A New Software for Spike Analysis
with Signal Averaging Technique. Ā®Japanese


F.Shichijo, H.Oki, T.Soga and K.Matsumoto
Department of Neurological Surgery, School of Medicine, The University of Tokushima, Japan

Introduction

Since the first report of Gibbs and co-workers 3), electroencephalogram (EEG) has been playing an important role in the field of diagnosis of epilepsy, localization of epileptic foci, evaluation of the treatment of epilepsy and so on.  Recently, the investigations of voltage topography during the course of a spike and wave have begun 4)7). Furthermore, the dipole tracing method has been applied to identify the epileptic foci2)5). The method for analyzing the spike discharge with EEG topography is called spike voltage topography (SVT) 2). The spatial distribution of spike voltage has been clearly displayed with this method. Sometimes it is very hard to get a stable SVT because of the instability of background EEG activities with spike discharges. Therefore a new software which is named "averaged spike voltage topography (averaged SVT)" for spike analysis with signal averaging technique has been devised. This software was applied in clinical cases to get a stable SVT without the influence of background EEG activities.
In this paper, the principle and clinical application of averaged SVT is presented and discussed.


Materials and methods

1. Subject

The subjects who had spikes or sharp waves in the ordinary EEG recording were investigated with the method of averaged SVT.

2. Method

1) EEG recording

Sixteen electrodes, placed according to the international 10-20 system (Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, Fz, Pz, T5, T6), were referred to both earlobes. In the case of an active reference electrode with temporal lobe epilepsy, the balanced non-cephalic reference electrode was preferred.

2) A principle of averaged SVT and recording procedure

A new software called averaged SVT has been devised with the signal processor 7T18 (NEC SAN-EI, Japan) using signal BASIC language (Fig. 1). The principles of this method are as follows.

Fig. 1: Schematic diagram illustrating the procedure of averaged spike voltage topography (averaged SVT).

The trigger signals were switched on manually when the spike discharge was detected during the monitoring of the on-line EEG recording or off-line EEG recording from the data recorder (Fig. 2-A). According to this procedure, the EEG recordings around spike discharge (from 3 seconds before trigger signal to 3 seconds after trigger signal) were stored in the memory of 7T18. After the storing of EEG data, all the data which was stored in the memory of 7T18 was visually identified again, one by one. Only the data, which had no artifact and was thought that the origin of spike was derived from the same source, were selected for averaging (Fig. 2-B). The middle cursor line which indicated the trigger points for averaging was shifted at the peak of spike and the spike was registered for averaging (Fig. 2-C). The EEG data which was not suitable for averaging was skipped to next data. According to this method, a stable averaged spike wave was displayed. On the CRT of 7T18, the averaged SVT was displayed not only in the pattern of waves (on the left screen) but also in the dynamic spike voltage topography (on the right screen) (Fig. 2-D).  
The bandpass digital filter was also available to distinguish a selected wave pattern without influence of the unstable background EEG activities. (Clinical case will be presented later.)

Fig. 2 : Representative process for making an averaged spike voltage topography.

A: The EEG recordings around spike discharge (6 seconds) are stored in the memory of 7T18 according to the trigger signals which are switched on manually when the spike discharge is detected during EEG recording.
B: All the data which is thought that the origin of spike is derived from the same source is selected for averaging.
C: The cursor line which indicated the trigger points for averaging is moved at the peak of spike.
D: The averaged data is displayed not only in the pattern of waves but also in the dynamic spike voltage topography.


Results

A couple of representable cases are presented.

Case 1: 20-year-old male with epileptic seizure (Fig. 3)

In the conventional EEG recording, a spike and wave complex was observed in the right frontal area. In averaged SVT, the raw spike pattern was displayed as SVT with a single average (Fig. 3-A). While the spike was distinguished with the averaging technique (8 times) (Fig. 3-B).

Fig. 3: Averaged spike voltage topographies.
Both the raw data (A) and the averaged data (B) are displayed as averaged SVT.

Case 2: 19-year-old lady with syncopal attack (Fig. 4)

In the conventional EEG recording, a very small spike was detected under careful observation of EEG (Fig. 4-A). Small spike discharge was clearly detected by the averaging technique (7 times). Though, the dynamic SVT was not stable because of the instability of the baseline around the spike discharge with wide bandpass digital filter (0.5 - 250 Hz) (Fig. 4-B). The special display with selected bandpass filter (8 - 50 Hz) was also facilitated to distinguish the specific wave form such as spike discharge (Fig. 4-C).

Fig. 4: EEG and spike voltage topographies.
Small spike discharge is clearly detected by the averaging technique.

A: According to the careful observation of EEG recording, very small spikes (arrows) are detected in the EEG.
B: Small spike discharge is clearly detected by the averaging technique. Though the base line is not stable with wide bandpass digital filter (0.5 - 250 Hz).
C: The special display with selected bandpass digital filter (8-50 Hz) is also possible to distinguish the specific wave form such as spike discharge.

Case 3: 8-year-old girl with epileptic seizure (Fig. 5)

In the conventional EEG recording, the independent multiple epileptic foci were supposed to be in certain places (Fig. 5-A). In averaged SVT, more than two types of epileptic foci were detected under visual selection of the data which were stored in the memory of 7T18. The average wave of 12 spikes which were dominant in C4, P4, O2 and T6 (Fig. 5-B) was one of the group of epileptic foci and the average wave of 10 spikes which were dominant in C3 and P3 (Fig. 5-C) was the other group of epileptic foci.

Fig. 5: EEG and averaged spike voltage topographies.
The independent multiple epileptic foci (A) are also detected as different type of averaged SVT (B,C) by this averaging technique under visual selection of the data which are stored in the memory of 7T18.

Case 4: 67-year-old male who had an operation scar for left parietal subcortical hemorrhage (Fig. 6)

This patient was having an epileptic attack such as speech disturbance once a month since the last three months after the time of surgery of the left parietal subcortical hemorrhage. In averaged SVT, the spike discharge was distinguished in T5 (Fig. 6-A). An equivalent dipole of the spike potential was placed in the left posterior temporal region with the neural network method for brain electric source localization 1) (Fig. 6-B). According to the comparison with the magnetic resonance images (Fig. 6-C), the equivalent dipole of the spike potential was estimated to be located around the anterior wall of the operation scar of the left parietal subcortical hemorrhage.

Fig. 6: Averaged spike voltage topography and the dipole tracing for the left posterior temporal spike discharge.

A: A left posterior temporal spike is clearly detected in an averaged SVT.
B: An equivalent dipole of the spike potential is placed in the left posterior temporal region with the neural network method for brain electric source localization.
C: The scar of subcortical hemorrhage is shown in magnetic resonance images (proton weighted images) .


Discussion

1) The significance of averaging method of spike voltage

Separation of the buried evoked potential wave forms from other electrical activity are accomplished by signal averaging technique. In contrast with the evoked potential wave forms, the spike potential may be recognized from the background EEG activities without averaging technique. Though to investigate the spike voltage topography for spatial analysis, it is important to diminish the influence of the background electrical activity and the baseline instability of EEG recording. Then the averaging technique of spike potentials was devised to distinguish the spike forms from other electrical activities. In 1984, authors devised one of the averaging technique for Rolandic spikes, the amplitude of which was higher than the background EEG activities. These spikes were used as a trigger and the averaged SVT was obtained automatically 8). However, there were some conditions which were necessary to be applied to this method. These are as follows --- 1) The amplitude of spikes must be higher enough than background activities, 2) The baseline of EEG activities must be stable and 3) The source of spike discharge should be single.
Now, it becomes hard to make an average automatically with the method of averaged SVT. However, it is easy to make an average of the specific type of the spike potentials by visual selection of the epileptic discharges. Even if the spike potentials were lower than the background activities or were derived from the multiple epileptic foci, it is possible to make an averaged spike potential with this method.
These averaged waves are not a real wave, therefore, the comparison between averaged spike potentials and a raw spike is necessary to confirm a reliability of wave forms. With this new method, a raw SVT will be obtained easily with a single average, then the comparison will be done easily.

2) SVT and dipole tracing method

After the development of revolutionary diagnostic techniques, such as CT, MRI, SPECT, PET and so on; detection of the anatomical and functional abnormalities of the brain has advanced. Development of the recording technique of depth and subdural electrodes and the dipole tracing method with EEG topography or MEG, stimulates the surgical treatment of epilepsy.
In the epileptic surgery, it is essential to know whether epileptogenic source is localized or diffuse, single or multicentric, or correlates an organic lesion or not. Then the clinical importance of the SVT is how the epileptogenic dipole source localization is estimated easily with the data of SVT. Recently various dipole source localization methods have been applied in the clinical cases 2)5). Authors have been studying the feasibility of using a neural network trained by using the Backpropagation algorithm as a dipole estimator 1). Recently, the dipole tracing methods with magnetencephalography (MEG) 6) have been reported, and these are expected to be a new investigative method for the epileptogenic source localization.


Conclusions

The characteristics and merits of this method are as follows;
1) Spike analysis can be done without the influence of background EEG activities.
2) Both the averaged data and the raw data can be compared with each other.
3) Small spike discharge can be detected with the averaging technique of the selected bandpass digital filter technique.
4) The independent multiple epileptic foci are also detected by this technique under visual control.
5) The dynamic topographical changes around spike discharge are easily investigated spatially and serially with the dynamic spike voltage topography.
6) The current dipoles of spike discharge are calculated with the neurocomputer technique.


< Homepage >

access No.=Ā@Ā@(96/11/27-)