Complementarity of different analysis methods

136

ELECTROENCEPHALOGRAPHY AND CLINICAL NEUROPHYS|OLOGY

COMPLEMENTARITY

OF DIFFERENT

ANALYSIS

METHODS ~ W . STORM VAN LEEUWEN

Neurophysiology Department, Institute of Medical Physics, T.N. 0., and EEG Department, N~ arological Clinic, University Hospital, Utrecht (The NetherlandM

In the previous communication Grey Walter demonstrated that, with different forms of analysis of the EEG, different aspects of the signal may be extracted. Thus the notion of complementarity of the different methods of analysis was developed. In the present communication an example of this notion is presented. The example concerns an investigation which has been carried out on the occurrence and spread of alpha rhythms over the head in normal subjects. For this investigation three forms of analysis have been applied. 1. Continuous frequency analysis (Bekkering e t al. 1958; Storm van Leeuwen and Bekkering 1958). 2. Auto- and cross-correlation (Brazier and Casby 1952; Barlow e t al. 1959). 3. Topographic analysis (Petsche and Marko 1955). METHODS

The investigations have been carried out in fifteen subjects from whom alpha rhythms could be recorded, in one of these subjects all three forms of analysis have been carried out. The electrodes were conventional silver-silver chloride discs, stuck on the skin with collodion. The electrode positions in the various investigations are indicated in the diagrams. Sixteen channel EEG instruments constructed by Van Gogh, Grass and Offner, have been used. In all investigations 0.3 sec time constants have been used. The subjects reclined in a comfortable chair or couch in a dimly lit room. Care was taken to prevent the subjects from sleeping during the recording. in the investigations with auto- and cross-correlation analysis a 7-channel Ampex tape recorder has been used. In the other investigations the 1 R e p o r t o n : The analysis a n d display of E E G information.

data were tape recorded with the 8-channel EEG magnetograph (Bekkering et al. 1957, 1958). RESULTS

The results obtained with the continuous frequency analyser have been described by Bekkering e t al. (1957). They may be summarised briefly as follows. In different a eas of the scalp alpha rhythms may be recorded the frequencies fluctuate around a main frequem:y. In some areas these frequency fluctuations occLJr simultaneously, the frequency distribution in I:ime being similar. In other areas the frequencies of the alpha rhythms fluctuate differently, although the dominant frequency may be the same lFig. 1). On the basis of these data, areas whet't, alpha rhythms show an identical frequency dis1 ribution in time have been distinguished from are as where the frequency distribution in time is d~fferent. In some of the subjects three areas on ,:~ach side of the head were found, in others foul areas were distinguishable. It has been possible to carr.,, out auto- and cross-correlation studies of the different alpha rhythm producing areas, thanks ~o the aid of Dr. Brazier and her collaborators (Massachusetts General Hospital, Boston, Mas~.) and Professor Rosenblith and his collabora!ors (Massachusetts Institute of Technology, Ca~nbridge, Mass.). The results have been described briefly by Storm van Leeuwen (1961). They can be summarised as follows. In each of the subjects consi(ierable auto-correlation of the alpha rhythms was observed. This is in agreement with many previous investigations by Brazier and Casby (1952), Brazier and Barlow (1956) and Barlow e t al. (1959). M oreover, marked cross-correlation of alpha rbsthms occurring in various areas was encountered, even in those areas differing in the frequency distribution in Eleetroenceph. elin. Neurophysiol., 1964, 16:136-139

COMPLEMENTARITY OF ANALYSIS METHODS

137

Fig. 1 Continuous frequency analysis of alpha rhythms occurring in eight areas. Leads are indicated in diagram (bipolar linkage). In various places the frequency 10 (c/sec) is marked. Note that the frequency distributions in the areas covered by electrodes 4-9 are similar. The frequency distributions in the areas covered by electrodes 1 3 are also similar but different from the previous.

time, as shown with c o n t i n u o u s frequency analysis. In a series o f b i p o l a r leads 2 c m parallel to the midline a n d in a series o f leads in an o c c i p i t o t e m p o r a l d i r e c t i o n 2 cm a b o v e the ear, g r a d u a l phase shifts are o b s e r v e d (Fig. 2). These phase shifts seem to indicate that, a v e r a g e d over 2½ min, a f r o n t o - o c c i p i t a l " s w e e p " o f a l p h a waves takes place. This finding is in a c c o r d a n c e with o b s e r v a t i o n s b y C o o p e r a n d M u n d y Castle (1960) o b t a i n e d with the use o f a spiral-scan t o p o s c o p e , with o b s e r v a t i o n s by Petsche and M a r k o (1955) with the m u l t i v i b r a t o r t o p o s c o p e , a n d with early o b s e r v a t i o n s b y D a r r o w et al. (1955), o b t a i n e d w i t h o u t a u t o m a t i c analysis. T h e c r o s s - c o r r e l a t i o n d a t a indicate t h a t the a r e a over which the phase shift t a k e s place differs f r o m one subject to a n o t h e r . I n some subjects the phase shift occurs in the o c c i p i t o - p a r i e t a l region, in others in the p o s t e r i o r p a r t o f the occipital area. A phase shift was n o t i c e a b l e in all subjects investigated.

We have been able to carry out t o p o g r a p h i c analysis o f the a l p h a r h y t h m s according to Petsche.

7

Fig. 2 Cross-correlation analysis of alpha rhythms in seven areas. Leads are indicated in diagram (common reference on forehead). Average of a 2½ rain period. Note gradual phase shift, the central areas leading over the occipital areas. Electroenceph. clin. Neurophysiol., 1964, 16 : 136-139

138

W. STORM VAN LEEUWEN

Fig. 3 Topographic analysis (Petsche) of alpha rhythms occurring in sixteen area s (same subject a~ in Fig. 2). Leads indicated in diagram (common reference on forehead). Note gradual phase shift in frontooccipital direction in most, but not in all waves. TABLE I Comparison of the value which the three analysis methods and the primary EEG have for studying ~he different aspects of the alpha rhythms*

Continuous frequency analysis Correlation analysis (Brazier et al.) Topographic analysis (Petsche) Primary EEG

Amplitude

Phase relations

Wave form

Frequency

I~tegrationperiod

poor poor

poor good

poor good

good good

1 sec 2½ min

poor good

good poor

poor good

poor poor

none none

* In the last column the integration period is added. A similar g r a d u a l s h i f t o f phase could be o b s e r v e d for each s e p a r a t e wave (Fig. 3), as was f o u n d with c r o s s - c o r r e l a t i o n a v e r a g e d over a p e r i o d o f 21 min. H o w e v e r , the d i r e c t i o n o f the shift, t h o u g h m a i n l y f r o n t o - o c c i p i t a l (in a g r e e m e n t with the c r o s s - c o r r e l a t i o n d a t a ) is s o m e t i m e s reversed, o c c i p i t o - f r o n t a l , whereas on o t h e r occasions the waves a p p e a r s i m u l t a n e o u s l y in all a r e a s (Fig. 3, 5th wave f r o m the left). The o b s e r v a t i o n s with the three forms o f analysis indicate the f o l l o w i n g :

1. Different areas on the scalI~ m a y be discerned where a l p h a r h y t h m s occuJ with f r e q u e n c y d i s t r i b u t i o n s which are n o t alwavs identical (continuous f r e q u e n c y analysis). 2. These areas nevertheless ~ihow c o n s i d e r a ble c r o s s - c o r r e l a t i o n when averaged over a period o f 2½ min. This suggests ' l o o s e c o u p l i n g " b e t w e e n these areas, as described by W i e n e r (1958) (correlation analysis). 3. " S w e e p i n g " o f a l p h a waves a p p e a r s to occur over the scalp in a m a i n l y f r o n t o - o c c i p i t a l Electroeneeph. clin. NeurophysioL

1964, 16:136-139

COMPLEMENTARITYOF ANALYSISMETHODS direction (correlation analysis), but sometimes in the opposite direction (topographic analysis). 4. The "sweeping" often occurs synchronously and in the same direction on both sides of the head. 5. Some alpha waves appear to occur, without "sweeping", simultaneously in all areas on both sides of the head, where alpha waves can be detected (topographic analysis). These conclusions demonstrate how information which is rejected by one form of analysis is selected by another method. As shown in Table I, one of the differences of the three analysis methods used in this investigation is the value of the integration period. With the correlator the integration period is 2½ min; the cross-correlation data reveal rhythmic phenomena occurring synchronously--possibly with phase differences-averaged over this period. With continuous frequency analysis changes occurring in the order of 1 sec can be detected. (The buildup and die-away characteristics of the filters, although not the same as in auto- and crosscorrelation, can be regarded as an integrating time.) In the topographic analysis no integration is applied; mutual relations of each separate wave occurring in different areas can be studied. The above comparison indicates that the combination of data obtained with various forms of analysis may yield an insight into the phenomena under investigation beyond simple addition or, in other words, that different forms of analysis may be complementary. SUMMARY The spread of alpha waves over the scalp in normal subjects has been investigated with three forms of analysis : 1. continuous frequency analysis; 2. auto- and cross-correlation analysis; 3. topographic analysis. The data obtained with these analyses indicate that different alpha rhythms may be discerned in different areas, that the alpha rhythm-pro-

139

ducing structures are probably loosely coupled, and that alpha waves appear to "sweep", usually in a fronto-occipital direction. The analysis data are shown to be complementary.

REFERENCES BARLOW, J. S., BRAZIER, M. A. B. and ROSEN,qL.ITH,W. A.

The application of autocorrelation analysis to electroencephalography. Proc. 1st nat. Biophysics Conf., 1957. Yale University Press, New H~ven, 1959: 622-626. BEKKERING,D. H., KAMP,A. and STORMVAN L EEUWEN,W. The EEG spectrograph. Electroenceph. olin. Neurophysiol., 1958, 10: 555-559. BEKKERING, D. n . , KUIPER, J. and STORM VAN LEEUWEN,

W. Origin and spread of alpha rhythms. 4cta physiol. pharmacol, neerl., 1957, 6: 632-640.

BRAZXER,M. A. B. and BARLOW,J. S. Some applications of correlation analysis to clinical problems in electroencephalography. Electroenceph. clin. ]~ urophysiol., 1956, 8: 325-331. BRAZIER, M. A. B. and CASBY,J. U. Cross-correlation and auto-correlation studies of electr()encephalographic potentials. Electroenceph. clin. ~europhysiol., 1952, 4:201-211. COOPER,R. and MUNDVCASTLE,A. C. Spalial and temporal characteristics of the alpha rhythm: A toposcopic analysis. Electroenceph. clin. Neuroplr~siol.. 1960, 12: 153-165. DARROW, C. W., WILCOTT, R. C., SIEGEL A. and STROUp M. Instrumental evaluation of anterior-po~terior EEG phase relationships. Electroenceph. clin. Neurophysiol., 1955, 7: 472~73. PETSCHE, H. und MARKO,A. Ueber die Ausbreitung der Makrorhythmen am Gehirn des Menschen und des Kaninchens auf Grund toposkopischer Untersuchungen. Arch. Psychiat. Nervenkr., 1955, 193: 177-198. STORM VAN LEEUWEN, W. Comparison of EEG data obtained with frequency analysis and with correlation methods. In: M. A. B. BRAZIER(Editor). Computer techniques in EEG analysis. Electroenceph. Neurophysiol., 1961, Suppl. 20: 37--40.

clin.

STORM VAN LEEUWEN W, and BEKKERING, I~). n . Some

results obtained with the EEG spectrograph. Electroenceph, clin. Neurophysiol., 1958, 10: 563-570. WIENER, N. Non-linear problems in random theory. John Wiley, New York, 1958, 131 p.

Reference: STORMVAN LEEUWEN, W. Complementarity of different analysis methods. Electroenceph. clin. Neurophysiol., 1964, 16: 136-139.