The History of Ambulatory EEG—A Personal Perspective

CHAPTER 1

THE HISTORY OF AMBULATORY EEG—A PERSONAL PERSPECTIVE

JOHN S. EBERSOLE, MD

PROLOGUE

Reviewing this history is useful not for the details of technologies that are now obsolete, but for the story of the struggle to bring a new diagnostic procedure to fruition. Hopefully, this will inspire young clinical investigators to accept and address similar challenges in future eras.

INTRODUCTION

To understand the significance of the development of ambulatory EEG (aEEG) and the problems faced and eventually overcome in its evolution, one must go back in time to the clinical neurophysiology of the 1970s. The Grass eight-channel, pen-writing EEG machine was the long-standing clinical workhorse. New on the horizon was a 16-channel EEG machine that would become the latest great thing at the best centers. The value of long-term EEG monitoring (LTM) with video recording was just becoming established. Now seizures could be recorded commonly, not simply spikes, as was usual for laboratory EEG. Early pioneers of LTM had to be creative regarding how to record both EEG and patient behavior in a synchronized fashion. Initially, two cameras were used with split-screen video technology. Remember, this was a predigital era. One camera imaged the moving EEG paper with newly penned waveforms, and the other recorded the patient. Engineering advances eventually included the “reformatter” that recorded multiplexed EEG on the edge of the videotape recording the patient. These advances in epilepsy diagnosis and characterization were monumental at the time, but they came with some drawbacks. There were only a handful of epilepsy monitoring units (EMUs) in the country, hospitalization was required, it was expensive to keep patients in hospital for days, and the normal day-to-day triggers for seizures were not present in a hospital setting. Patient activity and mobility were restricted often to sitting or lying in bed. Being in the hospital was clearly unnatural and likely to inhibit both real epileptic and nonepileptic episodes. In fact, it was soon realized that a patient’s typical seizure frequency often plummeted when he or she was hospitalized. If there was only a way to obtain the benefits of long-term EEG recording in an outpatient, rather than hospital, setting.

Cardiologists have always faced a similar problem to that of epileptologists, namely the detection of physiological abnormalities that are paroxysmal. Cardiac arrhythmias are often intermittent, and they are often not recorded on a standard duration ECG. Fortunately for them, the ECG is relatively large in amplitude, and one channel of data is usually sufficient to identify the problem. Accordingly, the development of ambulatory ECG monitoring was technically easier, and it preceded that of aEEG. Norman Holter in 1953 devised a system for the radio-telemetry of single-channel ECG from an ambulatory patient. Later this transitioned to long-term, single channel, cassette recordings of ECG, and the famous “Holter monitor” was born (1).

CONTINUOUS AEEG RECORDING

However, the development of any reasonable aEEG counterpart would have to await the solution of four technical problems—more channels than just one, preamplification of the much smaller EEG signal, a recording duration that was at least 24 hours, and a means to play back and analyze all the recorded data in an efficient fashion. One by one in the early 1970s these obstacles were overcome. A four-channel, portable tape recorder, weighing approximately 1.5 pounds, was developed by Marson and McKinnon for industrial purposes (2). Ives and Wood later showed that recording EEG on it was feasible (3). These recorders were analog devices that used 1/8-inch tape and four recording heads. Tape speed was reduced to 2 mm/sec, so that a standard C120 cassette could record at least 24 hours of continuous data. However, the problem of amplifying the EEG signals sufficiently before transcription onto tape remained. Theoretically, it made sense to try to amplify the EEG as close as possible to the head in order to minimize lead artifact from movements during wakefulness. In 1978, Quy developed an amplifier chip that could be glued onto the scalp with collodion (4). Each chip represented a single channel and had its own Grid 1 and 2 inputs. Electrode leads were plugged into the chip, so montages were necessarily derived on the head. Electrodes with bifurcated leads were necessary if montages with linked channels, such as a bipolar chain, were desired.

That still left the problem of reviewing hours of recorded data. One solution was to print out the entire recording with a jet ink writer at up to 20× real time. To reduce the volume of paper produced, which was still large, the recording paper speed could be reduced, but this meant that only generalized seizures could be discerned. The paginated, rapid video playback was the conceptual breakthrough that made efficient analysis of aEEG data feasible (5). This new video playback was also an analog device, that is, essentially an oscilloscope. Repeated refreshing of the screen allowed both a static page of data to be visualized and also sequential pages of data at 20× or 60× real time. Video page lengths could be either 8 or 16 seconds. At the fastest replay speed, 24 hours of data could be reviewed in 24 minutes. A critical additional feature of the playback unit was the simultaneous audio reproduction of one data channel. At 20× or in particular 60×, standard EEG frequencies would become audible. If an accurate time registration during the recording was desired, unfortunately one of the four channels had to be sacrificed for that function. Thus, in 1979, Oxford Medilog introduced a completed, commercially available, four-channel aEEG recorder, preamplification system, and paginated video playback (Figure 1.1). However, questions remained as to whether such a system would be clinically useful and how it should be used to extract the most information efficiently.

Now that epileptologists had become accustomed to eight- and soon 16-channel EEG, it was going to be difficult for them to accept four-, let alone three-, channel EEG data. That was clearly a giant step backward. It was accordingly easy to dismiss the new recorders as less than useful toys. It took a bit more insight to see the benefit that a longer recording brought, particularly as and outpatient, even with significantly fewer channels. Convincing neurologists that a three-channel EEG carried any worth would require a controlled comparison. The first study undertaken by Rob Leroy and me was to design rationally aEEG montages that would maximize the detection of focal ictal and interictal abnormalities (6). Proper montages were clearly necessary before even attempting to evaluate the efficacy of the technique. Ambulatory EEG montages had to fulfill two goals. Although the first was spike/seizure detection, equally important was that the data had to be displayed in a form that was conductive to perception on rapid video playback. Nothing would be gained by recording an event that was overlooked on review.

FIGURE 1.1 The Oxford Medilog 4–24 recorder was the first commercial aEEG cassette recorder. It utilized the HDX-82 on-head preamplifier chip. The Oxford Medilog page mode display unit (PMD12) was the first commercial device to enable analysis of rapidly replayed aEEG in both a visual and auditory format.

Source: From Ref. (15). Ebersole, JS (Ed.): Ambulatory EEG Monitoring. New York, NY: Raven Press, Wolters Kluwer Health; 1989:365.

We thought that it might not be necessary to cover the head uniformly, and thus fewer channels would be reasonable. We recorded interictal and ictal EEG abnormalities from over a hundred adult and adolescent patients sequentially admitted to the West Haven VA EMU for monitoring and determined, perhaps really not to anyone’s surprise, that the frontal and temporal head regions bore the greatest percentage of epileptiform abnormalities, namely 78% (6). The central, parietal, and occipital regions, particularly in adults, were relatively quiet. Given that preferential sampling was a technical necessity with a three- to four-channel aEEG, it became clear that we had to concentrate our montages in these areas.

We chose to develop three-channel montages because the fourth channel was commonly used for time/event marking or ECG. Montages with longitudinal temporal and transverse frontal derivations were found to be most useful. When these channels were organized into a chain in left-transverse-right sequence, surface negative potentials in the frontotemporal regions appeared as a phase reversal common to two channels. This afforded enhanced perception of the most common focal abnormalities. Generalized discharges were easily detected, particularly in the frontal channel. We quickly learned that it was not a good idea to use a truly linked three-channel montage utilizing only four electrodes. Loss or significant artifact in one of the frontotemporal electrodes would confound two thirds of the data. Accordingly, separate electrodes were used for each of the three channels.

Having our montage, we proceeded with a series of studies addressing the relative diagnostic yield of three-channel aEEG versus eight-channel LTM (which was standard at the time) (7,8

Only gold members can continue reading. Log In or Register to continue