314. The aforementioned sequence program controls the control unit 31 (Fig. 7) to generate the pulse as illustrated in Fig. 5D. The sequence program is stored in a hard disk HD connected to a central processing unit (CPU) 32 through a bus and installed in a memory M connected to the CPU 32 through the bus for use. The sequence program may be stored in a predetermined recording medium such as a CD-ROM CD. In such a case, the sequence program can be read from the recording medium each time used or installed in the hard disk HD . The sequence program can control the control unit 31 as described above by its single unit, or in cooperation with OS installed in the computer.

The control unit 31 is connected to the CPU 32 to perform a desired control sequence. The CPU 32 is for setting parameters on the control unit 31 in response to an input from a keyboard (not shown) in order to run a pulse sequence. The gradient waveform generator

312 is connected to a magnet coil assembly 35 via a gradient power amplifier 34. The gradient waveform generator 312 generates gradient pulses that are applied to the magnet coil assembly 35 after being amplified by the gradient power amplifier 34. In response to this, the magnet coil assembly 35 applies gradient pulses to a human subject (not shown) who is placed in a static magnetic field.

The functional MRI system 30 also includes a digital RF signal unit 33. The digital RF signal unit 33 has a system controller 330, a receiver 331, a transmission signal generator 332, and a digital synthesizer 333. The digital RF signal unit 33 is connected to a radio frequency (RF) tuner 36 and a radio frequency (RF) power amplifier 37. More specifically, the RF waveform generator 313 is connected to the transmission signal generator 332 and generates RF pulses which are supplied to the transmission signal generator 332. The transmission signal generator 332 is connected to the RF tuner 36 via the RF power amplifier 37. In response to an RF pulse from the RF waveform generator 313, the transmission signal generator 332 generates a transmission signal and supplies it to the RF power amplifier 37. The transmission signal is amplified by the RF power amplifier 37 and is applied to the RF tuner 36 as an RF signal. In response to reception of the RF signal, the RF tuner 36 produces RF pulses having a single high frequency and applies them to the subject. When hydrogen nuclei in the body of the subject are excited by the gradient and RF pulses, the hydrogen nuclei emit signals.

These signals are received by a receiver coil (not shown) contained in the RF tuner 36. The received signal is supplied to the receiver 331 where it is amplified and subjected to analog-to-digital conversion. The resulting digital signal is then supplied to the functional MRI data acquisition unit 314.

While not being illustrated, it is apparent to those skilled in the art that the magnet coil assembly 35 and the RF tuner 36 are separated from the rest of the components of the functional MRI system 30 via filter plates or other similar components. Since the most common RF frequency used for functional MRI systems is 64 MHz , the filter plates serve to remove all 64-MHz noise components from external sources . In this embodiment, the functional MRI system 30 is activated in response to a key input from the keyboard (not shown) . Upon receiving the key input, the control unit 31 activates a system controller 330. The system controller 330 notifies the transmission signal generator 332 and the receiver 331 of the initiation of a pulse sequence (in this embodiment, the stepping stone sequence) . Then, the RF waveform generator 313 supplies a trigger signal S3 to a clock divider (frequency divider) 45 indicating initiation of the pulse sequence. The transmission signal generator 332 is supplied with clock signals from the digital synthesizer 333 in order to provide a high frequency and phase with high accuracy and stability. The digital synthesizer 333 controls the time-related functions of the functional MRI system 30. The digital synthesizer 333 supplies clock signals at a predetermined frequency (e.g., 4 MHz) to the clock divider 45. The clock divider 45 is also connected to a device that measures bioelectric signals, i.e., the electroencephalograph 70 in this embodiment. The electroencephalograph 70 has a preamplifier 71, an EEG synchronization processor 72, and a switch 73. The pre-amplifier 71 is connected to EEG electrodes 74 affixed to the head of a subject placed within the static magnetic fields. For the purpose of simplification, only one EEG electrode 74 is illustrated in Fig. 7 but actually a total of thirty-two EEG electrodes are used in this embodiment . The number of EEG electrodes is not specifically limited and a physician should determine the appropriate number of electrodes to be used depending on the specific clinical application.

The electroencephalograph 70 receives an EEG signal from the EEG electrode 74. The EEG signal is supplied to the pre-amplifier 71. The EEG synchronization processor 72 has a low-pass filter 721, an amplifier 722, an anti-aliasing filter 723, an analog- to-digital (A/D) converter 724, a digital signal processing (DSP) unit 725, an EEG data acquisition unit 726, and a sampling pulse generator 727. The pre-amplifier 71 is connected to the low-pass filter 721 which, in turn, is connected to the amplifier 722. The amplifier 722 is connected to the anti-aliasing filter 723 which, in turn, is connected to the A/D converter 724. The A/D converter 724 is connected to the DSP unit 725 which, in turn, is connected to the EEG data acquisition unit 726. The EEG data acquisition unit 726 is connected to the EEG event analyzer/display 80.

The pre-ampli ier 71 pre-amplifies a weak signal from the body of a subject, such as the EEG signal. The pre-amplified EEG signal is supplied from the pre-amplifier 71 to the low-pass filter 721. The combination of the pre-amplifier 71 and the low-pass filter 721 improves the signal-to-noise (S/N) ratio and removes unnecessary high-frequency components from the signal. The output of the low-pass filter 721 is supplied to the amplifier 722 where it is further amplified into an amplified EEG signal at a signal level suitable for subsequent A/D conversion. The amplified EEG signal is then supplied to the anti-aliasing filter 723. The anti-aliasing filter 723 is a kind of low-pass filter that attenuates signal components which are greater than the "fold frequency" (e.g., sampling frequency/2) , thus preventing them from being aliased upon A/D conversion. The output of the anti-aliasing filter 723 is low-pass filtered by a factor of five in this embodiment and is supplied to the A/D converter 724. The A/D converter 724 converts the amplified EEG signal into a digital EEG signal. The digital EEG signal is then supplied to the DSP unit 725. The A/D converter 724 is selectively connected to the sampling pulse generator 727 and the clock divider 45 via the switch 73. The sampling pulse generator 727 is for generating sampling pulses which are used for conventional EEG recording. However, the system 25 according to the present invention does not use the sampling pulse generator 727. The present invention uses the single clock divider 45 to produce sampling pulses SP (10,000 Hz) . The sampling pulse SP is used as an external clock to the electroencephalograph 70. The electroencephalograph 70 down samples 10,000 Hz to 1,000 Hz which achieves and ensures EEG sampling at every 1 millisecond. For operation of the clock divider 45, the switch 73 disconnects the connection to the sampling pulse generator 727. Instead, the switch 73 is connected to the clock divider 45 to allow EEG signals to be recorded in synchronization with the functional MRI system according to the sampling pulses . Even when the nominal frequency of the sampling pulse generator 727 exactly equals the clock frequency of the functional MRI system, it is known that the true frequencies may differ slightly from device to device. A very slight timing shift could become greatly divergent as time passes , and may cause a break in synchronization between the functional

MRI system 30 and the electroencephalograph 70. Therefore, single clock driving is used for the functional MRI system 30 and the electroencephalograph 70 by driving the electroencephalograph 70 with clock signals from the f nctional MRI system 30. This arrangement assures reliable EEG sampling, which will be described further below.

The functional MRI system 30 and the electroencephalograph 70 are connected to the EEG event analyzer/display 80. The EEG event analyzer/display 80 has a data recording/analysis unit 81. The data recording/analysis unit 81 is connected to the functional MRI data acquisition unit 314 and the EEG data acquisition unit 726 to receive fMRI data and EEG data, respectively. The data recording/analysis unit 81 processes the fMRI data and EEG data for display on a display unit 82. The data recording/analysis unit 81 records fMRI data and EEG data individually and analyzes the data comprehensively. These data and analyses are displayed on the display unit 82 as an instantaneous EEG event mapped over an fMRI slice image.

Typical MRI systems and electroencephalographs each have their own internal clock (quartz oscillator) to achieve desired timing control. It is impossible, however, to maintain synchronization between these clocks . Synchronization of the stepping stone sequence and the EEG sampling at the scale of microseconds can be achieved only by the above-mentioned single clock driving scheme. More specifically, the switch 73 selects the output of the clock divider 45 in order to supply it to the

A/D converter 724 in the EEG synchronization processor 72 when simultaneous acquisition of EEG and fMRI is required. The output of the clock divider 45 (sampling pulses SP) has a frequency of

10,000 Hz. In response to reception of the output of the clock divider 45, the electroencephalograph 70 down samples 10,000 Hz to 1,000 Hz which achieves and ensures EEG sampling at every 1 millisecond. Details of this down sampling are described below. In this embodiment, a SynAmps (a digital amplifier available from Neuroscan Lab., Sterling, VA, USA), controlled by the SCAN4.2.1 program installed in a PC (Pentium III 600 MHz) , is used as the electroencephalograph 70. Amplification and A/D conversion allow the SynAmps to perform simultaneous phase-locked sampling via multiple individual sample-and-holds. In addition, the high frequency of the internal clock (e.g., 4 MHz) of the functional MRI system 30 can be modulated to acceptable frequencies for driving the electroencephalograph 70, by using a CD-2 clock divider (available from Physio-Tech Co., Ltd., Tokyo, Japan) as the clock divider 45. The CD-2 clock divider down samples the original frequency to provide several clock frequency options (e.g. , 1,000 Hz, 2,000 Hz, 10,000 Hz, and 20,000 Hz) to drive the SynAmps. In this embodiment, the clock divider down samples the 4-MHz frequency of the internal clock to 10 , 000 Hz . The 10, 000-Hz clock is supplied to the SynAmps. In response to reception of the clock, the SynAmps provides some sampling frequency options . The present inventors selected the 1,000 Hz option using the SCAN4.2.1 program on a PC (not shown) connected to the SynAmps. It should be noted that, in order to operate SynAmps with external clock signals , several steps must be followed. Details of the steps are available from the Neuroscan's website.

It should be noted that the present invention can be applied to a combined system using a digital amplifier different from SynAmps or a similar device so long as the candidate device has an input port through which external clock signals can be received and used as a clock for the device. Likewise, the clock divider

45 is not limited to using the CD-2 from Physio-Tech. A similar clock divider can be used. The functional MRI system 30 used in this embodiment is a Siemens Vision Plus MRI scanner but the present invention can be applied to other MRI scanners or MRI systems .

EEG sampling at the baseline level cannot be absolutely guaranteed even when the stepping stone sequence and the EEG sampling are synchronized exactly by the single common clock divider 45. To ensure EEG sampling at the baseline level, it is necessary to accurately align the starting point of the fMRI scanning and EEG sampling. In this respect, the combined system according to this embodiment uses a single operation initiated by an operator to achieve the coincident initiation of MRI acquisition and EEG recording. This may be achieved either mechanically or through software.

3. Clock Divider

Next, the configuration and operation of the clock divider 45 are described with reference to Figs. 7 to 9. Fig. 8 is a schematic block diagram showing essential components of the clock divider 45. Fig.9 is a timing diagram for signals in the components of the clock divider 45. The clock divider 45 in this embodiment receives three kinds of input signals (a manual reset signal SI, MRI clock signals S2, and a trigger signal S3) and produces sampling pulses SP . The following exemplified description is for the case where the MRI clock signals have a frequency of 4 MHz , and the sampling pulses SP, having a pulse width of 20 μs are produced at a frequency of 10,000 Hz.

The clock divider 45 has a clock counter 451, a delayed time counter 452 , a D flip-flops 453 , an RS flip-flop 454 , and OR circuits 455 to 458. The digital synthesizer 333 (Fig. 7) is connected to the clock counter 451 and the delayed time counter 452. MRI clock signals (4 MHz) S2 start being supplied from the digital synthesizer 333 to the clock counter 451 and to the delayed time counter 452 when the functional MRI system is energized (regardless of whether MRI images are being acquired) . The clock counter 451 is a counter used to determine the timing and duration (frequency) of the sampling pulses SP. The clock counter 451 counts the number of MRI clock signals S2. In this embodiment, the clock counter 451 has an "80" terminal and a "400" terminal. The "80" and "400" terminals dispatch signal packets of "80" cycles and "400" cycles, respectively. The signal of the "80" cycle mode rises when the clock counter 451 counts up eighty MRI clock signals S2 from a given timing. The signal of the "400" cycle mode rises when the clock counter 451 counts up four hundred MRI clock signals S2 from the same given timing. When the RESET of the clock counter 451 is kept high, the count is held at zero and no counting is performed. The delayed time counter 452 is a counter that is used to synchronize the initiation of counting of the clock counter 451 with initiation of the stepping stone sequence a ter the lapse of a certain period of time from the rising of the trigger signal S3, when the trigger signal S3 is generated by the functional MRI system 30. The delayed time counter 452 counts the MRI clock signals S2.

The delayed time counter 452 has a T-T terminal. The signal supplied from the T-T terminal rises when the delayed time counter

452 counts N plus 1 provided that a predetermined value N is set to this counter. The delayed time counter 452 also has a CCR terminal. The signal supplied from the CCR terminal falls at the same time that the T-T signal rises. The T-T signal falls at the next count while the CCR signal is kept low until a CCR reset signal rises. When the RESET of the delayed time counter 452 is kept high, the count is kept at zero and no counting is performed.

The D flip-flop 453 is a D flip-flop circuit connected to the RESET of the delayed time counter 452. The D flip-flop 453 is set at the rising of the trigger signal S3. This releases the "Reset" of the delayed time counter 452 allowing it to being counting the MRI clock signals S2. The D flip-flop 453 is reset in response to an output from the OR circuit 455, which resets the delayed time counter 452, preventing it from counting the MRI clock signals S2.

The RS flip-flop 454 is an RS flip-flop for generating the sampling pulses SP. The RS flip-flop 454 is set in response to an output of the OR circuit 458 and is reset in response to an output from the OR circuit 457.

The OR circuit 455 is used to reset the D flip-flop 453. The output of the OR circuit 455 rises when the manual reset signal Si rises or when the T-T signal from the delayed time counter 452 rises. The rise of the output of the OR circuit 455 resets the D flip-flop 453.

The OR circuit 456 is used to reset the clock counter 451. The output of the OR circuit 456 rises when the CCR signal from the delayed time counter 452 rises or when the "400" signal from the clock counter 451 rises. The rise of the output of the OR circuit 456 resets the clock counter 451.

The OR circuit 457 is used to reset the RS flip-flop 454.

The output of the OR circuit 457 rises when the manual reset signal SI rises or when the "80" signal from the clock counter 451 rises.

The rise of the output of the OR circuit 457 resets the RS flip-flop

454. Consequently, the sampling pulse SP rises.

The OR circuit 458 is used to set the RS flip-flop 454. The output of the OR circuit 458 rises when the T-T signal from the delayed time counter 452 rises or the "400" signal from the clock counter 451 rises. The rise of the output of the OR circuit 458 sets the RS flip-flop 454. Consequently, the sampling pulse SP falls .

Next , the flow of signals in the clock divider 45 is described. When an operator performs a predetermined single start operation with a software program, the functional MRI system 30 and the electroencephalograph 70 start their operations at the same time. In response, the manual reset signal SI is supplied to the D flip-flop 453 and the RS flip-flop 454. The manual reset signal SI resets the D flip-flop 453 and resets the RS flip-flop 454. When the stepping stone sequence is initiated, the RF waveform generator 313 supplies a trigger signal S3 to the D flip-flop 453. Then, the rising edge of the trigger signal S3 sets the D flip-flop 453 and releases the RESET of the delayed time counter 452. Release of the RESET causes the delayed time counter 452 to start counting the MRI clock signals S2. This corresponds to the timing tl in Fig. 9.

The delayed time counter 452 has a predetermined delay time N. When the delayed time counter 452 counts N number of MRI clock signals S2, the following MRI clock signal - i.e., the (N + l)-th

MRI clock signal - causes the T-T signal to rise at timing t2 in

Fig.9. This signal sets the RS flip-flop 454 through the OR circuit

458 at the same timing t2. In turn, the sampling pulse SP falls. Also, the CCR signal falls, and this causes the RESET of the clock counter 451 to fall through the OR circuit 456. Therefore, the clock counter 451 is allowed to begin counting the MRI clock signals

S2. The duration between the timing tl and the timing t2 corresponds to a delayed time T<j_eι_ay. In addition, the rising of the T-T signal resets the D flip-flop 453 through the OR circuit 455. Consequently, the delayed time counter 452 is reset and stops the counting of MRI clock signals S2. The CCR signal is kept low until the delayed time counter 452 is reset by the manual reset signal SI. When the clock counter 451 counts up a predetermined number (eighty in this embodiment) of MRI clock signals S2, it produces a pulse to the "80" terminal at the timing t3 in Fig.9. In response to this, the "80" signal rises and resets the RS flip-flop 454 through the OR circuit 457. This causes the sampling pulse SP to rise. In other words , the sampling pulse SP is kept low for a period of 20 μs . In this embodiment, one sampling pulse SP has a pulse width of 20 μs and the MRI clock signal S2 has a frequency of 4 MHz (corresponding to a duration of 0.25 μs) . Therefore, the clock counter 451 counts up eighty MRI clock signals S2 (20 μs/0.25 μs) from the timing t2. After counting the eighty MRI clock signals S2 , the clock counter 451 continues to count up the MRI clock signals S2.

When the clock counter 451 counts up a predetermined number of counts (400 in this embodiment) , it produces a pulse to the "400" terminal at timing t4 in Fig. 9. In response, the "400" signal rises and sets the RS flip-flop 454 through the OR circuit 458.

This causes the sampling pulse SP to fall. At the same time, the

"400" signal is supplied to the clock counter 451 through the OR circuit 456. In response to the "400" signal, the clock counter

451 resets to zero. This causes the "400" signal to fall immediately. The clock counter 451 again starts counting from zero.

In this embodiment, the sampling pulse SP has a frequency of 10,000

Hz (corresponding to a duration of 100 μs) and the MRI clock signal S2 has a frequency of 4 MHz (corresponding to a duration of 0.25 μs) . Therefore, the clock counter 451 counts up four hundred MRI clock signals S2 (100 μs/0.25 μs) from timing t2.

The duration between timing t2 and timing t4 corresponds to a sampling time T_sampii_ng. The pattern of operations during the sampling time T_sampi-_Lng can be repeated for subsequent operations .

As described above, the RS flip-flop 454 is reset through the OR circuit 457 in response to reception of the manual reset signal SI. This causes the sampling pulse to rise. The manual reset signal SI also resets the CCR terminal of the delayed time counter 452 and causes the signal supplied from it to rise. Then, the clock counter 451 is reset through the OR circuit 456 to stop counting and to await reception of a new trigger signal S3.

It should be noted that the pulse width and duration (frequency) of the sampling pulse SP can be varied by means of changing the number of counts (i.e. , 80 and 400) processes by the clock counter 451.

In the above-mentioned embodiment, the clock divider 45 divides the MRI clock signals having a frequency of 4 MHz into a frequency of 10,000 Hz. The divided frequency, however, is not limited to this value. When a different combination of a divided frequency and a sampling frequency is used, the count-up value of the clock counter 451 is varied.

Fig. 10 is a schematic diagram showing the configuration of a combined system 25A for simultaneous acquisition of EEG/fMRI according to a second embodiment of the present invention. The system 25A in Fig. 10 is similar to the one shown in Fig. 7 except that the clock divider 45 is contained in an electroencephalograph

70A. Otherwise the configuration and operation of each component is similar to the first embodiment. Therefore, a redundant description of these components will be omitted.

Fig. 11 is a schematic diagram showing the configuration of a combined system 25B for simultaneous acquisition of EEG/fMRI according to a third embodiment of the present invention. The system 25B in Fig. 11 is similar to the one shown in Fig. 7 except that the clock divider 45 is contained in a functional MRI system 30A. Otherwise the configuration and operation of each component is similar to the first embodiment. Therefore, a redundant description of these components will be omitted.

4. Clinical Examples

A clinical test was made on a human subject using a combined system for simultaneous acquisition of EEG/fMRI according to one of the embodiments described above. With stepping stone sampling, the result is as shown in Fig. 12. The term "stepping stone sampling" refers to a sampling method that involves (1) the stepping stone sequence according to the present invention, (2) one clock driving (of an MRI system and a device measuring bioelectric signals such as an electroencephalograph), and (3) start alignment (alignment of start timing between the sequence and sampling) . In

Fig. 12, white circles indicate EEG sample points 92. As apparent from the figure, EEG signals couldbe sampled at a sampling frequency of 1,000 Hz because the imaging artifact consistently returned to the baseline at 1-millisecond intervals. EEG signals may be sampled at regular intervals, e.g., at every other white circle or every third white circle, by adjusting the clock divider 45.

Fig. 13 shows a comparison between artifact waveforms obtained with a conventional pulse sequence and the stepping stone sequence using a combined system for simultaneous acquisition of EEG/fMRI according to one of the embodiments described above. The averaged peak-to-peak amplitude of an imaging artifact 550 is reduced by 1/5 to 1/10 with stepping stone sampling in comparison to the amplitude of an imaging artifact 560 from a conventional sampling method.

As to residual imaging artifacts having a low amplitude, they were averaged and subtracted from an average-artifact template. This provided the EEG record as shown in Fig. 14. This EEG post processing may be implemented on Matlab Tool (Mathworks, Inc. , MA, USA). For example, data sets from a 100-millisecond span of data prior to the first point of a "one volume scan", the entire data of a "one volume scan", and 300 milliseconds of data following a "one volume scan" , can be grouped together and averaged across all volumes to make an average-artifact template. Then, a 50- millisecond span of data prior to the starting point may be used to determine the baseline level. The template may eventually be subtracted from every artifact on data for each channel basis . Following the subtraction, the data of all channels may be filtered out by a low-pass filter. 5. Sampling Theorem

The sampling theorem is a fundamental law of conventional digital sampling. The sampling theorem states that a sampling frequency must be greater than twice the highest frequency of the input signal in order for the input signal to be sampled perfectly without aliasing. If the sampling f equency is less than mentioned above, the frequencies of the input signal that are above half the sampling frequency will be "aliased". Hence, to ensure that no frequency components greater than half the sampling frequency remain, an analog low-pass filter, called an "anti-aliasing filter" , is typically used before sampling.

The electroencephalograph according to an embodiment of the present invention has an anti-aliasing filter (see; Figs. 7, 10, and 11). However, because stepping stone sampling violates the sampling theorem, the acquired data will always have aliasing contamination wherein "aliasing contamination" is defined by the data components in which frequencies are higher than the Nyquist frequency. With typical low-pass filtering, filtered artifact waveform has a smaller amplitude and a lower frequency than the waveform before filtering. The lower frequency means a longer duration for each half-cycle of the artifact waveform. An extended duration, such as this, is a significant problem for stepping stone sampling because the absolute value of the amplitude remains high throughout its duration, inhibiting stepping stone sampling at the baseline level. In other words, stepping stone sampling requires the duration of each half-cycle to be as short as possible. To satisfy this requirement, the cut-off frequency of the low-pass filter should not be extremely low. In stepping stone sampling. therefore, the cut-off frequency is intentionally rather high.

Conventional electroencephalographs typically use two low-pass filters , an arrangement which applies also to the embodiments of the present invention described with reference to Figs .7 , 10 , and 11. Of these conventional electroencephalographs , the one (SynAmps, available from Neuroscan Lab. , Sterling, VA, USA) used as the electroencephalograph (70, 70A) for implementation of the present invention has the unique characteristic of being able to receive clock signals from outside as external clock signals. This electroencephalograph can handle two different frequencies : the frequency of the external clock signal (in the above embodiments , the frequency of the clock signal supplied from the clock divider 45); and the frequency used for EEG sampling (in the above embodiments, the sampling frequency at which the EEG data acquisition unit 726 performs digital sampling) . The electroencephalograph offers options for clock frequencies . Choosing one of the clock frequency options automatically determines a corresponding sampling frequency. The embodiments of the present invention use an external clock (sampling pulse SP) having a frequency of 10,000 Hz, which is down sampled, then, to a sampling frequency of 1,000 Hz.

Now, the relationship between the low-pass filter 721 and the anti-aliasing filter 723 becomes a subject of discussion. The cut-off frequency of the low-pass filter 721 is fixed at 3,500 Hz in the embodiments. The anti-aliasing filter 723 has a cut-off frequency which is one-fifth (1/5) of the frequency of the external clock, if any. For example, when the external clock has a frequency of 5,000 Hz, and the low-pass filter 721 has a cut-off frequency of 3,500 Hz, then the anti-aliasing filter 723 has a cut-off frequency of 1,000 Hz (5,000 * 1/5). In this case, the cut-off frequency of the anti-aliasing filter 723 is lower than that of the low-pass filter 721. Therefore, the anti-aliasing filter 723 becomes the "valid" low-pass filter, and the measurable range for EEG signals will be up to 1,000 Hz, the cut-off frequency being

1 , 000 Hz . On the other hand, when the external clock has a frequency of 20,000 Hz, and the low-pass filter 721 has a cut-off frequency of 3,500 Hz, then the anti-aliasing filter 723 has a cut-off frequency of 4,000 Hz (20,000 * 1/5). The cut-off frequency of the low-pass filter 721 is lower than that of the anti-aliasing filter 723. Therefore, in this case, the low-pass filter 721 is the "valid" low-pass filter, and the measurable range for the EEG signals is up to 3,500 Hz as determined by the cut-off frequency of 3 , 500 Hz . In this way, the choice of the external clock frequency determines whether the low-pass filter 721 or the anti-aliasing filter 723 will be used as the low-pass filter for subsequent digital sampling. In stepping stone sampling, the external clock has a frequency of 10 , 000 Hz , so the anti-aliasing filter 723 has a cut-off frequency of 2,000 Hz. Accordingly, the anti-aliasing filter 723, with a cut-off frequency of 2,000 Hz, becomes the valid low-pass filter for EEG sampling. As apparent from the above, the combined system, according to the embodiments of the present invention, operates with a cut-off frequency of 2,000 Hz and a sampling frequency of 1,000 Hz. In such cases, with a sampling frequency of 1,000 Hz, the frequency components above 500 Hz are aliased.

Next, the question arises as to which cut-off frequency is suitable for stepping stone sampling. The present inventors made various examinations and concluded that the cut-off frequency which is higher than the highest frequency of the artifact works best . The artifact having the highest frequency appears during the readout period, so the present inventors concluded that the cut-off frequency of the low-pass filter (either the low-pass filter 721 or the anti-aliasing filter 723 ) should be higher than the frequency of the artifact during the readout period (or the original readout gradient) . The frequency of the artifact during the readout period has a close relationship with the sampling frequency used in stepping stone sampling. In stepping stone sampling, since the sampling is performed at every half cycle of the artifact waveform during the readout period, the following relationship is given: (Frequency of Artifact during Readout Period) = (1/2 * Sampling Frequency) .

One half of the sampling frequency is equal to the Nyquist frequency, so (Cut-off Frequency of Low-Pass Filter) > (1/2 * Sampling Frequency) = (Nyquist Frequency) .

Therefore, stepping stone sampling inevitably has aliasing. To estimate the aliasing contamination, human EEG data in a magnet with and without fMRI acquisition at a sampling frequency of 20,000 Hz with a high cut-off frequency of 3,500 Hz determined by a low-pass filter in the SynAmps were recorded. The data collected during fMRI acquisition were artifact-corrected with the above-mentioned method for averaged artifact subtraction. In Figs. 15A and 15B, the power spectrum density of the data is shown. Both of the EEG data sets had 105 (μV)² of EEG signal peaks at around 10 Hz and less than 0.6 (μV)² of signal peak components up to 10,000 Hz except the one at 600 Hz. (See the magnified figures on the interior graphs of each figure) . The range from 1,000 Hz to 10,000 Hz is not shown in these figures. The underlying 600 Hz component was clearly visible and could cause substantial aliasing contamination in the data. However, signal components that are above about 100 Hz are unnecessary because no EEG signals ever approach such a high frequency as characterized by these artifacts. The aliasing by signal components at alias frequencies, including the outstanding

600 Hz component, can emerge at the frequency (1,000 minus alias frequency) Hz, such as 400 Hz. Actually, these frequencies are beyond the interest of the present inventors and thus can be eliminated by a standard low-pass filtering procedure. On the other hand, signals with a frequency ranging from 0 to 100 Hz as a result of aliasing are significantly few and can be considered negligible.

Next, the actual alias contamination was estimated by EEG inspection using the above-mentioned data obtained in a magnet at a sampling frequency of 20,000 Hz, from which two sets of estimation data were derived using different procedures . Alias-clean EEG was first low-pass filtered with a cut-off frequency of 80 Hz and then down sampled at a sampling frequency of 1,000 Hz. In contrast, alias-contaminated EEG was first down sampled to 1,000 Hz and then low-pass filtered with 80 Hz, a procedure which should leave alias contamination. The alias-clean data and the alias-contaminated EEG data are shown in parallel in Fig. 16. As can be seen, no apparent difference exists between these data sets. Thus, it can be concluded that aliasing contamination does not substantially interfere with EEG observation during fMRI acquisition using the stepping stone sampling.

6. Validation of the Stepping Stone Sequence

To verify that the stepping stone sequence is sufficiently sensitive for BOLD contrast , the present inventors made clinical experiments on a human subject using a visual checkerboard stimulation paradigm. More specifically, the present inventors compared functional MR images obtained with a conventional blip type pulse sequence and those obtained with the stepping stone sequence in order to determine whether the activation in the bilateral occipital areas can be extracted. In the clinical experiments, checkerboard visual stimulation was used for the task blocks, and hair-line cross fixationwas used for the control blocks . Stimuli were displayed on screen from an LCD projector, introduced into an MRI gantry, and presented to the subject via a mirror mounted on the head coil . Fig.17A shows the results obtained with a typical blip type pulse sequence while Fig. 17B shows the results obtained with the stepping stone sequence. Figs. 17A and 17B demonstrate that the conventional pulse sequence and the stepping stone sequence exhibited almost the same activation in the bilateral occipital areas with similar distributions. It is not unusual, even on the same human subject, that different areas are activated by the effect of alertness or in consequence of acclimation from successive experiments . Under the circumstances, such coincident results imply that the present invention can be applied in a similar manner to conventional typical functional MRI pulse sequences .

The combined system for simultaneous acquisition of EEG and fMRI according to an embodiment of the present invention enables mapping of spontaneous EEG phenomena onto MRI slice images from the EEG data obtained. Each EEG phenomenon, however, is an accumulation or collection of one-dimensional waveforms and in order to map the one-dimensional phenomenon onto MRI slice images, general linear model (GLM) analysis with SPM99 (Wellcome Department of Cognitive Neurology, London, UK) is required. First, EEG sampling and fMRI acquisition are initiated at the same time. An event of interest could appear at any instant. The examination of EEG after subjection to post-processing to remove imaging artifacts reveals the events of interest and the time instants when those events occurred.

Next, a model function is defined for modeling the time- course of the BOLD responses; and statistical analysis is performed for each voxel of an image between the BOLD responses and the model function. Through this procedure, brain regions relating to specific events can be identified; and spontaneous EEG phenomena can be investigated, thereby, through mapping onto an MRI slice image. By using such a mapping technique for spontaneous EEG phenomena onto MRI slice images, active states of the brain (as mentioned below, for example) can be evaluated.

Mapping of an Epileptic Focus

Fig.18A shows an MRI slice image on which an epileptic region is mapped using stepping stone sampling, and Fig. 18B shows a conventional magnetoencephalogram with an epileptic region. Conventional methods, such as the magnetoencephalogram, involve recording information only from the surface of the brain and, thus, clinical conditions in the subcortical gray and white matters of the brain cannot be investigated with suchmethods . In other words , because of poor temporal resolution, the abnormalities relating purely to epileptiform spikes cannot be portrayed by the conventional method. In contrast, stepping stone sampling of EEG makes it possible to map epileptic regions onto MRI slice images (white portions depicted by an arrow in Fig. 18A) . It has been confirmed that the results obtained with stepping stone sampling match the results obtained with a conventional MEG (white portions depicted by an arrow in Fig.18B) . As can be seen, epileptic regions can be portrayed on the basis of a magnetic phenomenon and on the basis of hemodynamics . The fact that there is a match between epileptic regions portrayed on the basis of completely different , independent events proves that stepping stone sampling is highly reliable. Hence, the present invention is considered advantageous in that extremely accurate information relating to the focal points of epileptic activity can be obtained prior to a proposed neurosurgery for treating epilepsy.

Application to Tasks Relating to Responses to Target Stimuli

An event related potential was measured through an "oddball paradigm" using a combined system according to an embodiment of the present invention. The oddball paradigm consists of the random presentation of non-target and target stimuli, in which the probability of a target stimulus is much higher than that of a non-target. Upon presentation of a target stimulus, the subject is required to respond by, for example, pressing a button. Figs. 19A and 19B show the characteristic waveform of an averaged event related potential on Cz and Pz electrodes, respectively, obtained during 15% presentation of a target stimulus. Figs. 19C and 19D show the characteristic waveform of an averaged event related potential on Cz and Pz electrodes, respectively, obtained during 30% presentation of a target stimulus. As can be seen from the characteristic waveforms of averaged event related potentials (P300 amplitudes) in Fig. 19A to 19D, P300 amplitude is clearly higher at a 15% target stimulus presentation than at a 30% presentation. It has been confirmed that these results do match generally recognized P300 waveform characteristics. The results also prove that the quality of EEG data in EEG and fMRI acquisition is highly reliable. Using the above-mentioned paradigm, an examination of the activated sites on a functional MR image, corresponding to the aforementioned spontaneous EEG events, was performed. Figs. 20A and 20B show mapping results on MRI slice images during 15% and

30% presentations of the target stimuli. From these figures, activated sites (white portions depicted by the white arrow) can be confirmed in each of the right parietal lobe and the right frontal lobe. What is important here is that the conventional system discusses such event related potentials only in terms of the EEG, but by means of the system of the present invention, such EEG events can be discussed from the viewpoint of having been mapped onto an MRI slice image based on hemodynamics. What is also important in this case is that the activation reflected in the BOLD signal is larger at a 15% target stimulus presentation (Fig. 20A) than at a 30% presentation (Fig.20B) . Furthermore, for both an electrical phenomenon and a BOLD signal, vigorous cell tissue activity is suggested more clearly at a 15% presentation (Fig. 20A) than at a 30% presentation (Fig.20B) . These results are highly suggestive that the magnitude of a given potential correlates directly to the expanse and extent of activation. While the above embodiments have thus been described in conjunction with a combined system comprising a functional MRI system and an electroencephalograph, the stepping stone sampling of the present invention can be applied to a system for simultaneous acquisition of other bioelectric signals and magnetic resonance images. Examples of such bioelectric signals include, but are not limited to, spirographic signals, electrocardiographic (ECG) signals, electrooculographic (EOG) signals, electromyographic

(EMG) signals, galvanic skin response (GSR) signals, electrogastrographic (EGG) signals, and pupillary reflex signals .

As described above, stepping stone sampling allows sampling within the periods in which the signal resides around the baseline and thus is unaffected by the amplifier's dynamic range. In other words, sampling can be made consistently within the amplifier's dynamic range regardless of the amplitude of the original artifacts because the sampling shelves are produced by null periods as described above.

While sampling is made at a sampling frequency of 1,000 Hz in the embodiments described above, other sampling frequencies can also be used, such as 2,000 Hz, 2,500 Hz, 5,000 Hz, 10,000 Hz, and 20,000 Hz, as long as the above-mentioned requirements of the present invention are satisfied. In addition, although the combination of a 300-microsecond ramp up segment, a 400-microsecond flat segment, and a 300-microsecond ramp down segment is used for the readout gradient pulses in the above mentioned embodiments to sample EEG signals within the sampling shelves at a sampling frequency of 1,000 Hz, the ramp up and ramp down segments can be shortened by high-speed gradient capability and higher sampling frequency options in an EEG amplifier. In other words, the faster gradient speed in combination with an appropriate sampling frequency facilitates a shorter readout gradient span leading to a shorter MRI acquisition time without concern for potential image distortion. For example, a 200 μs - 225 μs - 200 μs pattern in combination with a sampling frequency of 1,600 Hz results in a 40 μs readout gradient span. The same applies to other bioelectric signals.

Furthermore, the present invention is not limited to digital sampling performed at every 1 millisecond. The sampling may be made at other constant intervals (ranging between, for example,

0.1 and 10 milliseconds) within certain null segments, e.g. , every other null segment .

As apparent from above, the present invention makes it possible to obtain bioelectric signals of good quality even under the unfavorable conditions of MRI acquisition. Accordingly, a broad practical utilization of both high quality bioelectric signals and concomitant mapping of a part of the human body, especially the brain, through MRI or fMRI, can provide a new understanding of various spontaneous activities of the human body. Although illustrative embodiments of the present invention have been described herein with reference to the accompanying drawings , it is to be understood that the present invention is not limited to those precise embodiments , and that various other changes and modifications may be affected therein by one skilled in the art without departing from the scope or spirit of the invention.

Claims

1. A method performed in a combined system capable of acquiring a magnetic resonance image and recording bioelectric signals at predetermined intervals, the bioelectric signal being an electric signal from the body of a subject, said method comprising: an image acquisition step for acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis , a phase encoding axis, and a readout axis; and a bioelectric signal recording step for recording a bioelectric signal, each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment, the combination of a readout ramp up segment, readout null segment, and readout ramp down segment being repeated a predetermined number of times , the recording during the readout phase being performed within the readout null segment, the recording thereby being performed without imaging artifacts on the bioelectric signal.

2. The method as claimed in Claim 1 , wherein the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal. the time durations of all readout null segments are generally equal, and the time durations of all conjoined readout ramp down/up segments are generally equal, the recording during the readout phase being performed at constant intervals within predetermined readout null segments.

3. The method as claimed in Claim 2 , wherein recording during the readout phase is performed within all readout null segments.

4. The method as claimed in Claim 1 , wherein the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, and the time durations of all readout null segments are generally equal, the recording during the readout phase being performedwithin every other readout null segment .

5. The method as claimed in Claim 1 , wherein the bioelectric signal is selected from the group consisting of an electroencephalographic signal, an electrocardiographic signal, an electrooculographic signal, an electromyographic signal, a spirographic signal, a galvanic skin response signal, an electrogastrographic signal, and a pupillary reflex signal.

6. The method as claimed in Claim 1 , wherein the magnetic resonance image is a functional magnetic resonance image.

7. The method as claimed in Claim 1 , wherein recording is performed at constant intervals ranging between 0.1 milliseconds and 10 milliseconds.

8. The method as claimed in Claim 7 , wherein recording is performed at 1-millisecond intervals.

9. A method performed in a magnetic resonance imaging system capable of acquiring a magnetic resonance image which is used in combination with a predetermined device capable of recording bioelectric signals at predetermined intervals , the bioelectric signal being an electric signal from the body of a subject, said method comprising an image acquisition step for acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis, a phase encoding axis, and a readout axis , each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment, the combination of a readout ramp up segment, readout null segment, and readout ramp down segment being repeated a predetermined number of times , the recording during the readout phase being performed by the predetermined device within the readout null segment , the recording thereby being performed without imaging artifacts on the bioelectric signal.

10. The method as claimed in Claim 9, wherein the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, the time durations of all readout null segments are generally equal, and the time durations of all conjoined readout ramp down/up segments are generally equal, the recording during the readout phase being performed by the predetermined device at constant intervals within predetermined readout null segments .

11. The method as claimed in Claim 10, wherein recording during the readout phase is performed within all readout null segments .

12. The method as claimed in Claim 9, wherein the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, and the time durations of all readout null segments are generally equal, the recording during the readout phase being performedwithin every other readout null segment .

13. The method as claimed in Claim 9, wherein the bioelectric signal is selected from the group consisting of an electroencephalographic signal, an electrocardiographic signal, an electrooculographic signal, an electromyographic signal, a spirographic signal, a galvanic skin response signal, an electrogastrographic signal, and a pupillary reflex signal.

14. The method as claimed in Claim 9, wherein the magnetic resonance image is a functional magnetic resonance image.

15. The method as claimed in Claim 9, wherein recording is performed at constant intervals ranging between 0.1 milliseconds and 10 milliseconds.

16. The method as claimed in Claim 15, wherein recording is performed at 1-millisecond intervals.

17. The method as claimed in Claim 9, wherein the predetermined device is an electroencephalograph.

18. A method performed in a combined system having a magnetic resonance imaging system capable of acquiring a magnetic resonance image and a predetermined device capable of recording bioelectric signals at predetermined intervals , the bioelectric signal being an electric signal from the body of a subject, said method comprising: an image acquisition step for acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis , a phase encoding axis, and a readout axis; a clock generation step for generating clock pulses at a first predetermined frequency; a sampling pulse generation step for generating sampling pulses from the clock pulses by means of dividing the first predetermined frequency into a second predetermined frequency; and a bioelectric signal recording step for recording a bioelectric signal according to the sampling pulses , each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment, the combination of a readout ramp up segment, readout null segment, and readout ramp down segment being repeated a predetermined number of times, the recording during the readout phase being performed within the readout null segment, the recording thereby being performed without imaging artifacts on the bioelectric signal.

19. The method as claimed in Claim 18, wherein the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, the time durations of all readout null segments are generally equal, and the time durations of all conjoined readout ramp down/up segments are generally equal, the recording during the readout phase being performed by the predetermined device at constant intervals within predetermined readout null segments .

20. The method as claimed in Claim 19, wherein recording during the readout phase is performed within all readout null segments.

21. The method as claimed in Claim 18, wherein the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, and the time durations of all readout null segments are generally equal, the recording during the readout phase being performed within every other readout null segment.

22. The method as claimed in Claim 18, wherein the bioelectric signal is selected from the group consisting of an electroencephalographic signal, an electrocardiographic signal, an electrooculographic signal, an electromyographic signal, a spirographic signal, a galvanic skin response signal, an electrogastrographic signal, and a pupillary reflex signal.

23. The method as claimed in Claim 18, wherein the magnetic resonance image is a functional magnetic resonance image.

24. The method as claimed in Claim 18, wherein recording is performed at constant intervals ranging between 0.1 milliseconds and 10 milliseconds.

25. The method as claimed in Claim 24, wherein recording is performed at 1-millisecond intervals.

26. The method as claimed in Claim 18, wherein the predetermined device is an electroencephalograph.

27. A combined system comprising: a magnetic resonance imaging system capable of acquiring a magnetic resonance image, said magnetic resonance imaging system having a magnet coil assembly adapted to apply gradient pulses to a slice selection axis, a phase encoding axis, and a readout axis to acquire a magnetic resonance image; and a predetermined device capable of recording bioelectric signals at predetermined intervals , the bioelectric signal being an electric signal from the body of a subject, each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with readout ramp up segments during which the amplitude of the gradient pulse increases, readout null segments during which the amplitude of the gradient pulse does not change, and readout ramp down segments during which the amplitude of the gradient pulse decreases, each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment, the combination of a readout ramp up segment, readout null segment, and readout ramp down segment being repeated a predetermined number of times, said predetermined device being adapted to record bioelectric signals within the readout null segment, the recording thereby being performed without imaging artifacts on the bioelectric signal.

28. The system as claimed in Claim 27, wherein the magnet coil assembly is adapted to apply gradient pulses to the readout axis, during a readout phase when the magnetic resonance image is being acquired, in such a manner that the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, the time durations of all readout null segments are generally equal, and the time durations of all conjoined readout ramp down/up segments are generally equal, said predetermined device being adapted to record bioelectric signals at constant intervals within predetermined readout null segments .

29. The system as claimed in Claim 28, wherein recording is performed within all readout null segments .

30. The system as claimed in Claim 27, wherein the magnet coil assembly is adapted to apply gradient pulses to the readout axis, during a readout phase when the magnetic resonance image is being acquired, in such a manner that the time durations of all readout ramp up segments are generally equal, the time durations of all readout ramp down segments are generally equal, and the time durations of all readout null segments are generally equal, said predetermined device being adapted to record bioelectric signals within every other readout null segment .

31. The system as claimed in Claim 27, wherein the bioelectric signal is selected from the group consisting of an electroencephalographic signal, an electrocardiographic signal, an electrooculographic signal, an electromyographic signal, a spirographic signal, a galvanic skin response signal, an electrogastrographic signal, and a pupillary reflex signal.

32. The system as claimed in Claim 27, wherein the magnetic resonance image is a functional magnetic resonance image .

33. The system as claimed in Claim 27, wherein recording is performed at constant intervals ranging between 0.1 milliseconds and 10 milliseconds.

34. The system as claimed in Claim 33, wherein recording is performed at 1-millisecond intervals.

35. The system as claimed in Claim 27 , wherein said predetermined device is an electroencephalograph.

36. A pulse sequence program used in a magnetic resonance imaging system capable of acquiring a magnetic resonance image by applying gradient pulses to a slice selection axis, a phase encoding axis, and a readout axis , said program comprising causing the magnetic resonance imaging system to acquire the magnetic resonance image, wherein each gradient pulse applied to the readout axis during a readout phase when the magnetic resonance image is being acquired having a waveform with a readout ramp up segment during which the amplitude of the gradient pulse increases , a readout null segment during which the amplitude of the gradient pulse does not change, and a readout ramp down segment during which the amplitude of the gradient pulse decreases , each readout null segment being sandwiched between a readout ramp up segment and a readout ramp down segment , the combination of a readout ramp up segment , readout null segment, and readout ramp down segment being repeated a predetermined number of times .

37. The pulse sequence program as claimed in Claim 36, wherein the magnetic resonance image is a functional magnetic resonance image .

38. The pulse sequence program as claimed in Claim 36 in the simultaneous acquisition of a bioelectric signal and a magnetic resonance image.

39. The pulse sequence program as claimed in Claim 38, wherein the magnetic resonance image is a functional magnetic resonance image .

40. The pulse sequence program as claimed in Claim 38, wherein the bioelectric signal is selected from the group consisting of an electroencephalographic signal, an electrocardiographic signal, an electrooculographic signal, an electromyographic signal, a spirographic signal, a galvanic skin response signal, an electrogastrographic signal, and a pupillary reflex signal.