Practical guide: recording, analysis and interpretation
Subdural electrocorticographic recordings and minimally invasive alternatives
Electrocorticography (ECoG) with a linear subdural platinum electrode strip is the current gold standard for monitoring spreading depolarizations in the clinic. The most commonly used electrode contains six platinum contacts spaced at 10mm along the strip (Wyler, 5mm diameter, Ad-Tech, Racine, WI, USA). Subdural placement on the cortical surface allows monitoring of viable tissue at risk for secondary injury. Recordings run for up to 14 days in aneurysmal subarachnoid hemorrhage (aSAH), and 7 days in other conditions.
Subdural strip electrode and subdermal needle ground/reference electrodes
Subdural linear strip electrode (left) and subdermal needle electrodes (right)
In patients that require an open craniotomy after traumatic brain injury, intracerebral hemorrhage, or malignant hemispheric stroke, the electrode strip is placed on peri-contusional, peri-hematomal, or peri-infarct tissue, respectively. In patients that require an open craniotomy after aSAH, the electrode strip is targeted at the vascular territory of the aneurysm-carrying vessel. This area is often a predilection site for delayed cerebral ischemia because it bears the most blood clots.
Correct/poor electrode placement in relation to the injury (open craniotomy)
These two images illustrate both correct and poor placement of subdural strip electrodes. Each of these images show a brain injury with a central contusion and surrounding penumbra. The pink line emerging from the injury indicates the hypothetical path of a spreading depolarization. Since spreading depolarizations have difficulties crossing sulci and major fissures, the strip should preferably be placed along a single gyrus so that the depolarization can contact most of the embedded electrodes. This allows the best documentation of the propagation. Increasing the distance of the strip electrode from the injury decreases the chance of detecting spreading depolarizations.
The tail of the electrode strip is tunneled subcutaneously beneath the scalp and exited 2–3 cm from the craniotomy scalp incision. It should then be coiled and sutured to the scalp to provide strain relief and guard against accidental displacement. The previously removed bone flap may be re-secured with a titanium clamp or plating system, followed by standard wound closure (paying special attention not to place any sutures around the electrode).
Electrical ground is provided by a platinum needle (Technomed Europe, Maastricht, Netherlands or Natus Neurology - Grass, Warwick, RI, USA), silver/silver chloride (Ag/AgCl) scalp electrode, or a self-adhesive Ag/AgCl patch electrode on the shoulder. For direct current (DC) referential recordings, a platinum needle or Ag/AgCl sticky electrode is placed as a reference on the mastoid or frontal apex, away from muscle attachments.
Beware of the following pitfalls:
Gentle traction may not be sufficient to remove the strip after the monitoring period when it is trapped/pinched by the bone flap or the titanium fixation, or if the subcutaneous tunnel is too tight.
A cerebrospinal fluid (CSF)-fistula may develop. Therefore, the following precautions should be taken:
- a sufficient bone should be removed with an osteotome or rongeur where the tail of the strip exits, and through which the electrode can be withdrawn by gentle traction at the end of monitoring;
- the tail of the strip should exit the craniotomy and scalp in line with the rest of the electrode, and not be curved or bent at an angle;
- no plating hardware should be used next to the location of the strip; and
- the subcutaneous tunnel should be made sufficiently long and dilated, e.g. with a Halsted-Mosquito clamp.
- an additional fully penetrating skin suture should be performed with some distance to the scalp exit point, and a sufficient amount of absorbable hemostatic gelatin sponge should be placed under the bone flap - particularly in the area of the strip. Antibiotic treatment beyond standard preoperative prophylaxis is not recommended.
Not all patients require a craniotomy. Here, it is also possible to place a subdural electrode strip through a burr hole, or to monitor with an intraparenchymal electrode array (Spencer, 1.1mm diameter, Ad-Tech, Racine, WI, USA) that is placed through a burr hole or a multi-lumen bolt. This is customary practice for intracranial pressure (ICP) and tissue partial pressure of oxygen (ptiO2) monitoring.
Advantages and disadvantages of intraparynchemal sensors and strip electrodes
Intraparenchymal sensors
| Advantages |
|---|
| - good safety profile |
| - lower risk of CSF fistula compared to a subdural strip electrode |
| Disadvantages |
|---|
| - cause necrosis at insertion site |
| - cause local disruption of the blood-brain barrier (upregulation of inflammatory cell types, extravasation of plasma proteins) |
| - fail to detect spreading depolarizations in the wider vicinity |
| - subject to spreading depolarization-triggered pH and ptiO2 changes that may influence the DC potential |
Strip electrodes
| Advantages |
|---|
| - less invasive |
| - permits recording of a larger cortical area |
| Disadvantages |
|---|
| - risk of CSF fistula |
| (- subject to spreading depolarization-triggered pH and ptiO2 changes that may influence the DC potential) |
Resistive and nonpolarizing Ag/AgCl and calomel electrodes are ideal for the recording of low-frequency potentials. Their toxicity, however, precludes an invasive use in patients. Platinum electrodes are safe to use in patients, but it has been assumed that they distort low-frequency and DC potentials due to their polarizable character. In the clinical setting, alternating current (AC) coupled amplifiers with a 0.01–0.02 Hz lower frequency limit were used until it became clear that DC coupled amplifiers could be reliably used for continuous ECoG monitoring as well. This has the great advantage that the durations of the DC shifts can be measured. Practical experience of DC recordings suggests that they are a suitable substitute for AC techniques.
Noninvasive technologies such as continuous scalp electroencephalography (EEG) are not yet sufficient to detect spreading depolarizations without simultaneous subdural ECoG monitoring. However, correlates of spreading depolarizations have previously been identified in combined scalp EEG/ECoG recordings. Scalp EEG may hold particular promise for noninvasive monitoring of spreading depolarization, if it is combined with other noninvasive technologies that measure regional cerebral blood flow (rCBF) or its surrogates.
Technical equipment for the recording and monitoring of spreading depolarizations
| 1 | Brainamp amplifier | 3 |
Laptop with software a) LabChart b) Brain Vision Recorder |
5 | Powerlab 16/SP analog/ digital converter |
| 2 | GT205 amplifier, 0.01–50 Hz | 4 | Integra Licox Brain Tissue Oxygen Monitoring System | 6 | Component Neuromonitoring System (CNS) Monitor |
- BrainAmp amplifier, 0–50 Hz
BrainProducts GmbH, Munich
Amplifier for DC-ECoG recordings. - GT205 amplifier, 0.01–50 Hz
ADInstruments, New South Wales, Australia
Electrodes 2-6 of the strip electrode are connected in sequential unipolar fashion to this amplifier, each referenced to an ipsilateral subgaleal platinum electrode. Electrode 1 of the strip electrode serves as ground. - Labtop with software
LabChart ADInstruments, New South Wales, Australia
BrainVision Recorder BrainProducts GmbH, Munich - Brain tissue oxygen monitoring system with intraparenchymal oxygen sensor
Licox, Integra Lifesciences Corporation, Plainsboro, NJ, USA - Powerlab 16/SP analog/digital converter
ADInstruments, New South Wales, Australia
This device samples data at 200 Hz. Recommendations for PowerLab connections are shown on the close-up view of the amplifier’s front panel. If any of the monopolar electrodes (2 or 5) break, they should be exchanged to allow calculation of as many monopolar channels as possible (e.g. if electrode 5 breaks, exchange it for electrode 6). If you are not recording Laser Doppler flowmetry (LDF), you may use those channels to connect other devices.
- Component Neuromonitoring System (CNS) Monitor
Moberg ICU Solutions
The CNS Monitor is a neuro-focused patient monitor for data acquisition, display, and integration that supports interoperability with multiple external devices. Depending on the site, many other things can be monitored in addition to ECoG and conventional EEG, such as ptiO2 or rCBF. The CNS monitor also synchronizes with the bedside physiology monitor that has other patient vitals such as electrocardiogram, arterial pressure, respiration, plethysmography, etc.
Analysis of the Electrocorticogram (ECoG)
The hallmark signature of spreading depolarization in the ECoG is a negative DC shift with sequential onset in adjacent electrodes. The raw DC can only be seen with DC-coupled amplifiers (see section above). AC-coupled amplifiers with a lower frequency limit of 0.01 or 0.02 Hz distort the cortical DC shift of spreading depolarization but a typical slow potential change is still visible. We can use the AC-recorded slow potential change to identify spreading depolarizations. However, only the unfiltered DC shift allows assessment of the local duration of spreading depolarization, and it is a true measure of tissue energy status and risk of injury. In the following section, the term DC shift also refers to the slow potential change in the AC-ECoG.
Spreading depolarization-induced spreading depression presents as a more or less rapidly developing, propagating reduction of the raw amplitude of spontaneous brain electrical activity in the 0.5-45 Hz band, or any derived measure based on amplitude. Review of the raw signal alongside a leaky integral of power of the bandpass filtered AC-ECoG most reliably shows this loss of amplitude.
ECoG signal conversion
1. Trace: Near-DC/AC ECoG (0.01-45 Hz) shows a spreading depolarization as a characteristic negative, slow potential change. (Negative DC shift is only seen with DC-coupled amplifiers, and is not shown here.)
2. Trace: 0.5-45 Hz bandpass-filtered ECoG shows spreading depolarization-induced spreading depression as an amplitude reduction.
Arithmetic calculation in LabChart:
bandpass(rch[No. of channel in raw recordings];0.5,45)
3. Trace: Squared 0.5-45 Hz bandpass-filtered ECoG (also called AC-ECoG power) provides better visualization of amplitude loss during spreading depression. Changes of power are less sensitive to artifacts than changes of the integral of power (see below).
Arithmetic calculation in LabChart:
bandpass(rch[No. of channel];0.5;45)^2
4. Trace: The integral of power is based on a method of computing time integrals over a sliding window according to a time decay function. This provides a smooth curve that facilitates the assessment of changes in the AC-ECoG power, i.e. the exact start and end of spreading depolarization-induced depression periods.
Arithmetic calculation in LabChart:
integrate(bandpass(rch[No. of channel];0.5;45)^2;”decay”;60)
Data analysis software such as LabChart (ADInstruments, Oxford, UK)provide various filtering and signal processing functions and allow multiple display views. DC shifts and depression durations can be observed in either monopolar (unipolar montage against a reference electrode) or bipolar recordings (each electrode subtracted from neighboring electrode). Monopolar recordings are superior to bipolar recordings when local information on individual electrodes is of interest, such as the duration of the DC shift. The bipolar montage provides more stable recordings in case of reference electrode loss due to patient movement or nursing procedures.
Basic analysis of baseline activity, spreading depolarizations and spreading depression, and ictal epileptiform events
For the basic analysis, it is at certain points necessary and/or helpful to convert the raw ECoG signal. In the following list you will find a step-by-step tutorial for the analysis, together with the arithmetic LabChart conversion codes needed for the individual tasks.
Data post-processing
Data post-processing provides summary measures that help to examine the relationship between recorded data and clinical events such as baseline injuries, interventions, lesion development (as assessed by neuroimaging), patient outcome, and late epilepsy. All summary measures should be normalized to the duration of valid recording.
- Determine date and time of the initial insult, and define days post injury: the first 24-hour period counts as ‘day 0’, the second 24-hour period as ‘day 1’ and so on.
- Calculate the total spreading depolarization-induced depression duration of each recording day (TDDD). This is the sums of the longest depression durations of all individual spreading depolarizations during each 24-hour period. If there is temporal overlap between depression periods of successive depolarizations at different electrodes, the overlapping period counts once. TDDD is defined as the value after normalization to the total time of valid recordings during that 24-hour period.
- Example: 240 minutes (=4 hours) of total depression time are recorded in 22 hours of a 24-hour period. Normalized TDDD is 262 minutes [(4/22) x 24 x 60].
- Determine the peak total spreading depolarization-induced depression duration per recording day (PTDDD). This is the longest TDDD among all recording days.
- Calculate the sum of all types of spreading depolarization (spreading depolarizations with spreading depression, isoelectric spreading depolarizations, and spreading convulsions) and peak numbers per recording day in analogous fashion.
- Assess isolated spreading depolarizations and clusters of spreading depolarizations: the current working definition of a cluster of spreading depolarizations is the occurrence of at least three spreading depolarizations within three or fewer consecutive hours of recording.
- Calculate the total duration of IEE of each day: the sums of the longest IEE durations among all channels during each 24-hour period is calculated. Determine the peak total duration of IEE among all recording days as per the PTDDD above. Calculate sum and peak number of IEEs per recording day in analogous fashion.
Troubleshooting
With a bit of training, the identification of spreading depolarizations is usually straightforward and it is quite easy to distinguish the characteristic waveforms from other confounding physiologic changes or artifacts that may present as similar waves. In individual patients, spreading depolarizations often show a stereotypical pattern which means that in cases of doubt, it is useful to screen all recordings from that patient to get used to the specific spreading depolarization pattern. Once that pattern is understood, it is easier to recognize all spreading depolarizations, and to score events that would normally be missed (e.g. spreading depolarizations co-occurring with signals caused by artifacts). In addition to that, typical responses of ptiO2 or rCBF to spreading depolarization – when recorded – are also useful. If significant doubts remain, it is recommended that questionable events should not be scored as spreading depolarizations to avoid false positives. In this section, you will find some examples of frequent artifacts.
Minimal requirements to diagnose spreading depolarization
- when using DC-coupled amplifier: characteristic DC shift with corresponding spreading depression of spontaneous activity in a single ECoG channel, OR
- nonsimultaneous DC shift / slow potential change recorded at two or more electrodes within 10 minutes of each other; in addition, spreading depression of activity in at least one channel, unless tissue is electrically inactive, OR
- similar pattern of DC shift / slow potential change as recorded at other times throughout the monitoring period of the same patient. Depression of activity or shape of DC shift / slow potential change may be more irregular or distorted by artifacts.
The following slide show will guide you through some of the most common troubleshooting scenarios.
-
This image shows a 20 minute section from the recording of a patient ECoG. We see a signal that resembles the slow potential change of a spreading depolarization at all six electrodes.
Why is this not a true spreading depolarization?
First, we see no signs of a propagation. The observed signal appears simultaneously at all recording electrodes. Secondly, there is no depression of spontaneous activity in the AC-ECoG traces.
What is it then?
This is probably just a noise signal.
-
This image shows a 30 minute section from the recording of a patient ECoG. We see a depression of the spontaneous activity in the AC-ECoG.
Why is this not a true spreading depression?
First, there is no sign of a spreading depolarization in the near-DC/AC-ECoG. Spreading depression cannot occur without spreading depolarization. Secondly, the onset depression occurs simultaneously in all channels, there is no sign of a propagation.
What is it then?
Drug administration (e.g. sedatives) sometime causes simultaneous depression of the spontaneous activity.
-
This is a 45 minute section from the recordings of a patient ECoG. We see the slow potential change of a spreading depolarization that is interrupted and distorted by noise. It is therefore difficult to identify the depolarization and its propagation pattern when looking at the near-DC/AC-ECoG alone.
The corresponding AC-ECoG traces show the depression of the spontaneous activity and help with the identification of spreading depolarization and its propagation.
Spreading depolarization and spreading depression always propagate in a similar fashion.
-
In these recordings, we see a scenario that looks like a spreading convulsion: a spreading depolarization with epileptic field potentials rising on the tail end of the DC shift. Instead of spreading depression, the AC-ECoG traces show a signal that resembles ictal epileptic activity.
Why is this not a true spreading convulsion?
The next slide shows close-ups of an unusual synchronicity between the bandpass-filtered AC-ECoG and blood pressure recordings.
What is it then? In isoelectric tissue, we sometimes see pulse artifacts that mimic spreading convulsions. Their cause is unknown. One hypothesis is that severe cytotoxic edema results in shrinkage of the extracellular space. Therefore, the decreased distance between neurons and capillaries facilitates mechanical artifacts.
-
