Annals of Indian Academy of Neurology
: 2007  |  Volume : 10  |  Issue : 2  |  Page : 121--127

Technical standards for digital electroencephalogram recording in epilepsy practice

Dinesh S Nayak, P Sajeesh 
 Department of Neurology, R. Madhavan Nayar Center for Comprehensive Epilepsy Care, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Trivandrum, India

Correspondence Address:
Dinesh S Nayak
Sree Chitra Tirunal Institute for Medical Sciences and Technology, Trivandrum - 695 011


With the advent of digital technology in the recording of the electroencephalogram (EEG) in the last decade, analogue paper-EEG machines have all but disappeared. While there are several advantages of digital EEG over its analog counterpart, like being paperless and therefore easy to store and the ability to change montages and filter settings during review, there is wide disparity in the standards of EEG recording, display and reporting in laboratories across the country. Colorful brain maps conveying little meaning are usually appended to reports. This article reviews the minimum standards that must be observed for recording digital EEG as recommended by the International Federation of Clinical Neurophysiology (IFCN) and illustrates the importance of use of appropriate derivations, montages, filters and gains during recording and review of digital EEG in the context of evaluation of patients with suspected epilepsy.

How to cite this article:
Nayak DS, Sajeesh P. Technical standards for digital electroencephalogram recording in epilepsy practice.Ann Indian Acad Neurol 2007;10:121-127

How to cite this URL:
Nayak DS, Sajeesh P. Technical standards for digital electroencephalogram recording in epilepsy practice. Ann Indian Acad Neurol [serial online] 2007 [cited 2022 Jun 29 ];10:121-127
Available from:

Full Text


Digital technology has changed the way electroencephalogram (EEG) is recorded in the last two decades. Digital EEG machines are now manufactured in India and are being used extensively across the country. These machines have all but replaced conventional paper analog EEG machines. There are several advantages of digital EEG over the conventional analog EEG: digital EEG is paperless. It can be stored in electronic media and therefore saves storage space. Post hoc changes in filter settings, recording speed and montage reformatting is possible with digital technology. EEG data can be electronically sent to an expert anywhere in the world for obtaining a second opinion. Synchronized digital video-EEG technology has made long-term monitoring easier and is used in the presurgical evaluation of patients with medically refractory epilepsy, in the classification of the type of epilepsy and to differentiate between epileptic and nonepileptic events. It is also used in intensive care units, mainly to look for nonconvulsive seizures. Continuous ambulatory EEG akin to Holter monitoring can also be performed, especially for evaluation of frequent paroxysmal events in children who may not cooperate for hospital admission for long-term monitoring. It is beyond the scope of this paper to discuss long-term video EEG, ambulatory EEG and continuous EEG in the intensive care setting.

It has been observed that standards of digital EEG recordings in our country vary widely, as is their interpretation. Proper recording and interpretation of digital EEG is of crucial importance in the appropriate management of patients who undergo the test. The aim of this paper is to highlight the technical standards for digital EEG recording, the use of various derivations and montages in epilepsy practice.

 Minimum Standards for Digital EEG Recording

The International Federation of Clinical Neurophysiology (IFCN) has set certain standards for digital recording of clinical EEG.[1],[2] These recommendations are discussed under the following headings:

Patient information

Patient demographic details such as name, age, clinical diagnosis, state of the patient, medication details, test number and comments have to be entered.

Documentation during recording

At the start of the recording, square wave calibration signals of 100 V, of 1-2 s duration, in the referential derivation must be carried out to assess the integrity of the amplifiers and analog-to-digital (A-D) conversion. The technologist must be able to enter comments such as eye opening and closing, hyperventilation, photic stimulation, drowsiness, sleep, artifacts, etc and there must be provision for entering free text.


The tiny electrical signals produced by the brain need to be amplified tens of thousands of times in order to be displayed on the computer screen or on paper. The amplification factor is called gain and the unit of gain is the decibel (dB). Amplifiers receive input voltages within a certain range called dynamic range, which can be controlled by means of sensitivity units of either microvolts per millimeter or millivolts per centimeter. The differential amplifier measures the potential difference between two electrodes. The difference between the two voltages is then amplified. If the voltage between the two electrodes is the same, the voltage difference is zero and hence there is no amplification. This is called common mode rejection and is able to reject extraneous noise to a great extent. Amplification and channel acquisition must be available for at least 24 channels and preferably 32 channels, of recording EEG along with artifact channels. The sampling rate must be at least 200 samples/s. Higher rates are preferable and must be in multiples of 50 or 64, e.g., 500 or 512. The low frequency filter (high-pass filter) should be set at 0.16 Hz or less for recording. Higher settings of the low filter for recordings are discouraged during routine recordings, except in certain specific settings. The low frequency filter is labeled as hertz (Hz) or can be expressed as time constant in seconds . The relationship between the turnover frequency of the low frequency filter and time constant is given in [Table 1]. The 50 Hz notch filter must be available to cut off external electrical disturbances, but should not be used routinely, since it can distort sharp components. It should therefore be used only when all measures against 50 Hz noise fail such as eliminating mains interference by means of proper earthing and positioning of the equipment and reducing electrode impedances. Noise is defined as small fluctuating output even when there is no input signal. Noise can contaminate any sensitive electrical measuring system.

Since filters distort both amplitude and interchannel phase of signals, it is better to minimize the use of filters during EEG recording. Recording should be made on a referential montage to facilitate subsequent montage reconstruction. Digitization with 12 bits will provide a dynamic range from 0.5 V to 1023 V. Electrode impedances must be below 5 kU and preamplifier input impedances must be more than 100 MU. Common mode rejection ratio must be at least 110 dB for each channel measured at amplifier input.


Digital EEG equipment must be able to display the EEG with the same temporal and spatial resolution traditionally used for paper recordings. Remontaging with bipolar and referential montages should be possible. Digital low frequency (high pass) filters of 0.5, 1.0, 2.0 and 5.0 Hz and digital high frequency (low pass) filters of 15, 35 and 70 Hz must be available. For majority of EEG investigations the recorded signal ranges between 1 and 70 Hz. Information will be lost if the frequency range (band pass) is narrowed. It is a common practice among many EEG laboratories to give prints-outs of digital EEGs with filter setting at 5 Hz (low frequency filter) and 15 Hz (high frequency filter). Such a narrow band-pass will cut out the slower and faster frequencies and can considerably distort epileptiform activity [Figure 1]. A low frequency filter set at higher than 1 Hz should not be used routinely to attenuate slow-wave artifacts. On doing so, pathological delta activity may be missed. During review, the system should be able to display montage designations, gain and filter settings, technologist's comments and event markers along with raw and transformed EEG data. The screen display scaling should be set such that 1 s occupies 30 mm with a minimum display resolution of 120 data points per second, per channel. More compressed and more expanded time scales should also be available, e.g., 7.5, 15, 30 or 60 mm/s. For 60 mm/sec display, at least 200 data points per second should be presented for each channel. The standard interchannel separation is 10 mm. The standard video screen must have a minimum resolution of 4 pixels per vertical millimeter. For paper printouts, at least 300 dots per inch (dpi) resolution is needed.

Electrode nomenclature

When additional scalp sites are required, these must be placed between the traditional 10-20 electrode system sites and are named the 10% system or the extended 10-20 electrode system. In the coronal row, AF lies midway between rows Fp and F; FC between F and C; CP between C and P and so on; PO between P and O. Thus, CP1 will be between CPz and CP3; C1 will be between Cz and C3; C5 will be between C3 and T3.

Exchange of clinical EEG

The manufacturer must use a standard file format so that the EEG can be reviewed in any computer apart from the manufacturer's equipment. Clinical EEG data belongs to the health care providers or to the patients, not to the manufacturers.

 The Use of Derivations and Montages

Multichannel systems are able to simultaneously record cerebral activity from many electrodes. Strategies for connecting various electrodes to the various channels are called ' derivations ' or ' methods '. Within these general strategies, one can make different combinations and sequences of electrode connections, which is referred to as ' montages '. Two types of derivations are commonly used: bipolar and referential.

Bipolar derivation

In this widely used method, rows of electrodes are connected to consecutive channels, such that an electrode is attached to lead 2 of one channel and lead 1 of the next channel (e.g., Fp2-F4 in channel one and F4-C4 in channel two. F4 is connected to lead 2 of channel one and lead 1 of channel two). The polarity convention in recording with differential amplifiers is the following: if lead 1 is negative with respect to input 2, there is an upward deflection. If lead 1 is positive, there is a downward deflection; if lead 2 is negative with respect to lead 1, there is downward deflection, if lead 2 is positive, there is an upward deflection. The linking of serial pairs of electrodes in a bipolar chain can be used to locate localized peaks of activity by phase reversals [Figure 2]a. In this figure, the sharp transient deflects downwards at F8-T2, indicating that T2 is more negative than F8. The deflection in the next channel (T2-T4) is upward, indicating that T2 is more negative than T4. Therefore, there is a localized negativity at T2, which is seen as a phase reversal between two channels in which T2 is the common electrode.

Common reference derivation

This is the simplest multi-channel montage to understand. Each electrode is connected to one input of the amplifiers (usually lead 1) and all the lead 2 inputs are joined together and connected to another electrode. The common reference electrode is chosen such that the possibility of picking up signal and/or artifact is minimal, but lack of activity can never be assumed, because there is no truly inactive area over the scalp or the body. If there is very little activity at the site of the reference electrode, it is truly representative of the voltage differences between the two electrodes and the EEG is easy to interpret [Figure 2]b, c. On the other hand, if the reference electrode site is active, it will appear in all the channels, making it difficult to interpret the EEG [Figure 2]d.

Common average reference derivation

Although an ideal 'inactive' reference is difficult to achieve, it is possible to devise a reference which is adequate for most EEG recordings.[3] This is the 'common average reference', This uses the average of the electrical activity at all electrodes as reference against which individual activities are measured and is similar to the 'Wilson' electrode used in electrocardiography which assumes the mean potential of all electrodes in use. The arithmetic mean of a set of values will obviously lie between the highest and the lowest values in that set. The sum of the differences of these values from the arithmetic mean will be zero. Thus, the average reference will be near zero and therefore will reflect the electrical field fairly accurately. One of the problems with this recording is that if one of the electrodes shows a very high potential, this will affect the average so that equal deflections of one Nth its magnitude and of opposite polarity will appear in all the electrodes, where N is the total number of electrodes [Figure 3]a, b. The problem can be solved if the electrode showing the large potential is removed from the average [Figure 3]c.

Source reference

This is another type of referential derivation, in which each electrode is referred to the weighted average of the activities of the immediately neighboring electrodes, using the Laplacian equation. This is a kind of local average reference for each electrode. This is somewhat similar to the common average reference, but it displays focal epileptiform activity better.

 Comparison of Different Methods of Derivations

No single method of derivation is uniquely ideal for displaying all types of activity. Although both bipolar and referential derivations are equally good to demonstrate focal epileptiform activity, the phase reversal of this activity in bipolar derivation makes it ideal for demonstration. On the other hand, if a phenomenon produces gentle potential gradients over a large area of scalp, the potential differences between adjacent closely spaced electrodes of bipolar chains will be small and less conspicuous, than when a referential montage is used. It is therefore, prudent to use different derivations and montages to study an activity of interest [Figure 4]a, b, c. The user must take advantage of this facility provided by the digital EEG. Another occasion where referential derivation is perhaps more appropriate than bipolar is in the evaluation of occipital spikes. This is because in the commonly used bipolar montages, O1 and O2 electrodes are at the end of the chain, e.g., T5-O1 or P3-O1. O1 is subsequently not connected to another electrode. Therefore, when O1 is negative with respect to P3 and T5, the deflection will be positive or down (as per the convention Pos-Neg=dowN ). No phase reversal will be seen despite the focal activity because of the 'end of the chain' phenomenon, whereas in the referential montage, it will be seen as negative [Figure 5]a, b. A similar feature can be expected when focal epileptiform activity is confined to Fp1/Fp2, due to 'beginning of the chain' phenomenon.

The ideal reference electrode must be 'inactive' and may be cephalic or extracephalic (e.g., shoulder). However, it is impossible to find a truly inactive area in the body. For example, the shoulder reference is not expected to pick up any cerebral activity and therefore theoretically ideal, but it can pick up ECG artifacts, which will then contaminate the EEG. Nevertheless, it is possible to judiciously use a relatively inactive reference. During wakefulness, the midline electrodes such as Cz and Pz may serve as good reference electrodes, but these are not ideal during sleep because they pick up sleep activity such as vertex sharp waves and therefore become active. Similarly, in the evaluation of suspected temporal lobe epilepsy, the commonly used earlobe reference (A1/A2) is not ideal because of its proximity to the fields generated by temporal lobe spikes, which will make it 'active' [Figure 2]d.

 Brain Mapping

This facility is available in almost all digital EEG systems. Brain mapping is the topographic display of digitally quantified EEG information on a stylized head outline.[4] It displays the electrical field over the scalp at one moment in time. It must be emphasized that brain mapping is an extension of carefully analyzed primary EEG data and all it does is to display the primary data in a visually pleasing manner. It does not substitute careful analysis of the EEG by the electroence-phalographer. The technique is fraught with danger, as artifacts may be overlooked or may be generated during the process of analysis and mapping. It is common practice among laboratories to give colorful printouts of brain maps that carry information that is of no relevance or is actually misleading. Nevertheless, it can display electrical fields when the transient is chosen properly and filter settings are correctly used. For example, it is appropriate to increase the low frequency filter to 3 or 5Hz to remove background slow wave activity when we want to study the field of a spike [Figure 6]a, b. Another use of the brain map is to elegantly display the tangential dipole in benign rolandic epilepsy (BRE). The spikes in BRE are located in mid-temporal and central regions, while the positive end of the dipole is made out as a small downward deflection in the frontal regions. This positivity must be seen in a referential derivation and not in the bipolar derivation. In [Figure 6]b, the brain map displays the tangential dipole nicely, with the maximal negativity in the temporal and central regions and the positivity in the frontal regions. Again, low frequency filter in this instance has been kept at 3 Hz to cut off the slow activity to highlight the spike activity.

 EEG in the Evaluation of Some of the Epilepsies

Temporal lobe epilepsy : In mesial temporal lobe epilepsy (MTLE), which is the most common surgically remediable epilepsy syndrome, the epileptiform discharges are maximal at the anterior temporal electrodes [Figure 2], whereas in neocortical temporal lobe epilepsy, the spikes are maximal at the mid-and/or posterior temporal locations (T3/T4 and T5/T6). In the standard 10-20 electrode system, the antero-basal temporal lobe is not well covered and therefore it is important to use additional electrode placements such as the anterior temporal (Silverman) electrodes designated T1/T2. This is located at a point 1 cm above the junction between anterior 1/3 rd and posterior 2/3 rd of a line connecting the outer canthus and the tragus. In unilateral MTLE, the incidence of bilateral independent temporal epileptiform abnormalities is around 30% during routine scalp EEG recording, whereas it increases to around 70% during prolonged video-EEG monitoring.[5]

Frontal lobe epilepsy : A discrete epileptiform focus is found in only about 9% of patients, while some 59% will show more widespread discharges over fronto-central or fronto-temporal areas.[6] The difficulty of EEG localization in frontal lobe epilepsy may be due to the fact that mesial frontal and orbitofrontal areas are relatively inaccessible. There is also a tendency for discharges to rapidly spread both within the frontal lobes and also secondarily generalize. Sometimes, the epileptogenic region within the frontal lobe is itself extensive. Mesial frontal and orbitofrontal foci may show secondarily generalized spike -wave discharges. Persistent asymmetry between the sides may be the only way to distinguish from the 3 Hz spike-wave activity of idiopathic generalized epilepsy. Patients with a focus in the supplementary motor area may show interictal epileptiform activity confined to the vertex [Figure 5]. It must be noted that montages most often used in the laboratories for routine EEG tend to omit mid-frontal and vertex (Fz and Cz) electrodes. Therefore, an epileptic focus at the midline vertex will be missed

Idiopathic generalized epilepsy : The EEG findings in childhood absences consist of typical 3/sec, frontally dominant generalized spike-wave activity, best brought on by hyperventilation. The child must be asked to keep tapping the couch with his finger during hyperventilation. During a burst of epileptiform activity, the child ceases to tap for the duration of the discharge, after which he resumes tapping, indicating that the child was absent during that time. In juvenile myoclonic epilepsy (JME) the abnormality consists of generalized, frontally dominant spike, polyspike and wave discharges at greater than 3 Hz (fast spike and wave discharges). Again, it may be necessary to use a referential derivation to accurately assess the regions of maximal amplitude [Figure 7]. Focal spikes are being increasingly recognized in JME, but these usually shift from one hemisphere to another. There are no established criteria concerning what degree of asymmetry of generalized spike-wave activity should be regarded as evidence of a focal onset.

 Sleep and Epilepsy

In general, there is an increase in epileptiform activity during sleep when compared to the awake state.[7] Generalized epileptiform discharges occur chiefly in non-REM sleep and are maximum in the early hours of sleep or just before awakening. Patients with myoclonic seizures like juvenile myoclonic epilepsy show increased discharges during early stages of non-REM sleep and just before awakening. Focal discharges of temporal, frontal and occipital origin also typically increase during drowsiness and stages 1 and 2 non-REM sleep. In about 20% of patients with temporal lobe epilepsy, epileptiform activity occurs exclusively in non_REM sleep.[8] It is therefore, vitally important to record the EEG in both awake and sleep states in patients suspected to have epilepsy. In our institution, we ask patients to sleep only for three or four hours on the night prior to the scheduled EEG appointment. We record at least 20 min of EEG in awake state and another 20 min in the sleep state. In some patients who have seizures exclusively or predominantly in sleep, we usually record the EEG for one hour in the sleep state. In children sedatives such as chloral hydrate or promethazine can be used to induce sleep.

Diagnostic sensitivity and specificity of EEG in epilepsy

The diagnostic sensitivity of a single awake EEG with photic stimulation and hyperventilation is about 50% in adults with epilepsy.[4] The sensitivity increases to 92% after 4 EEGs.[9] A single awake plus sleep record will be positive in about 80%.[10],[11]

The diagnostic specificity of the EEG depends on the proper interpretation of the EEG. Misinterpretation of normal variants such as benign sporadic small spikes (BSSS), 14 and 6 positive spikes, wicket waves, 6 Hz phantom spikes, as well as artifacts can lead to a wrong diagnosis of epilepsy and unnecessary antiepileptic drug therapy. It is also important to understand that a normal EEG does not exclude epilepsy. The EEG may remain normal despite repeated EEG recordings in 5% to 10% of patients. Such patients usually have only simple partial seizures or mesial frontal or parietal lobe epilepsy.

Interpretation of the EEG must always be done in the clinical context. There must be a dialogue between the referring physician and the electroencephalographer regarding the clinical problem and the questions EEG is expected to answer. In general, the interictal EEG cannot exclude epilepsy; and the amount of epileptiform activity does not provide a guide to seizure frequency or therapeutic response despite widespread belief to the contrary.


Digital EEG is here to stay. It has many advantages over conventional analog paper EEG machines. Technologists must strictly adhere to the minimum technical standards of digital EEG recording as laid out by the IFCN. The reporting neurologist or clinical neurophysiologist must be familiar with the various derivations and montages that are commonly used and their advantages and limitations. It is important to review the EEG in different derivations and montages in order to get a clear idea of the electrical field generated by an epileptiform discharge. Brain mapping is a frequently misused tool, but when judiciously used, can display electrical fields that can be understood more easily than the primary EEG. It cannot however, replace careful analysis of the primary EEG data by the electroencephalographer. Technologists and electroencephalographers must receive adequate training before they set up long-term video-EEG units for evaluation of patients with epilepsy.

Every effort must be made to get an awake and sleep EEG. In our laboratory, we ask patients to sleep only for three to four hours the previous night. Attempts must be made to obtain natural sleep. In case this fails, we use Triclofos, a derivative of chloral hydrate at a dose of 25 mg/kg for children and 1000 mg for adults.Recording of at least 20 min each of awake and sleep must be carried out. If for some reason only awake record is possible, it must be for at least 30 minutesInitial parameters for recording include time constant of 0.5 Hz, HFF of 70 Hz, sensitivity of 7.5 V/mm and chart speed of 30 mm/cm.Electrode impedence must be below 5 K for all leads.Activation procedures such as hyperventilation and photic stimulation must be carried out. Hyperventilation is contraindicated in patients aged >65 years, cardiovascular and respiratory diseases, advanced pregnancy and recent stroke.Photic stimulation must be carried out for different frequencies ranging from 1 Hz to 60 Hz, each lasting for about 5 seconds.


1Nuwer MR, Comi G, Emerson R, Fuglsang-Frederiksen A, Guerit JM, Hinrichs H, et al . IFCN standards for digital recording of clinical EEG. In : Deuschl G, Eisen A, editors. Recommendations for the Practice of Clinical Neurophysiology: Guidelines for the International Federation of Clinical Physiology (EEG suppl. 52). International Federation of Clinical Neurophysiology. Elsevier Science B.V: 1999. p. 11-4.
2Ebner A, Sciarretta G, Epstein CM, Nuwer M. EEG instrumentation. In : Deuschl G, Eisen A, editors. Recommendations for the Practice of Clinical Neurophysiology: Guidelines for the International Federation of Clinical Physiology (EEG suppl. 52). International Federation of Clinical Neurophysiology, Elsevier Science B.V: 1999. p. 7-10.
3Cooper R, Binnie CD, Osselton JW, Prior PF, Wisman T. EEG technology. In : Binnie CD, Cooper R, Mauguiere F, Osselton JW, Prior PF, Tedman BM, editors. Clinical Neurophysiology Vol 2. EEG, paediatric neurophysiology, special techniques and applications. Elsevier B.V: Amsterdam; 2003. p. 8-104.
4Binnie CD, Stefan H. Modern electroencephalography: Its role in epilepsy management. Clin Neurophysiol 1999;110:1671-97.
5Ergene E, Shih JJ, Blum DE, So NK. Frequency of bilateral independent interictal epileptiform discharges in temporal lobe epilepsy. Epilepsia 2000;41:213-8.
6Quesney LF, Constain M, Rasmussen T, Olivier A, Palmini A. Presurgical EEG investigation in frontal lobe epilepsy. In : Surgical Treatment of Epilepsy: Epilepsy Res Suppl. 5th ed. Theodore WH: 1992. p. 55-69.
7Billiard M. Epilepsies and the sleep-wake cycle. In : Sleep and epilepsy. Sterman MB, Shouse MN, Passouant P, editors. Academic Press: New York; 1982. p. 269-86.
8Nayak SD, Radhakrishnan K, Venugopal A. Predictive value of scalp EEG in correctly localizing the seizure focus in patients with temporal lobe epilepsy. In : Radhakrishnan K, editor. Medically refractory epilepsy. Sree Chitra Tirunal Institute for Medical Sciences and Technology: Trivandrum; 1999. p. 113-8.
9Salinsky M, Kanter R, Dashieff RM. Effectiveness of multiple EEGs in supporting the diagnosis of epilepsy and operational curve. Epilepsia 1987;28:331-4.
10Ellingson RJ, Wilken K, Bennett DR. Efficacy of sleep deprivation as an activation procedure in epilepsy patients. J Clin Neurophysiol 1984;1:83-101.
11Westmoreland BF. Epileptiform electroencephalographic patterns. Mayo Clin Proc 1996;71:501-11.