DETECTION OF FOCAL EPILEPTIFORM ACTIVITY

Abstract

The invention relates to detection of focal epileptiform activity. In order to accomplish an EEG-based mechanism that enables the detection and localization of focal epileptiform activity, a first plurality of brain wave signals is obtained from a subject and a signal-specific measure indicative of the degree of epileptiform activity is determined for at least some of the first plurality of brain wave signals, thereby to obtain a second plurality of signal-specific measures. An indication of the presence of focal epileptiform activity is provided based on the second plurality of signal-specific measures. This may involve, for example, a comparison of the signal-specific measures. If significant mutual differences are detected in the measures, focal epileptiform activity is detected.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

In the following, the invention and its preferred embodiments are described more closely with reference to the examples shown in FIG. 1 to 6 in the appended drawings, wherein:

FIG. 1 is a flow diagram illustrating an embodiment of the method of the invention;

FIG. 2 illustrates one possible electrode configuration for measuring the brain wave signals from a subject;

FIG. 3 illustrates a further embodiment of the method of the invention;

FIG. 4 a to 4d illustrate,respectively, four EEG channels measured from a patient;

FIG. 4 e illustrates the entropies of the wavelet coefficients calculated over a freguency band of 4 to 8 Hz for each of the EEG signals shown in FIGS. 4a to 4d;

FIG. 4 f illustrates the standard deviation of the entropy values shown in FIG. 4e;

FIG. 5 illustrates one embodiment of the apparatus or system according to the invention; and

FIG. 6 illustrates the operational units of the control unit of FIG. 5 for detecting and localizing focal epileptiform activity.

DETAILED DESCRIPTION OF THE INVENTION

FIG. 1 is a flow diagram illustrating one embodiment of the method of the invention. In the present invention, N (N=2, 3, . . . ) brain wave signals are measured from a subject. As is common in the art, each incoming brain wave signal is sampled and the digitized signal samples are processed as sets of sequential signal samples representing finite time blocks or time windows, commonly termed “epochs”. Here, each brain wave signal is also referred to as a channel, i.e. each brain wave signal is obtained through a corresponding channel.

Based on each brain wave signal, a measure indicative of the degree of epileptiform activity is determined (steps 11₁, . . . , 11_N), whereby N signal-specific values are obtained for the measure in each time window. As mentioned above, the measure here refers to a quantitative measure of epileptiform activity, which changes in a monotonic manner on a continuous scale according to the changes in the epileptiform activity.

Next, some or all of the N signal-specific values obtained during a time window are compared with each other at step 12 to examine whether there are significant differences between the signal-specific values obtained within the time window. If this is the case, the process decides at step 13 that focal epileptiform activity is present. If no significant differences are detected between the signal-specific values, the process decides that focal epileptiform activity is not present.

After step 13, the process provides the user of the system information about the results (step 15). As discussed below, this step may involve provision of diversified information based on which the user may perceive whether focal epileptiform activity is present or not.

In a typical embodiment of the invention, step 13 further includes the localization of the focus of the epileptiform activity when focal epileptiform activity is detected. This is performed based on the electrode positions of the individual signal channels: the electrode position(s) of the signal(s) with the strongest indication(s) of epileptiform activity represent(s) the focal area.

The quantitative measure determined for each brain wave signal in steps 11₁ to 11_N may be derived, for example, as the entropy or as a normalized central moment, such as kurtosis, of the signal components on a desired subband of the brain wave signal. The subband may be derived by filtering the time-domain brain wave signals by a filter bank, for example. As discussed below, one possibility to implement the filter bank is to use a wavelet filter to decompose each brain wave signal into subbands so that at least one of the subbands corresponds to the epileptic waveforms of interest. In this case the entropy and/or kurtosis of the wavelet coefficients of the said at least one subband may serve as the quantitative measure defined in steps 11₁ to 11_N.

The quantitative measures may also be derived, for example, from absolute or relative signal amplitudes detected during each successive time window. In order to improve the specificity of the method, the signal amplitudes may be examined on a certain frequency band. For example, the absolute EEG spike amplitudes or the EEG spike amplitudes relative to the background amplitudes may be determined, which occur on a frequency band on which epileptiform spikes occur. A spike here refers to a sharp transient with duration up to a certain maximum time length, such as 200 ms. Alternatively, the quantitative measure may be determined as the signal power on the said certain frequency band. For this, a Fourier transform may be employed to obtain the signal components on the said frequency band. The said certain frequency band may also be a band on which phasic epileptiform activity occurs. That is, the quantitative measure may also be determined as the absolute or relative signal amplitude or as the signal power on a band on which phasic epileptiform activity occurs.

The operations performed in the comparison step 12 may depend on the number of channels used. In case of only two channels (signals) the difference of the quantitative measures may be determined, whereas a measure of the amount of variability, such as the standard deviation, may be calculated if a greater number of channels is used.

In simplified embodiments of the invention, the system of the invention may only display the channel-specific quantitative measures and/or at least one difference of the said measures, whereby the user of the system may decide on the presence of focal epileptiform activity. In FIG. 1, these embodiments are illustrated by dashed arrows leading to step 15. Thus, in these embodiments the system does not make a decision on the presence of focal epileptiform activity, but the system only provides information based on which the user is able to make the said decision.

The measurement electrodes may be positioned to various positions around the scalp of a patient so that signals are obtained from both hemispheres. However, it is advantageous to employ an electrode configuration comprising one or more electrode pairs, so that the electrodes of each pair are positioned symmetrically onto opposite hemispheres. This is due to the fact that certain artifacts caused by eye movements, electrical activity of the heart and blood pulsation appear similarly in symmetrically positioned electrodes and thus tend to cancel out in the comparison step.

FIG. 2 illustrates one possible electrode configuration for measuring the brain wave channels from a subject. The figure shows a seven-channel circumferential bipolar configuration viewed from above the head 20 of a patient. This kind of electrode configuration is illustrated in the above-referred article of S. L. Bridgers and J. S. Ebersole, and it is advantageous in the sense that all the electrodes may be positioned below the hairline, which enables a quick and easy placement of the electrodes. The number of electrodes used for the N channels may vary, since two or more channels may have a common reference electrode. In the embodiment of FIG. 2, adjacent channels (numbered from 1 to 7) have a common electrode. The dashed curves provided with reference signs CH1 to CH4 relate to the example discussed below in connection with FIGS. 4a to 4f.

FIG. 3 illustrates an embodiment of the invention, in which the signal-specific quantitative measure is the entropy or kurtosis of the wavelet coefficients of a subband of the frequency range on which epileptiform activity may appear.

In the embodiment of FIG. 3, a wavelet transform is utilized to separate two subbands from the frequency band on which epileptiform activity may appear, each subband corresponding to a specific type of epileptiform waveforms. Epileptiform EEG activity may include spiky waveforms. Although the frequency contents of the spikes may reach up to about 70 Hz, epileptiform EEG components are typically below 30 Hz. In the embodiment of FIG. 3, two bands are selected, which may contain epileptiform activity and a wavelet transform is employed to decompose each EEG signal obtained from a subject into the said two subbands (steps 31₁ to 31_N). The corresponding wavelet filter bank is denoted with reference numeral 30.

As a result of each decomposition process, two sets of wavelet coefficients are obtained. Each step 31_j (j=1, . . . , N) thus outputs a first set of wavelet coefficients for the first subband (output A) and a second set of wavelet coefficients for the second subband (output B). The entropy of the respective wavelet coefficients is then determined at each step 32A_i and 32B_i (i=1, . . . , N), where A refers to the first subband and B to the second subband, i.e. in step 32A_i the entropy of the wavelet coefficients of the first subband is determined for brain wave signal i and in step 32B_j the entropy of the wavelet coefficients of the second subband is determined for brain wave signal j.

The standard deviation of the N simultaneous entropy values obtained for the first subband is then calculated at step 33A, whereas the standard deviation of the N simultaneous entropy values obtained for the second subband is calculated at step 33B. Both values of the standard deviation are then examined in separate processes, steps 34A and 34B, in order to detect whether the standard deviation meets at least one predetermined criterion set for the occurrence of focal epileptiform activity. Steps 34A and 34B thus serve as decision-making steps in which a decision is made on the occurrence of focal epileptiform activity specific to the subband in question, and they may include a comparison in which the calculated deviation value is compared with a predetermined threshold value. If the standard deviation exceeds the threshold value, the process decides that focal epileptiform activity is present in the signal data. If this is the case for any of the subbands, the process continues by sorting the entropy values of that subband and selects the signal with the lowest entropy value to represent the damaged brain area (steps 35A and 35B). The lowest value is selected since the entropy decreases during epileptiform activity. Based on the electrode position(s) corresponding to the selected value, the process may then indicate the focus of the epileptiform activity (steps 36A and 36B).

In one embodiment of the invention, the presence or absence of focal epileptiform activity is not indicated in each time window, but the process may make the decision based on the comparisons made in several successive time windows. The decision may be made, for example, based on a predetermined majority rule.

As discussed above, the N quantitative measures obtained for a subband may also be compared to each other pair-wise, and focal epileptiform activity may be detected if the difference (or another like variable) between the quantitative measures obtained from symmetrically positioned electrodes exceeds a predetermined threshold.

A method in which subband-specific entropies of wavelet coefficients are utilized to detect epileptiform activity is disclosed in Applicant's EP Patent Application No. 06110089.7-2305 (not public at the filing date of the present application). As discussed therein, the wavelet transform may be employed to decompose the EEG signal into subbands so that at least one of the subbands corresponds to the epileptic waveforms of interest. Thus, in steps 31₁ to 31_N each EEG signal may be decomposed to one or more subbands of interest.

The subbands that may be employed include subbands on which epileptiform spikes occur, such as 16 to 32 Hz and 32 to 64 Hz, and subbands on which phasic waves (triphasic, diphasic or monophasic) occur, such as 2 to 4 Hz and 4 to 8 Hz. In the embodiment of FIG. 3, the first subband may be, for example, a subband on which phasic waves occur, while the second subband may be a subband on which spikes occur. If subbands related to different types of epileptiform activity are employed simultaneously, the system of the invention thus detects when focal epileptiform activity of a specific type is present, and may thus indicate both the focus and the type of the epileptiform activity detected.

The mother wavelet to be used for the wavelet transform belongs preferably to the Daubechies (db) family or to the Symmlet (symm) family, since these families include wavelets that have a good match for actual epileptiform waveforms. Furthermore, it is advantageous to employ a basis function of a relatively low order, such as two or three, since the low order basis functions of a family represent epileptiform patterns better than the high order basis functions of the same family. This is because the basis functions become smoother and more oscillatory, i.e. less spiky, when the order increases. The specificity of the basis functions to spiky and phasic waveforms thus declines as the order increases.

As also discussed in the above-mentioned EP Patent Application, instead of the entropy of the wavelet coefficients, the kurtosis of the wavelet coefficients may also be used as the quantitative measure, i.e. instead of the entropy of the wavelet coefficients, the kurtosis of the wavelet coefficients may be determined in steps 32A_i and 32B_i. Furthermore, kurtosis, which is a normalized form of the fourth central moment, may also be replaced by a normalized form of a central moment of an order higher than four. If kurtosis is employed as the quantitative measure, the signal with the highest kurtosis value is selected to represent the damaged brain area in steps 35A and 35B (kurtosis increases during epileptiform activity).

The advantage of the use of the wavelet entropy or any of the above central moments is that the said parameters are specific to epileptiform activity whereas the other quantitative measures mentioned above may be sensitive to other EEG features, such as normal EEG slowing, i.e. Delta and Theta activation, or to electromyographic (EMG) activity.

Next, the operation of the invention is illustrated with reference to the real-life examples illustrated in FIGS. 4a to 4f. FIGS. 4a to 4d illustrate, respectively, four EEG channels measured from a patient during epileptiform activity. The electrode arrangement used for the measurement corresponds to the arrangement shown in FIG. 2. The four channels, CH1 to CH4, are denoted by dashed curves in FIG. 2, each curve connecting the electrodes from between which the respective channel is measured. FIG. 4e illustrates the entropies of the wavelet coefficients calculated over a freguency band of 4 to 8 Hz for each of the EEG signals of FIGS. 4a to 4d, i.e. the analysis seeks to detect phasic focal epileptiform activity. The mother wavelet used for the wavelet transform was Daubechies 3. FIG. 4f illustrates the standard deviation of the entropy values shown in FIG. 4e. FIG. 4f also shows a threshold (0.05) used in the comparison step. When focal epileptiform activity is present, the entropy of one of the channels (CH4) drops, which causes the standard deviation of the entropy values to rise above the threshold value. The focal epileptiform activity is in this example a short focal disorder in the area of the right temporal lobe, a so-called periodical lateralized epileptiform discharge (PLED). As can be seen from FIG. 4f, the said seizure lasts about one minute.

FIG. 5 illustrates one embodiment of the system according to the invention, assuming that the brain wave data acquired from a subject 100 is EEG signal data. At least two signals are measured by positioning the corresponding electrodes on different parts of the scalp of the subject. If only two signals are utilized, they are measured from opposite hemispheres.

The EEG signals obtained through the corresponding electrode arrangement are supplied to an amplifier stage 51, which amplifies the signals before they are sampled and converted into digitized format in an A/D converter 52. The digitized signals are supplied to a computer unit 53 which may comprise one or more processors.

The signal path of each EEG channel may also be provided with various pre-processing stages, such as filtering stages, which serve to remove non-idealities from the measured signal or otherwise improve the quality of the signal. In one embodiment of the invention, electrocardiographic (ECG) and/or electro-oculographic (EOG) signal data are measured simultaneously from the patient to remove ECG-based artefacts and/or eye movement artefacts from the EEG signal data obtained from the patient.

The signal processing operations may be implemented by dedicated (EEG channel-specific) processing units or by processing units common to at least two channels. Therefore, FIG. 5 only illustrates basic operations applied to the signals, without taking a position on the actual hardware implementation.

The control unit is provided with a database or memory unit 55 holding the digitized EEG signals. The actual recording of the EEG signals thus occurs in a conventional manner, i.e. a measurement device 50 including the above elements serves as a conventional EEG measurement device that acquires a plurality of EEG signals from the subject. However, if wavelet transforms are utilized in the control unit, the sampling frequency of the device may be set according to the requirements of the transform.

Additionally, the control unit is provided with the above-described algorithms for detecting focal epileptiform activity. As shown in FIG. 6, this embodiment of the control unit may include at least three successive operational entities: a calculation module 61 for deriving a sequence of the quantitative measure for each channel, a comparison module 62 for comparing the simultaneous quantitative measures of at least two channels, and a localization module 63 for localizing the focus of the epileptiform activity. Additionally, the control unit may comprise an artifact suppression module 60, which may receive EOG and ECG signal data in order to remove ECG-based artefacts and/or eye movement artefacts from the brain wave channels. As discussed above, the calculation module may include a filter bank which may carry out a Fourier transform or a wavelet transform, for example. As to the comparison module, it may not be necessary to process the measures of all channels, but the comparison module may indicate presence of focal epileptiform activity immediately upon detection of a significant difference between two measure values within entire set of values.

Although one control unit (processor) may perform the calculations needed, the processing of the brain wave signals may also be distributed among different processors (servers) within a network, such as a hospital LAN (local area network). For example, a conventional measurement device may record the brain wave signals and an external processor may be responsible for the detection of the focal epileptiform activity based on the signals. The term control unit here refers to any system, processor, circuit, or computing entity which is capable of carrying out the above operations based on brain wave signal data, so that either the system or its user may decide whether or not focal epileptiform activity is present.

The control unit may display the results on the screen of a monitor 54 connected to the control unit. This may be carried out in many ways using textual and/or graphical information about the presence of certain waveforms or patterns and their focus. For example, the monitor may display the skull and indicate the area corresponding to the focus for each type of epileptiform activity detected. However, in a simplified embodiment the apparatus may only indicate the channel-specific quantitative measures, in which case a physician may decide on the occurrence of focal epileptiform activity.

The system further includes user interface means 56 through which the user may control the operation of the system.

As discussed above, the brain wave data may also be acquired through a standard MEG recording. The measurement device 50 may thus also serve as a conventional MEG measurement device, although a MEG measuring arrangement is far more expensive than an EEG measuring arrangement.

The software enabling a conventional EEG or MEG measurement device 50 to detect epileptiform waveforms may also be delivered separately to the measurement device, for example on a data carrier, such as a CD or a memory card, or through a telecommunications network. In other words, a conventional EEG or MEG measurement device may be upgraded by a plug-in unit that includes software enabling the measurement device to detect focal epileptiform activity based on the brain wave signals measured by the device from the subject. The software module thus determines, when run, the signal-specific quantitative measures and provides, based on the measures, an indication of the presence of epileptiform activity.

Although the invention was described above with reference to the examples shown in the appended drawings, it is obvious that the invention is not limited to these, but may be modified by those skilled in the art without departing from the scope and spirit of the invention. For example, the quantitative measure may be formed by combining two or more measures.

Claims

1. A method for detecting focal epileptiform activity, the method comprising the steps of: obtaining a first plurality of brain wave signals from a subject;determining a signal-specific measure indicative of the degree of epileptiform activity for at least some of the first plurality of brain wave signals, whereby a second plurality of signal-specific measures are obtained; andproviding an indication of the presence of focal epileptiform activity based on the second plurality of signal-specific measures.

2. A method according to claim 1, wherein the providing step includes a sub-step of comparing at least some of the second plurality of signal-specific measures with each other.

3. A method according to claim 2, further comprising the steps of: positioning a third plurality of electrodes onto predetermined areas of the scalp of the subject to obtain the first plurality of brain wave signals through the third plurality of electrodes, whereby each signal-specific measure is associated with a predetermined area of the scalp;localizing the focus of the epileptiform activity based on the predetermined areas and the second plurality of signal-specific measures.

4. A method according to claim 3, wherein the localizing step includes the sub-steps of: sorting the second plurality of signal-specific measures to obtain a sorted order of the signal-specific measures;selecting one of the outermost signal-specific measures in the sorted order; andfinding the predetermined area associated with the selected signal-specific measure.

5. A method according to claim 1, wherein the determining step includes a sub-step of employing a filter to decompose each brain wave signal into at least one predetermined subband.

6. A method according to claim 5, wherein determining step further includes a sub-step of determining the entropy of each brain wave signal with respect to each subband, whereby at least one entropy value is obtained for each brain wave signal.

7. A method according to claim 5, wherein the sub-step of employing a filter includes employing a wavelet filter, whereby wavelet coefficients corresponding to each of the at least one predetermined subband are obtained for each brain wave signal; andthe determining step further includes a sub-step of determining the entropy of the wavelet coefficients.

8. A method according to claim 1, wherein the first plurality of the brain wave signals are selected from a group including electroencephalogram (EEG) signals and magnetoencephalogram (MEG) signals.

9. A method according to claim 1, wherein the determining step includes the sub-steps of: defining spike amplitudes for the at least some of the first plurality of brain wave signals; anddetermining the second plurality of signal-specific measures based on the spike amplitudes.

10. A method according to claim 2, wherein the comparing sub-step includes calculating a measure indicative of the amount of variability in the second plurality of signal-specific measures.

11. A method according to claim 3, wherein the providing step includes indicating the focus of the epileptiform activity.

12. An apparatus for detecting focal epileptiform activity, the apparatus comprising: measurement means for obtaining a first plurality of brain wave signals from a subject;calculation means for determining a signal-specific measure indicative of the degree of epileptiform activity for at least some of the first plurality of brain wave signals, whereby a second plurality of signal-specific measures are obtained; andindicator means, responsive to the calculation means, for providing an indication of the presence of focal epileptiform activity.

13. An apparatus according to claim 12, wherein the indicator means comprise comparison means for comparing at least some of the second plurality of signal-specific measures with each other.

14. An apparatus according to claim 13, further comprising: a third plurality of electrodes to be positioned onto predetermined areas of the scalp of the subject for obtaining the first plurality of brain wave signals through the third plurality of electrodes, whereby each signal-specific measure is associated with a predetermined area of the scalp;localization means for localizing the focus of the epileptiform activity based on the predetermined areas and the second plurality of signal-specific measures.

15. An apparatus according to claim 14, wherein the localization means comprise: sorting means for sorting the second plurality of signal-specific measures to obtain a sorted order of the signal-specific measures;selection means for selecting one of the outermost signal-specific measures in the sorted order; andassociation means for finding the predetermined area associated with the selected signal-specific measure.

16. An apparatus according to claim 12, wherein the calculation means comprise a filter bank for decomposing the at least some of the first plurality of brain wave signals into at least one predetermined subband.

17. An apparatus according to claim 16, wherein the calculation means are configured to determine the entropy of the at least some of the first plurality of brain wave signals with respect to each subband.

18. An apparatus according to claim 16, wherein the filter bank comprises wavelet filters.

19. An apparatus according to claim 18, wherein the wavelet filters are configured to perform a discrete wavelet transform.

20. An apparatus according to claim 18, wherein the wavelet filters are provided with a basis function from a group of Daubechies wavelets.

21. An apparatus according to claim 18, wherein the wavelet filters are provided with a basis function from a group of Symmlet wavelets.

22. An apparatus according to claim 12, wherein the calculation means are configured to define spike amplitudes for the at least some of the first plurality of brain wave signals and determine the second plurality of signal-specific measures based on the spike amplitudes.

23. An apparatus according to claim 13, wherein the comparison means are configured to calculate a measure indicative of the amount of variability in the second plurality of signal-specific measures.

24. An apparatus according to claim 23, wherein the measure is the standard deviation of the second plurality of signal-specific measures

25. An apparatus for detecting focal epileptiform activity, the apparatus comprising: a measurement module configured to obtain a first plurality of brain wave signals from a subject;a first computing module configured to determine a signal-specific measure indicative of the degree of epileptiform activity for at least some of the first plurality of brain wave signals, whereby a second plurality of signal-specific measures are obtained; andan indicator module configured to provide an indication of the presence of focal epileptiform activity based on the second plurality of signal-specific measures.

26. A computer program product for detecting focal epileptiform activity, the computer program product comprising: a first program code portion configured to determine a signal-specific measure indicative of the degree of epileptiform activity for a plurality of brain wave signals, thereby to obtain a corresponding plurality of signal-specific measures; anda second program code portion configured to provide an indication of the presence of focal epileptiform activity based on the second plurality of signal-specific measures.