Prof. Dr. Wilfried Dimpfel
At the end of the nineteenth century (1875) an English medical doctor from Liverpool named Richard Caton (* 1842; † 1926) discovered that the brain of animals produced electricity. His work led German Hans Berger in Jena to discover that also human brains produced potential differences which he was able to measure in 1924. In 1929 he published his first paper on this discovery where he also cited the work of Caton. Since then “encephalography or EEG” as he named this methodology spread all over the world and there is hardly a neurologist who is not familiar with the basic methodology. But the size of these potential changes (1 – 250 mV) is very small and asks for special amplifiers to record them. The methodology of EEG recording has conquered the neurological ambulances and hospitals worldwide and is regarded as a standard diagnostic procedure since then. Nevertheless, interpretation of the electric changes is still a matter of debate. Despite highly sophisticated technical equipment and good artifact-free recordings the physiological meaning of signal changes very often remains unknown. This book aims at the description of current knowledge especially on the interpretation of EEG changes after transformation of the signals from time dependence into frequency dependence by a mathematical procedure called Fast Fourier Transformation (FFT). Already Hans Berger had recognized the usefulness of frequency analysis of his ”Enkephalogramm” , since he published a manuscript together with Dietsch in 1932, when it took him weeks to perform the mathematical calculation without computer aid. Today, FFT is performed by computers online and real time. However, the documentation of the results varies from software to software resulting in more or less easy understandable depictions of the electric brain activity. We have spent extremely much time and money to develop a hardware-software combination in order to present the data in a meaningful and easy interpretable manner. Years of recording of the EEG under a whole variety of physiological and pathological conditions using the newly developed technology “CATEEM” brought new insights into this fascinating activity of the brain which is available non-invasively. Relations of the electric activity to biochemical neurotransmitter activity as well as relations to cognitive and emotional features have been recognized. Also pathological findings can be documented in a more precise way. Finally, quantitative assessment of sleep and anaesthesia has become feasible based on the quantitative description of the EEG using source density analysis.
In order to measure the potential differences produced by the brain one needs at least two different locations. Preferably, one of those two places is electrically silent. For many recordings the nose or ears are used as one pole. The term “ monopolar” has been used for this type of electrode montage which is however not correct. It should be called “referenced” recording. One uses one ear or even linked ears as reference for one or more other electrode positions. Each of the different locations over the skull as recorded against the reference is called a channel. A disadvantage of this kind of recording is the fact that the distances between several locations and the reference electrode differ quite a bit, making comparisons of the channels among each other difficult. An alternative method consists in recording from two “active” electrodes in order to evaluate the difference between them. This has been called “bipolar” recording. Many different “montages” of this kind have been published with the aim of documenting more or less specific pathological features of brain activity. The disadvantage is that it takes an enormous effort to interpret the particular electric patterns or grapho-elements in relation to disease. Tons of publications and books deal with this kind of EEG analysis. A more reliable and standardized way to document electric brain activity consists in using the so-called common averaged reference like based on the electrode position Cz. In this case, firstly all potential differences from all other electrode positions are measured against Cz and then the averaged signal is taken as reference.
Historically, electrode positions are named according to the position at the skull where C stands for central, F for frontal, T for temporal, P for parietal and O for occipital (Fig. 1). Even numbers like C4 stand for the right hemisphere, uneven numbers like C3 for the left hemisphere. A standard distribution of electrode positions over the skull is called the 10/20 system according to 10 and 20% distances used between single electrodes.
Fig. 1 Classic schema of electrode positions according to the 10-20 system.
The recording against the common averaged reference Cz is the only reasonable way to obtain meaningful data for electric mapping of the brain as realized in CATEEM. An example of the recorded signals under this electrode montage is given in Fig. 2 for the recording condition “eyes open”.
Fig. 2 EEG curves recorded against common average reference at Cz under the condition of “eyes open”.
As one can imagine, to read these signals and to interpret either single grapho-elements like so-called k-complexes or steep waves or the whole pattern as such asks for many years of experience and finally remains very subjective. One of the reasons is that the electric pattern very much depends on the recording conditions under which the signals are taken. For example, there is a big difference in the pattern if recorded under the condition of “eyes closed” or “eyes open”.
Fig.3. Curves recorded against common average reference at Cz under the condition of “eyes closed”.
Therefore this has always to be documented before interpretation is possible. Furthermore, if mental work is performed or an exciting video clip is watched has a tremendous effect on the focal electric patterns in different brain regions. Again, during sleep and anaesthesia completely different patterns of changes are seen. Examples are given in Fig. 3-7. It is therefore easy to understand that classic evaluation of the recorded analog signals has come to its limits despite it is still practized throughout the world. A more meaningful approach seems to be to find a way of quantitation of the EEG signal as Hans Berger has suggested together with Dietsch (mentioned above). This is explained within the next section.
Fig.4. Curves recorded against common average reference at Cz under the condition of “memory test”.
In order to quantify the signals a mathematical tool called Fast Fourier Transformation has to be used for transformation from time dependence into frequency dependence. The result of this calculation is called a spectrogram where frequencies are depicted with their respective electric power. This can be done using a particular resolution for example of 0.25 Hz. Most electric power produced by the brain can be seen in the range of 0.1 to 35 Hz. Since it would be very difficult to follow each of the single 0.25 Hz broad peaks and evaluate them consecutively, several of those 0.25 broad frequency ranges are summed up to give what we call a frequency band. This can be done arbitrarily for example by definition of the first band reaching from 1 to 4 Hz, second one from 4 to 8 and so forth. But in order to be able to relate these frequency ranges to physiological features one can also chop the total frequency range of 0 to 35 Hz into several ranges where frequencies seem to cumulate from time to time. After obtaining thousands of EEG spectrograms and analyzing them for cumulating frequencies we decided to define 6 consecutive ranges. They were named according to Greek letters from delta representing the slowest frequencies from 1 to 4.5 Hz, followed by the theta frequency range representing frequencies from 4.75 to 6.75 Hz. Next, alpha1 waves were defined from 7 to 9.5 Hz and alpha2 from 9.75 to 12.5 Hz. The fast frequencies historically named as beta waves were chopped into beta1 from 12.75 to 18.5 Hz and beta2 from 18.75 to 35 Hz. One of the biggest advantages of quantitative assessment of EEG recordings is not only the objective manner of analyzing this kind of data, but the possibility to average those data over time. Thus, it is possible to accumulate data over a time period of for example 5 minutes and determine the median electric power for each of the six frequency ranges. A certain pattern of spectral power arises which now can be compared on a quantitative base among different recording conditions like “eyes open”, “eyes closed”, “mental work”, sleep or in the presence of drugs. For example the “Berger Effect” which describes parietal and okzipital increases of alpha waves can now precisely be quantitated.
Examples are given in Fig. 5-6.
Fig. 5 Quantitative description of electric power in absolute frequency ranges under the recording condition of “eyes open” for 5 minutes. Definition of frequency ranges (delta through beta2) is given in the text. Average electric power of single electrode positions over this time is given on the ordinate.
Fig. 6 Quantitative description of electric power in absolute frequency ranges under the recording condition of “eyes closed” for 5 minutes. Definition of frequency ranges (delta through beta2) is given in the text. Average electric power of single electrode positions over this time is given on the ordinate. Please remark different scaling of the ordinate in comparison to Fig. 5.
The next advantage of describing the EEG quantitatively by FFT is the possibility to construct a data base containing EEG`s from several hundred normal healthy people and use it to discover deviations from them in terms of increased or decreased spectral power at particular electrode positions. Determination of the error probability of this deviation or aberration from normality might be very useful to discriminate a pathologically changed EEG from the physiological condition. Finally, the numeric deviation can also be used for therapy control since a smaller deviation after treatment with a particular drug might indicate a return to normality. This has been shown to be the case for example in Parkinson`s disease after administration of rasagiline instead of selegiline after 4 months treatment. The search for a physiological meaning of these specially defined frequency ranges was conducted in rats, where we followed the action of highly specific drugs acting on particular neurotransmitter receptors. Interfering for example with the cholinergic system of the brain in an agonistic or antagonistic manner resulted mainly in changes of delta waves. But it turned out to be nearly impossible to change only this frequency range alone. However, administering quite a number of different compounds all acting at this cholinergic transmission system brought up enough evidence to relate this frequency range of delta waves to the activity of acetylcholine. After obtaining data from several compounds acting on the presynaptic norepinephrine alpha2 receptor it became clear that theta waves followed the drug action. Since agonists at this presynaptic norepinephrine receptor lead to attenuation of norepinephrine release, increases of theta power seemed to be indicative for increases tiredness and sedation. This was later confirmed also in humans. The next consecutive frequency range containing the alpha1 waves could successfully related to serotonergic transmission. Here also increases of spectral power were related to attenuation of serotonergic transmission in the rat brain succeeding in a more relaxed physiological state. Likewise alpha2 waves were modulated by drugs acting at the dopaminergic transmission. Several agonists and antagonist at dopamine D1 and D2 receptors were recognized to attenuate or increase spectral power in the alpha2 range. Newest results with respect to beta1 spectral power seem to indicate a relationship to glutamate-ergic transmission in the brain. Finally drugs tackling at the GABA-ergic system like benzodiazepines and barbiturates led to increases of beta2 spectral power. Discovering these relationships between quantitatively assessed spectral EEG power and several neurotransmitter actions might help to interpret pathological patterns the EEG and at the same time contribute to find specific drugs for treatment of some diseases. It could be the base of a rational pharmacotherapy based on pathological EEG findings in the future.
One of the problems in quantitative EEG assessment is the documentation and depiction of the results. Earlier attempts to map the results of spectral power consisted in the construction of single maps for each frequency range. According to the number of frequency ranges defined one ended up with several maps for only one recording condition. This approach is for sure not suitable for every day evaluation of the EEG in the hospital or neurological ambulance. We found an alternative way in transforming the spectral power into colors. Chopping the frequency range from 0 to 35 Hz into 140 single frequencies (0.25 Hz resolution of the spectrogram) we used an additive color mixture like that used in TTV (RGB for red-green-blue) to obtain one single map containing all frequencies according to their spectral power which was programmed into the intensity of the color. Information on colour coding is given in Fig. 7.
Fig. 7 Color coding of EEG Fast Fourier transformed data for map construction is documented on the left side. Mixture of colors is documented on the right side.
Continuous presentation of the results of single sweeps of 0.364 s duration results in an online real-time brain mapping. This has been combined with eye-tracking in order to serve for analysis of individual reactions of the brain to exposure of video clips or internet serving. The resulting pictures are called “Enkephaloglyphes”. Examples are given in Fig. 8 and Fig. 9. A video presentation can be watched on “youtube” by search for “wdimpfel” or “Enkephaloglyphen und Tennis”.
Fig. 8 Example of combination of eye-tracking (red spot on video) with fast dynamic quantitative EEG. Sweep length is 364 milliseconds.
Fig. 9 Example of combination of eye-tracking (red spot on video) with fast dynamic quantitative EEG. Sweep length is 364 milliseconds.
There is also the possibility to average data for 5 seconds and display them in an analog manner. An example taken during mental performance of a complex calculation task is given in Fig. 10 and during performance of a memory task in Fig. 11.
Fig. 10 Example of a brain map recorded during the performance of a concentration performance test (CPT). Blue “%G” indicates that this map of 5s duration is calculated against the median of a reference period of 5 minutes during which the test was performed continuously (set at 100%).
Fig. 11 Example of a brain map recorded during the performance of a memory test (Memory). Blue “%G” indicates that this map of 5s duration is calculated against the median of a reference period of 5 minutes during which the test was performed continuously (set at 100%).
Like any other organ of our body the brain is also prone to a number of disturbances exerted by inner and external disturbances. Internal reasons might be disturbances of the blood flow and by it supplementation of oxygen. External influences might come from intrusion of bacteria and viruses. But also traumatic influences are of considerable meaning. As multifunctional disturbances can be as heterogeneous are the consequences. This is a trivial statement, which shows special consequences, since we are dealing with a highly complex system. In many organs partial disturbance of a particular region only leads to quantitative diminution of total organ capacity. With respect to the brain one might suffer from a variety of different qualitative and quantitative failures. This is quite obvious in the case of stroke or tumors, where particular regions suffer from local disturbances or even lack their function totally. This high complexity together with the network structure as mentioned above asks for evaluation of its total functionality. Traditionally, one discriminates between the so called “somatic” part of the brain and its “psychic” components. Breakdown of the former part is covered by the field of neurology, whereas breakdown or disturbances of the latter are covered by the field of psychiatry. Fortunately, both areas are very often covered by one and the same doctor, at least within Germany. Differences are also reflected by different diagnostic tools. Neurologists mainly use methods based on the visualization of the brain`s “hard ware”, but psychiatrists still depend on the verbal communication of subjectively experienced symptoms.
Broadest use in neurology is made of x-ray, computer tomography and magnetic resonance tomography (MRT). In the second case electroencephalography (inclusive evoked potentials, see later) and functional MRT. In addition, ultra sound technologies are used currently more and more. I mention this because the development of technical methods for assessment of possible “somatic” causes of disease has made large progress, whereas for example electroencephalography as “soft ware check” remained far behind its possibilities. The mathematical description of the electric brain activity by means of the Fast Fourier Transformation (FFT) as well as the definition of particular frequency ranges according to their physiological meanings, as realized in the CATEEM system, seem to be very promising.
After extensive use of the CATEEM system for characterization of drugs and after its successful use in sleep research it was decided to learn more about disturbances of electric brain activity during disease processes. Based on the assumption, that disease origins from disturbances of electro-chemical communication within the brain, the goal emerged to describe the current state of the brain in terms of local frequency patterns. All what was needed now was a data base of healthy subjects characterized in the same manner giving median numbers of all six frequency ranges at each of the 17 locations. A total of 500 healthy volunteers of both sexes aged 18 to 80 participating in our phase 1 medication studies (“first in man studies”) were used to construct a “normal” data base. The distribution of these 102 values (17 electrode positions times 6 frequency ranges) now serves for comparison of single patients to this data-base. According to the distribution function, error probabilities for each single value are calculated and given as numbers from 1 to 4 representing error probabilities from 10:1 to 10000 to 1 (Fig. 12documents the mathematical base of the calculation).
Fig. 12 Calculation base of the aberration index.
This value is called “aberration index”. In medicine “normal values” are based on average numbers collected from a larger population. Even if your blood pressure is not exactly 120 to 80, it can still be normal. But the larger the deviation of individual values the higher the probability of having a pathological feature. For mapping the same type of colour code is used as for depiction of the original data but now they represent the statistical deviation from normality. Since EEG recording is used very often in epileptic patients the following example documents the electric changes in an epileptic patients during normal behavior. Data are averaged over a time of 5 minutes (Fig. 13).
Fig. 13 Recording from an epileptic patients during seizure free interval. Obviously, this patient suffers from temporal lobe epilepsy.
Numerous EEG recordings have been taken in the presence of different diseases. Fig. 14shows an example of the recorded electric disturbance following a stroke.
Fig. 14 Clinical case of brain haemorrhage.
Using this quantitative comparison from patient data with the norm data-base many aberrations of electric brain activity during quite a number of different diseases have been collected up to now. Documentation of the aberration index of single local frequencies during particular diseases is given in Fig. 15.
Fig. 15 Variety of electric aberrations from normality recorded in different diseases.
With this index an objective measure for the occurrence of a functional disturbance of the brain has been developed, which not only can be used as a diagnostic aid, but also could serve as control for a therapeutic success. If the aberration index becomes smaller this is interpreted as an approach to normality. During the course of degenerative disease also a progressive development can be detected using this parameter. From this is becomes obvious that this parameter is useful for the evaluation of the effect of medications as could be shown recently in patients suffering from Parkinson`s disease.
Consequently, a large number of patients have been recorded using the CATEEM technology in numerous hospitals. We always made the experience, that discovery of “brain hardware” abnormalities was accompanied by disturbances of electric activity. Of special interest were disturbances of electric brain communication, which up to now have escaped objective quantitative measurements. I like to mention headache and especially migraine, since we – using the method developed by us – succeeded for the first time to record clear, statistically significant, reproducible changes of electric patterns of this disease. Recording from more than 600 patients suffering from migraine indicated that in about 90% of the cases a deviation from normality was detected. But results also revealed the heterogeneity of headache and migraine, since deviations form normality not only occurred within different brain areas but also with respect to different frequencies. These deviant electric patterns were recorded during the ache-free interval and support the neurogenic origin of the disease. However, many textbooks refer to the vascular origin of migraine, based on disturbance of the blood flow. Since higher nervous systems activity at the same time recommends higher oxygen demand, it is very difficult to see where the original disturbance comes from, the famous egg-hen problem.
CATEM® is also used for neuromonitoring purposes on the intensive care unit or during surgery in the operating theater. Deviations from normality are depicted within minutes “on line” including statistics of deviation. Numerical values are depicted as bar graphs at the same time. During neuromonitoring also the time course of changes is depicted for one channel of choice. This kind of documentation of the quantitative EEG enables one to follow any physiological and pathological changes of the EEG in an absolute objective manner.