Clinical Neurophysiology 127 (2016) 33293334

Contents lists available at ScienceDirect

Clinical Neurophysiology
journal homepage: www.elsevier.com/locate/clinph

Burst firing of single neurons in the human medial temporal lobe

changes before epileptic seizures
Heidemarie Gast a,1, Johannes Niediek a,1, Kaspar Schindler b, Jan Bostrm c, Volker A. Coenen c,
Heinz Beck d, Christian E. Elger a, Florian Mormann a,
a
Cognitive and Clinical Neurophysiology, Department of Epileptology, University of Bonn, Bonn, Germany
b
Department of Neurology, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland
c
Stereotaxy and MR-Based OR Techniques, Department of Neurosurgery, University of Bonn, Bonn, Germany
d
Laboratory for Experimental Epileptology and Cognition Research, Department of Epileptology, University of Bonn, Bonn, Germany

a r t i c l e i n f o h i g h l i g h t s

Article history:
 Ipsilateral MTL neurons show an increased degree of burstiness during interictal recordings.
Accepted 15 August 2016
 This effect is found only for putative principal cells, not for interneurons.
Available online 20 August 2016
 Burstiness may serve as a protective factor and its absence may in turn facilitate seizure emergence.

Keywords:
Burstiness
Interictal a b s t r a c t
Pre-ictal
Interneurons
Objective: To better understand the mechanisms that lead to the sudden and unexpected occurrence of
Principal cells seizures, with the neuronal correlate being abnormally synchronous discharges that disrupt neuronal
function.
Methods: To address this problem, we recorded single neuron activity in epilepsy patients during the
transition to seizures to uncover specific changes of neuronal firing patterns. We focused particularly
on neurons repeatedly firing discrete groups of high-frequency action potentials (so called bursters) that
have been associated with ictogenesis. We analyzed a total of 459 single neurons and used the mean
autocorrelation time as a quantitative measure of burstiness. To unravel the intricate roles of excitation
and inhibition, we also examined differential contributions from putative principal cells and interneu-
rons.
Results: During interictal recordings, burstiness was significantly higher in the seizure onset hemisphere,
an effect found only for principal cells, but not for interneurons, and which disappeared before seizures.
Conclusion: These findings deviate from conventional views of ictogenesis that propose slowly-increasing
aggregates of bursting neurons which give rise to seizures once they reach a critical mass.
Significance: Instead our results are in line with recent hypotheses that bursting may represent a protec-
tive mechanism by preventing direct transmission of postsynaptic high-frequency oscillations.
2016 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights
reserved.

1. Introduction at a neuronal population level using EEG or local field potential

Epilepsy is not a disease in and of itself, but a chronic condition
of the brain characterized by paroxysmal and recurrent epileptic
events that occur as a result of chronic structural and/or functional
changes. A number of studies have investigated seizure initiation
Corresponding author at: Department of Epileptology, University of Bonn,
Sigmund-Freud-Strasse 25, 53105 Bonn, Germany. Fax: +49 228 287 19351.
E-mail address: florian.mormann@ukb.uni-bonn.de (F. Mormann).
1
These authors contributed equally to this work.
(LFP) recordings both in humans and animal models of epilepsy
(Szabo et al., 2015). However, it is not possible to extrapolate
meaningfully from this population level to the behavior of its cel-
lular elements, even though they underlie this activity. In order to
obtain mechanistic insight, it is therefore necessary to record cellu-
lar activity, in vivo, during seizure onsets. This technique has only
recently become available in humans (Bower et al., 2012;
Mormann and Jefferys, 2013 and references therein).
A specific neuronal discharge pattern termed bursting, which is
characterized by the repeated firing of discrete groups of

http://dx.doi.org/10.1016/j.clinph.2016.08.010
1388-2457/ 2016 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
3330 H. Gast et al. / Clinical Neurophysiology 127 (2016) 33293334
high-frequency action potentials, has been associated with chronic
epilepsy both in experimental models of epilepsy and in humans
(Wyler and Ward, 1986; Sanabria et al., 2001; Schindler et al.,
2006). To date only a few in vivo studies in humans have investi-
gated bursting and its relevance to the pathophysiology of epi-
lepsy. These few studies have yielded conflicting results. Earlier
studies compared seizure onset zones to contralateral brain
regions during seizure-free (interictal) periods and reported burst
discharges to be less frequent ipsilaterally (Colder et al., 1996). In
contrast, other studies found higher burst rates in seizure onset
zones in MTL epilepsy (MTLE) (Staba et al., 2002). Single cells
recorded in close proximity to the seizure onset zone tended to
have higher bursting rates if they showed pre-ictal and ictal
changes in spiking rates (Keller et al., 2010). Finally, burst inter-
spike interval ratios have been used to predict seizure onset zones
from interictal recordings (Valdez et al., 2013).
The relation between interictal epileptiform EEG activity and
neuronal bursting has also been investigated. Neurons whose firing
rates were modulated by interictal discharges had a significantly
higher bursting rate than neurons not showing this modulation
(Keller et al., 2010). Alarcn and colleagues reported an increased
occurrence of burst firing during spontaneous interictal epilepti-
form discharges and after single pulse electrical stimulation
(Alarcn et al., 2012).
The analysis methods employed by these studies, particularly
the definition of bursting, varied considerably, and most were car-
ried out on limited data sets. This might explain why no consistent
relationship between burst firing and localization of the seizure
onset zone has been demonstrated to date.
We here examined the role of bursting neurons in the emer-
gence of clinical seizures. We analyzed simultaneous recordings
from the MTLs ipsi- and contralateral to the seizure onset zone
in patients with MTLE both during the seizure-free interval and
immediately before seizures. To unravel the intricate and poten-
tially counterintuitive roles of excitation and inhibition, we also
studied the differential contributions of putative principal cells
and interneurons.
2. Methods
2.1. Patients
22 consecutive patients with pharmacologically intractable epi-
lepsy were included in this study. In these patients, non-invasive
methods had been unable to unequivocally localize a region of sei-
zure onset, and they chose to undergo intracranial long-term EEG
monitoring to be evaluated for epilepsy surgery (Rosenow and
Lders, 2001). All patients had given written informed consent to
participate in this study, which was approved by the Medical Insti-
tutional Review Board in Bonn. In 11 of the 22 patients, seizure
onset was bilateral, extratemporal, or could not be unequivocally
determined. Data from the remaining 11 patients were analyzed
in this study (Table 1).
2.2. Electrodes and recordings
LFPs were recorded with hybrid depth electrodes (AdTech,
Racine, WI) containing a bundle of eight microwires protruding
approximately 4 mm from their tips. The differential LFP signals
were amplified using a 256-channel Neuralynx ATLAS system
(Bozeman, MT), filtered between 0.1 and 9000 Hz, and sampled
at 32 kHz. Action potential (spike) detection and sorting was per-
formed using the WaveClus package (Quiroga et al., 2004). Sorted
units were manually confirmed and classified as single units
(SU), multi-units, or artifacts based on spike shape and variance,
peak-amplitude-to-noise level, the inter-spike interval (ISI)
distribution of each cluster, and presence of a refractory period.
Multi-units and artifacts were excluded from further analysis. A
total of 249 SU were recorded interictally and 210 pre-ictally
(mean number per patient 23 interictally and 30 pre-ictally).
Fig. 1 shows an example of an intracranial LFP recorded with a
microwire capturing two different SU.
Electrode locations were determined by clinical criteria and ver-
ified by MRI and CT. All patients had bilaterally implanted depth
electrodes, each equipped with 8 cylindrical contacts, in multiple
sites of the MTL, including hippocampus, amygdala, entorhinal cor-
tex and parahippocampal cortex. Total recording time analyzed
from these patient (interictal and pre-ictal) was 300 min (see
Table 1).
Recordings were performed continuously for the entire dura-
tion of the epilepsy monitoring (typically 714 days). Eleven inter-
ictal sessions were recorded one or two days after electrode
implantation. In 8 of the 11 patients, a total of 21 seizures were
recorded and analyzed. All seizures occurred at least one day after
the interictal recordings. Pre-ictal recordings covered the period of
10 min before seizure onset as determined from the intracranial
macro electrode recordings via joint visual inspection by two
board-certified electroencephalographers (H.G. and F.M.). Electro-
clinical evaluation of interictal and ictal recordings confirmed that
seizures typically started near the most medial contacts of the clin-
ical depth electrodes (Supplementary Table S1).
Based on electrographical seizure onset, or postoperative sei-
zure control in those cases where no seizure could be recorded,
electrodes were labeled ipsilateral or contralateral, depending on
their location with respect to the seizure onset and/or resection
zone. Note that not all of the 11 patients included in this study
went on to have surgery, and of those who did, not all became
seizure-free (Table 1). Nevertheless, at the time of electro-clinical
evaluation all 11 patients were diagnosed as unilateral MTLE, and
therefore included in this study.
2.3. Analysis of electrophysiological data
To assess the neurons spontaneous firing patterns, we com-
puted autocorrelation histograms with 1 ms bins and time lags
ranging from 100 to 100 ms (Fig. 1C). To quantify burstiness,
i.e. the firing of dense clusters of action potentials, we calculated
the mean autocorrelation time (MAT) for every SU. MAT denotes
the mean of each neurons autocorrelogram (Csicsvari et al.,
1998). Low MAT values thus characterize high burstiness, regard-
less of individual firing rates (Supplementary Fig. S1).
Following Ison and coworkers (Ison et al., 2011), we classified
all SU based on their action potential duration (spike width, mea-
sured from maximum to minimum, Fig. 1B), as either putative
interneurons (width 6 0.65 ms) or putative principal cells
(width > 0.65 ms).
Two-way ANOVA was used as an omnibus test to check for any
statistical difference in burstiness (as characterized by MAT) with
respect to location (ipsi- or contralateral) and time of recording
(interictal or pre-ictal). Post hoc analysis between groups was per-
formed using Wilcoxon ranksum tests with Bonferroni correction
for multiple (four) comparisons.
3. Results
Classification of SU as either principal cells or interneurons
based on spike width yielded a total of 415 principal cells and 44
interneurons. The plausibility of this classification was verified by
comparing mean baseline firing rate and burstiness using a
Wilcoxon ranksum test. Since it is a priori unclear how these
 H. Gast et al. / Clinical Neurophysiology 127 (2016) 33293334 3331

Table 1
Summary of clinical data. H.S.: hippocampal sclerosis; SAH: selective amygdalo-hippocampectomy. Seizure outcome according to Engel classification.

Pat. No. Gender Age MRI Resection Histo-pathological Seizure Number Total pre-ictal Number of Seizure
(years) findings findings outcome of time micro- onset
seizures analyzed wires hemisphere
(min) in MTL
1 m 37 Normal Left SAH Left H.S. 1A 3 30 64 Left
2 f 26 Normal No surgery n/a n/a 1 10 80 Left
3 f 56 Left H.S. No surgery n/a n/a 0 0 64 Left
4 f 43 Normal Left SAH Focal loss of neurons 3 4 30 80 Left
5 f 48 Normal Right SAH Focal loss of neurons 1B 0 0 80 Right
6 f 29 Normal No surgery n/a n/a 5 50 88 Right
7 f 38 Right H.S. Right SAH Right H.S. 1A 0 0 64 Right
8 m 33 Normal No surgery n/a n/a 2 20 80 Left
9 m 44 Normal Left SAH + Cell loss, gliosis, heterotopia. 1A 2 20 72 Left
lesionectomy
10 m 54 Right H.S. Right SAH Right H.S. 3 1 10 64 Right
11 f 42 Normal Right SAH Gliosis 1A 3 20 64 Right

regions ipsilateral to the seizure onset zone than in contralateral

Fig. 1. Quantifying burstiness in extracellular recordings from putative principal
cells and interneurons (see Methods). Blue and red colors represent a putative
principal cell and interneuron, respectively. (A) 80 ms of unfiltered signal recorded
from a microwire in the left anterior hippocampus of patient #10 (see Table 1) with
action potentials from two different single units marked by blue and red vertical
lines, respectively. (B) Action potentials (spikes) recorded from these units during a
10-min epoch. White and black lines represent mean SD. Green lines denote the
spike width (1.07 ms and 0.25 ms for putative principal cell (blue) and interneuron
(red), respectively). (C) Autocorrelograms, smoothed with a three-point moving
average for graphical display. Vertical green lines denote mean autocorrelation
times (19 ms and 49 ms for putative principal cell (blue) and interneuron (red),
respectively). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
regions during the interictal period (p = 0.0009), but not during
the pre-ictal period (Fig. 2A).
Performing the same analysis separately for principal cells and
interneurons, we observed again a significant main effect of loca-
tion for principal cells (F1455 = 7.05; p = 0.008), with post hoc tests
showing increased burstiness of the ipsi- vs. contralateral side dur-
ing the interictal period only (p = 0.009; Fig. 2B). Again, neither a
significant main effect of time (F1455 = 0.45; p = 0.50) nor a signifi-
cant interaction between time and location (F1455 = 1.78; p = 0.18)
were observed. For interneurons we found no significant main
effect of location (F1455 = 1.09; p = 0.30), but a weakly significant
main effect of time (F1455 = 6.03; p = 0.019) with no significant
interaction between time and location (F1455 = 1.86; p = 0.18). This
effect was not confirmed in individual post hoc tests for the ipsi- or
contralateral hemisphere (Fig. 2C).
To test whether the observed difference in statistics between
principal cells and interneurons was caused by a difference in pro-
portion of these cell types, we compared the ratios of principal
cells to interneurons recorded in interictal vs. pre-ictal recordings
using Fishers exact test. No significant difference was found in
either the ipsi- or contralateral hemisphere (Supplementary
Fig. S3). To furthermore rule out an influence of firing rate under-
lying these results, we ran the same two-way ANOVA as above
using the average firing rates of each neuron as dependent vari-
able. No significant main effect of time or location and no signifi-
cant interaction were observed (Supplementary Fig. S4). Likewise,
the time courses of putative principal cells and interneurons dur-
ing interictal and pre-ictal periods, ipsi- and contralaterally,
showed no dynamical trends, particularly not upon approaching
the seizure onset (Supplementary Fig. S5).

dynamic features might change just before a seizure, we restricted
this analysis to interictally recorded neurons (Supplementary
Fig. S2A). Interneurons showed higher firing activity than principal
cells (p = 0.012, Wilcoxon ranksum test, Supplementary Fig. S2B)
while principal cells exhibited a stronger tendency for burstiness
(p < 10 4, Supplementary Fig. S2C).
As a first exploratory analysis, we applied a two-way ANOVA
with factors time (interictal vs. pre-ictal) and location (ipsi- vs.
contralateral) to all 459 recorded neurons. Results yielded a signif-
icant main effect of location on burstiness (F1455 = 12.0;
p = 0.0006). Neither a significant main effect of time (F1455 = 0.09;
p = 0.77) nor a significant interaction between time and location
(F1455 = 1.72; p = 0.19) were observed. Post hoc Wilcoxon ranksum
tests with Bonferroni correction confirmed higher burstiness in
4. Discussion
A fundamental question in epileptology is what mechanisms
are responsible for changing the state of the epileptic brain from
interictal to ictal. We here investigated this question at the cellular
level by assessing the role of burst firing by single neurons. We
analyzed interictal and pre-ictal recordings from regions ipsi-
and contralateral to the seizure onset zone in the MTLs of 11
patients. To investigate the roles of excitation and inhibition, we
examined the differential contributions of putative principal cells
and interneurons. We found that burst firing was higher in the ipsi-
lateral MTL during interictal recordings. Interestingly, the signifi-
cant difference in burst firing between the ipsi- and contralateral
MTL was found exclusively in principal cells, not in interneurons,
and disappeared in pre-ictal recordings.
3332 H. Gast et al. / Clinical Neurophysiology 127 (2016) 33293334

Fig. 2. Burstiness of different cell types as quantified by the mean autocorrelation time (MAT). (A) MAT histograms of all recorded cells for the conditions interictal/ipsilateral
(top left, red), interictal/contralateral (bottom left, green), pre-ictal/ipsilateral (top right, black), and pre-ictal/contralateral (bottom right, blue), flanked by results of posthoc
Wilcoxon ranksum tests after Bonferroni correction. (B) Same as (A), but for putative principal cells only. (C) Same as (A), but for putative interneurons only. N, number of
neurons; int-ict, interictal; pre-ict, pre-ictal; n.s., not significant; p < 0.05; p < 0.01; p < 0.001. (For interpretation of the references to color in this figure legend, the reader
is referred to the web version of this article.)
 H. Gast et al. / Clinical Neurophysiology 127 (2016) 33293334
Increased burst firing in the ipsilateral MTL during interictal
recordings agrees with previous observations (Staba et al., 2002).
These findings deviate from an older study in which authors
describe a weakly significant overall reduction in interictal burst
discharges in ipsilateral MTL structures (Colder et al., 1996). This
study, however, used the area under the first significant peak in
the autocorrelogram within the first 150 ms as a measure of bursti-
ness. This measure is strongly biased by the average firing rate and
will pick up any cell with preferred ISIs up to 150 ms, many of
which would not be considered bursty in a stricter sense. Thus,
the reported difference may merely be an artifact of firing rate,
which was indeed reported to be higher in the contralateral MTL
(Colder et al., 1996).
How can a pre-ictal reduction of bursting be interpreted in
terms of a transition from the interictal to an ictal brain state?
The fact that interictal burst firing was significantly more pro-
nounced ipsilaterally but then decreased and was no longer differ-
ent from the contralateral hemisphere in the minutes before
seizure onset contradicts traditional views of ictogenesis. Accord-
ing to these views, excitation, e.g. in the form of bursting principal
cells, should steadily build up until a critical mass is reached,
which then leads to uncontrolled activity in the form of a seizure
that dissipates and resets the accumulated excitation (Lennox,
1941). Instead, our study demonstrates that the excess burstiness
in the ipsilateral MTL is significantly reduced during the transition
to seizure. Could differences in pre-ictal synaptic input dynamics
lead to reduced bursting? A characterization of input patterns that
lead to synaptically driven bursting has shown that fast oscillations
of input currents >100 Hz are most likely to generate bursting
(Schindler et al., 2006). Thus, any change in spectral patterns of
input that transitions out of this range might cause reduced burst-
ing. This does not preclude that inputs may become more syn-
chronous in the pre-ictal period, giving rise to highly structured,
regular firing as a precursor to seizures.
Bursting could thus help to dissipate fast oscillations by pre-
venting their direct transmission and thereby reflect a seizure-
protective mechanism. Increased interictal occurrence of high-
frequency oscillations (HFOs) has been shown to be highly specific
for the seizure onset zone in patients with MTLE (Haegelen et al.,
2013). However, their occurrence rate does not change systemati-
cally before seizures (Jacobs et al., 2009), and recent evidence sug-
gests that underlying mechanisms of interictal HFOs differ from
those of high-frequency activity during seizure onset (de Curtis
and Avoli, 2016). One may therefore speculate that the pre-ictal
decrease in bursting in the ipsilateral MTL may play a causal role
in the generation of seizures.
Note that for most patients, anti-epileptic drugs (AEDs) were
gradually tapered off in order to increase the likelihood of seizure
occurrence a procedure that is standard practice for epilepsy
monitoring. AED blood levels were therefore in general lower dur-
ing pre-ictal recordings than during interictal recordings. Reduced
AED levels, however, would be expected to result in an increased
rather than a decreased burstiness as observed in our study
(Chen et al., 2011; Uebachs et al., 2012). It therefore appears unli-
kely that the changes in neuronal dynamics found in this study are
merely attributable to AED blood levels per se, but instead are more
likely to reflect general mechanisms of ictogenesis.
As our recordings continue, we will be able to collect a succes-
sively larger number of pairs of interneurons/principal cells
recorded from the same microwire bundle. This will allow us to
scrutinize their interaction and shed light on the intricate interplay
between these inhibitory and excitatory groups of neurons before
seizures. It will furthermore provide us with the necessary statisti-
cal power to individually assess the contributions of different MTL
subregions to the changes described in this study.
3333
The fact that we found interictal burst firing to be significantly
more pronounced in the focal MTL may bear clinical relevance
since it could provide diagnostically important information even
before an actual seizure has occurred. Gaining all clinically relevant
information from interictal recordings has been a long-standing
goal in presurgical evaluation because this could potentially reduce
the number of seizures to be recorded. A smaller number of sei-
zures reduces the duration of intracranial monitoring and thus
decreases the risk of complications such as infections.
In conclusion, our study demonstrates that seizures in MTLE are
preceded by altered burstiness of principal cells with respect to
interictal baseline recordings in the ipsilateral MTL. Rather than
supporting conventional views of ictogenesis that propose
slowly-increasing aggregates of bursting neurons, our findings
are in line with recent hypotheses that bursting may represent a
protective mechanism by preventing direct transmission of postsy-
naptic high-frequency oscillations. Future studies investigating the
interplay between interneurons and principal cells as well as the
correlation of neuronal activity with LFP events such as epileptic
spikes and HFOs might offer possibilities to entertain alternative,
non-canonical mechanistic models for seizure initiation in human
epilepsy.
Acknowledgements
We thank Christopher John for help with the clinical data. This
research was supported by grants from the Volkswagen
Foundation (Lichtenberg Program), the German Research Council
(SFB 1089 and MO 930/4-1), and the European Commission (FP7
602102 EPITARGET).
Conflict of interest: None.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.clinph.2016.08.
010.
References
Alarcn G, Martinez J, Kerai SV, Lacruz ME, Quiroga RQ, Selway RP, et al. In vivo
neuronal firing patterns during human epileptiform discharges replicated by
electrical stimulation. Clin Neurophysiol 2012;123:173644.
Bower MR, Stead M, Meyer FB, Marsh WR, Worrell GA. Spatiotemporal neuronal
correlates of seizure generation in focal epilepsy. Epilepsia 2012;53:80716.
Chen S, Su H, Yue C, Remy S, Royeck M, Sochivko D, et al. An increase in persistent
sodium current contributes to intrinsic neuronal bursting after status
epilepticus. J Neurophysiol 2011;105:11729.
Colder BW, Frysinger RC, Wilson CL, Harper RM, Engel J. Decreased neuronal burst
discharge near site of seizure onset in epileptic human temporal lobes. Epilepsia
1996;37:11321.
Csicsvari J, Hirase H, Czurko A, Buzski G. Reliability and state dependence of
pyramidal cell-interneuron synapses in the hippocampus: an ensemble
approach in the behaving rat. Neuron 1998;21:17989.
de Curtis M, Avoli M. GABAergic networks jump-start focal seizures. Epilepsia
2016;57:67987.
Haegelen C, Perucca P, Chtillon C-E, Andrade-Valena L, Zelmann R, Jacobs J, et al.
High-frequency oscillations, extent of surgical resection, and surgical outcome
in drug-resistant focal epilepsy. Epilepsia 2013;54:84857.
Ison MJ, Mormann F, Cerf M, Koch C, Fried I, Quiroga RQ. Selectivity of pyramidal
cells and interneurons in the human medial temporal lobe. J Neurophysiol
2011;106:171321.
Jacobs J, Zelmann R, Jirsch J, Chander R, Dubeau C-ECF, Gotman J. High frequency
oscillations (80500 Hz) in the preictal period in patients with focal seizures.
Epilepsia 2009;50:178092.
Keller CJ, Truccolo W, Gale JT, Eskandar E, Thesen T, Carlson C, et al. Heterogeneous
neuronal firing patterns during interictal epileptiform discharges in the human
cortex. Brain 2010;133:166881.
Lennox WG. The causes of seizures. Science and seizures. New York: Harper; 1941.
Mormann F, Jefferys JGR. Neuronal firing in human epileptic cortex: the ins and outs
of synchrony during seizures. Epilepsy Curr 2013;13:1002.
3334 H. Gast et al. / Clinical Neurophysiology 127 (2016) 33293334
Quiroga RQ, Nadasdy Z, Ben-Shaul Y. Unsupervised spike detection and sorting with
wavelets and superparamagnetic clustering. Neural Comput 2004;16:166187.
Rosenow F, Lders H. Presurgical evaluation of epilepsy. Brain 2001;124:1683700.
Sanabria ER, Su H, Yaari Y. Initiation of network bursts by Ca2+-dependent intrinsic
bursting in the rat pilocarpine model of temporal lobe epilepsy. J Physiol (Lond)
2001;532:20516.
Schindler KA, Goodman PH, Wieser HG, Douglas RJ. Fast oscillations trigger bursts of
action potentials in neocortical neurons in vitro: a quasi-white-noise analysis
study. Brain Res 2006;1110:20110.
Staba RJ, Wilson CL, Bragin A, Fried I, Engel J. Sleep states differentiate single neuron
activity recorded from human epileptic hippocampus, entorhinal cortex, and
subiculum. J Neurosci 2002;22:5694704.
Szabo GG, Schneider CJ, Soltesz I. Resolution revolution: epilepsy dynamics at the
microscale. Curr Opin Neurobiol 2015;31:23943.
Uebachs M, Albus C, Opitz T, Isom L, Niespodziany I, Wolff C, et al. Loss of b1
accessory Na+ channel subunits causes failure of carbamazepine, but not of
lacosamide, in blocking high-frequency firing via differential effects on
persistent Na+ currents. Epilepsia 2012;53:195967.
Valdez AB, Hickman EN, Treiman DM, Smith KA, Steinmetz PN. A statistical method
for predicting seizure onset zones from human single-neuron recordings. J
Neural Eng 2013;10:16001.
Wyler AR, Ward AA. Neuronal firing patterns from epileptogenic foci of monkey and
human. Adv Neurol 1986;44:96789.