Spatiotemporal dynamics of direct cortical response depression during spreading depolarizations

Ethical approval

Animal care and procedures were in accordance with EU Directive 2010/63/EU for animal experiments, and all animal-use protocols were reviewed and approved by the Local Ethical Committee of Kazan Federal University (#24/22.09.2020).

Animal preparation

Experiments were designed, conducted, and reported in accordance with the ARRIVE guidelines. Wistar rats of both sexes aged from 4 to 8 weeks were used. Animals were prepared under isoflurane anesthesia at the surgical level (4% for induction, 2% for maintenance, Aerrane (Baxter, UK)). The skin and periosteum above both hemispheres and cerebellum were removed. The neck muscles were detached from the occipital bone, then the wound was treated with bupivacaine (0.25%) for local anesthesia. The skull surface was dried and covered with cyanoacrylate glue. A metal bar was attached to the skull along the sagittal suture using cyanoacrylate glue and dental cement (Meliodent, Heraeus Kulzer, Germany). At the same time, isoflurane anesthesia was gradually reduced, and the animal was administered urethane (Sigma, USA) at the surgical level of anesthesia (1.5 g/kg, i.p.). Surgical depth of anesthesia was confirmed by immobility and negative toe pinch reflexes throughout the experiment. Body temperature was maintained at 37 °C using a heated platform (TC-344B; Warner Instruments, Hamden, CT, USA). Of note, urethane may cause moderate metabolic acidosis with partial respiratory compensation and depression of the cardiocirculatory system. ²⁵ However, arterial blood gases and arterial pressure were not monitored, which represents a limitation of the study. The metal bar was attached to a fixation system via a ball joint. A craniotomy of 4 × 5 mm, centered over the left parietal cortex, was performed to accommodate the ECoG array. To prevent herniation, the fourth ventricle was then punctured and drained. The dura mater was incised with a 30 G needle. At the incision site, the dura was carefully lifted and separated from the pial surface using fine forceps, then trimmed along the perimeter of the cranial window with microscissors, taking care to avoid damaging the underlying cortex and blood vessels. A second cranial window of ~0.3 mm diameter for KCl application was drilled over the primary motor cortex (~1–2 mm rostral and ~3–4 mm lateral from bregma) and the 1–2 mm high dental cement wall was made around the cranial window. Overall, the animal preparation and recordings lasted for 6–8 h. No animals died during animal preparation and recordings.

Electrophysiological recordings

ECoG recordings of the local field potential (LFP) from the cortical surface were performed using custom ECoG (hereafter ECoG) electrode arrays (6 × 10 chromium-gold 50 µm in diameter electrode grids on a polyimide film base with 0.4 mm separation distance between the electrodes and pre-fabricated holes of 0.2 mm diameter ²⁶ ) placed epipially on the parietal cortex and covered with a liquid light paraffin (Panreac AppliChem, Spain). ECoG arrays were oriented with their long axis along the sagittal suture, and the rostral-distal corner of the array was positioned approximately 1 mm caudal and 6 mm lateral to bregma. Intracortical LFP and multiple unit activity (MUA) recordings were performed using a linear multichannel silicon probe with iridium electrodes: 413 μm² surface area, 100 μm separation distance (А1х16-5-100-413-А16; Neuronexus Technologies, USA). The probe was inserted vertically into the barrel cortex through a hole in the ECoG film at coordinates approximately 2 mm caudal and 5 mm lateral to bregma, to a depth of 1.6–1.8 mm. The signals from ECoG and intracortical electrodes were amplified, low-pass filtered at 9 kHz and digitized at 32 kHz using a DigitalLynx SX amplifier and Cheetah 6.3.2 software (Neuralynx, USA). Recordings were performed using full-band recordings with inverse filtering for signal reconstruction based on a hybrid AC/DC-divider filter. ²⁷

Direct cortical responses (DCRs) were evoked by electrical stimuli (20–50 V square pulse, 50 µs duration, 3 s inter stimulus interval) delivered through tungsten wire bipolar electrodes of 0.5–1 MOhm resistance (Microelectrodes Ltd., UK) placed either within or at different positions outside the ECoG array at depth of 100–200 µm from the cortical surface. Somatosensory evoked potentials (SEPs) were evoked by brief (2–5 ms) principal whisker (PW) deflection as described earlier. ²⁸ PW was identified by the shortest latency MUA response and maximal LFP amplitude in the L4. PW stimulation was performed at 1 s delay after stimulus evoking DCR.

SDs were induced by epidural KCl (0.5 M) application on the frontal cortex. After the successful SD induction confirmed by a characteristic DC potential shift across the ECoG array, the KCl was immediately washed out three times with physiological saline. Intervals between successive KCl-induced SDs were at least 15 min. Typically, four SDs were induced per experiment. Fifteen min after the final KCl-induced SD, an additional SD was triggered using a pinprick by fast 26 G needle insertion into the cortex in the same craniotomy over the frontal cortex. Because both SD induction models provided similar results, the data from high-potassium and pinprick-induced SDs were pooled together.

Data analysis

Raw data were preprocessed using a custom-developed suite of programs in the MATLAB environment. Positive polarity is graphed as up throughout all manuscript. The original DC signal was downsampled to 1 kHz and then used to analyze the LFP signal. SD onset was determined from the peak of the first derivative (SD’) of the 1 Hz lowpass filtered LFP signal during the initial SD depolarization phase. ²⁹ DCR N1 was detected as a trough in a time window 3–30 ms after the stimulus. Because stimulus artifacts interfered with the response, these were corrected by subtraction of blank responses during maximal SD-induced DCR depression during DCR analysis. DCR N1 peak amplitude was calculated from a line connecting positive peaks P1 and P2 within 2–40 ms time window after the stimulus. Sites with control values of DCR N1 amplitude less than 0.5 mV were excluded from analysis of SD-induced DCR depression. Current source density analysis of DCRs and SEPs during intracortical recordings was performed using inverse CSD method. ³⁰ Color-coded maps of DCRs, CSD and MUA were generated using linear interpolation. The SD propagation vector across the ECoG array was calculated as the average vector of the delays in SD-related DC shifts, and the DCR onset and offset times between pairs of electrodes at an angle to the long axis of the ECoG array in the rostral-caudal direction.

For the extracellular multiple unit spike detection, the original wide-band signal was filtered (300–4000 Hz), and negative local peaks >4 standard deviations of the quietest 100 s fragment in the control period were considered as spikes. The MUA frequency during DCRs and SEPs was calculated within 5–14 and 8–30 ms intervals following stimulus, respectively. The total MUA frequency was calculated as the sum of MUA frequencies across all 16 channels of the probe.

Spontaneous activity was assessed within a time window −900 to −10 ms before the DCR-stimulus. Power of delta oscillations was calculated in the 0.5–4 Hz frequency band.

Statistical analysis

Statistical analysis was performed using the MATLAB Statistics toolbox. Wilcoxon rank sum and signed-rank two-sided tests were used to assess the significance of differences between samples. The Pearson correlation coefficient (R) was used to measure the strength and direction of the linear relationship between two continuous variables. The level of significance was kept at p < 0.05. Pooled data are presented as median, 25th (Q1) and 75th (Q3) percentiles, in the form of boxplots or violin plots.

Spatial–temporal features of the direct cortical responses

Under control conditions, DCRs were characterized by distinct horizontal and vertical depth profiles. On ECoG electrodes, the most prominent negative N1 component of the DCR was largest at electrodes 1–2 mm from the stimulation site, where its amplitude reached 3.6 (2.6–4.2) mV (median (Q1–Q3); Figure 1(b); n = 15 animals). Hereafter, DCR descriptions will be based on the features of the N1 component. As shown in example recordings from one row of the ECoG array in Figure 1(c), DCR amplitudes decreased with distance from the stimulation site, which was also evident across the entire ECoG array (Figure 1(d)). Of note, DCRs were smaller in amplitude in close vicinity (<1 mm) to the stimulation site, which may be due to the activation of local inhibition^31,32 (Figure 1(c) and (d)). At the group level, the DCR amplitude decreased from its maximum at a rate of 0.8 (0.5–1.2) mV/mm (n = 13) with a distance from the stimulation site (Figure 1(e)). Along with the reduction in amplitude, the latency of the DCR N1 peak also progressively increased with distance from the stimulation site (Figure 1(f)–(h)). The rate of horizontal DCR propagation along the ECoG array was 0.28 (0.20–0.33) m/s (n = 13), which is compatible with the conduction velocity via non-myelinated cortico-cortical axonal fibers. ³³ During intracortical silicon probe recordings, DCRs at the top electrodes near the cortical surface were identical to the responses at nearby ECoG electrodes (Figure 1(i)). Across cortical depth, the N1 component of DCRs showed maximal amplitude in the superficial cortical layers 1/2, where, according to current-source density analysis, its main sink was also located; another prominent N1 sink was observed in cortical layer 5 (Figure 1(i)). These findings suggest that DCRs involve the direct activation of cortico-cortical excitatory synaptic inputs, primarily in superficial and deep cortical layers, but do not involve activation of excitatory inputs to the granular cortical layer 4, which is the main target of thalamo-cortical inputs. To verify this hypothesis, we compared DCRs with sensory-evoked responses, which are driven in the barrel cortex by thalamo-cortical inputs. Consistent with previous studies, ³⁴ brief deflection of the principal whisker elicited a typical sensory evoked potential (SEP) in the corresponding cortical barrel column, with main sinks in the thalamorecipient layer 4 and layers 5/6 border, and delayed sink in the superficial layers 2/3, supported by L4 to L2/3 connections Figure 1(i)). Consistent with these observations, DCRs were associated with maximal activation of units in L1/2 and L5, whereas SEPs caused the most of excitation in L4 and L5/6 border (Figure 1(j)). Taken together, these findings point two different networks activated during DCRs and sensory-evoked responses, which primarily involve cortico-cortical and thalamo-cortical circuits, respectively.

SD induces depression of DCRs on ECoG electrodes

We further explored changes in DCRs during SDs evoked by remote, local epidural application of a high-potassium solution or by a pinprick in the frontal cortex. SDs propagated through the ECoG array from the induction side at a rate of 5.7 (4.9–6.2) mm/min (Figure 2(a) and (b)). SDs were associated with a wave of negative DC voltage shift attaining 19.3 (15.9–21.8) mV (n = 13 rats), and a suppression of spontaneous cortical activity notably in the delta frequency range (Figure 2(c)), that is, similar to previous SD descriptions with ECoG recordings in rodents.^35–38 DCRs were transiently suppressed during SD alongside the depression of spontaneous activity, as shown in recordings from a single ECoG site (Figure 2(d)). For further assessments of the dynamics of this depression, we measured the delay of depression onset and offset at a 10% threshold of the control values, referenced to the SD onset, and the total duration of depression as indicated on the normalized DCR amplitude time plot in Figure 2(d). Notably, in the experiments shown in Figure 2, stimuli evoking DCRs were applied to the cortex downstream of the SD propagation trajectory (i.e. SD propagated from left to right, with the stimulation site located on the right side of the ECoG array).

OPEN IN VIEWER

Figure 2.

Depression of ECoG DCRs during SD: (a) color-coded isochrones of the SD front evoked by distal KCl application on the left side of the ECoG array and propagating across the cortex below the ECoG array. Vectors indicate the instantaneous direction and velocity of SD propagation, (b) group data on the velocity of SD propagation through the ECoG array (pooled data form n = 13 rats), (c, top) DC-ECoG LFP recordings of SD on electrodes b3, f3, and j3 (see panel (a)) and (bottom) corresponding AC-ECoG LFP bandpass-filtered at 1–50 Hz, and (d, top to bottom) DC-ECoG LFP recordings of SD on ECoG electrode h5; DCRs evoked by electrical stimulation on the right side of the ECoG array (as indicated by the black dot in panel (a)); corresponding DCR N1 amplitude normalized to the control values; and the algorithm for determining the depression onset and offset times relative to SD onset and the total depression duration.

SD-induced depression of DCR, SEP, and spontaneous activity: Intracortical recordings

We next compared the dynamics of SD-induced depression of DCRs, SEPs, spontaneous activity, and associated neuronal firing during intracortical silicon probe recordings. SD induced full suppression of DCRs through the entire cortical depth, including suppression of LFP signals and MUA. Similarly to DCRs, SEPs and spontaneous activity were also fully suppressed during SD in all layers (Figure 3(a)). SD displayed top-down propagation pattern (Figure 3(b)). Therefore, to compare the dynamics of various depression parameters during SD, we pooled data obtained from the surface layers, including all channels located in layers L1–L4 where the SD onsets did not differ significantly from the L1 SD onset (Figure 3(b)). The L1 SD onset was chosen for further comparisons with the ECoG data. DCRs and the delta power of spontaneous activity were sampled from L1, SEPs from L4, and MUA from all layers from L1 to L4. As illustrated by the group data presented in Figure 3(a) and the corresponding statistical comparisons in Figure 3(c), depression of DCRs, SEPs and spontaneous activity occurred close to the SD onset. However, the depression of activity lasted longer than the negative DC shift associated with the SDs. Overall, DCRs and SEPs recovered faster after SD compared to delta activity and spontaneous MUA (Figure 3(a) and (c)). We also noted that while the depression of spontaneous activity and SEPs is tightly locked to the SD, the dynamics of DCR depression varied depending on the position of the stimulating electrode and the intracortical probe along the SD trajectory. This dependency was explored in further detail during ECoG recordings.

OPEN IN VIEWER

Figure 3.

SD-induced depression of DCRs, SEPs, and spontaneous activity during intracortical recordings: (a, top to bottom) average SD during L1 DC-LFP recordings from the top channels of the silicon probe in the barrel cortex. SDs were averaged with reference to the SD onset (red circle and red vertical dashed line). The shaded area indicates the Q1–Q3 interval. Below are the L1 DCR and L4 SEP amplitude, L1 power in the delta-band frequency (0.5–4 Hz), integral MUA frequency in L1–4 during DCR and SEP in the time intervals indicated in the Figure 1(j), and spontaneous L1–4 MUA frequency (note pre-SD excitation at the SD onset^39,54), all normalized to control (pre-SD) values. Insets show examples of DCRs, SEPs and episodes of spontaneous activity at different times relative to the SD onset, (b) SD onset delays at different cortical depths relative to the SD onset in L1 (vertical dashed red line), and (c) depression onset and offset times (at 10% threshold) relative to the L1 SD, and total depression duration. SD LFP depression offset refers to the recovery of the negative DC shift of L1 LFP to 20% of its maximal amplitude. Boxplots show the median and Q1–Q3. (a–c) 47 SDs from 13 rats.

DCR depression depends on the position of the stimulating and recording electrodes along the SD trajectory

We further explored how the spatiotemporal dynamics of SD-induced DCR depression depend on the position of the stimulating and recording electrodes along the SD trajectory during ECoG recordings. In experiments where the stimulating electrode was placed downstream of the SD propagation relative to the ECoG array, the DCR depression onset coincided with the SD at the recording site. Conversely, in experiments where the stimulating electrode was placed upstream, DCR depression began well before the SD arrived at the recording site (Figure 4(a) and (c)). We reasoned that in the latter case, the early DCR depression is caused by the SD inhibiting neurons and neural fibers at the stimulation site. To estimate the arrival time of the SD at the stimulating site, we first approximated the progression of the SD onset across ECoG electrodes. A linear fit of this progression then provided an estimate of the SD onset at the stimulating site based on its position along the SD trajectory (Figure 4(b)). This estimated SD onset at the stimulating electrode matched the early DCR depression onset that preceded the SD at the recording electrode in cases where the stimulating electrode was positioned proximal to the SD initiation site (Figure 4(c), top panel). We next grouped SDs with comparable delays of the SD at the stimulating electrodes relative to the SD at the ECoG electrodes (−20, 5, and 28 s). We found that DCR depression indeed systematically began before the SD arrived at the recording electrode when the SD reached the stimulating electrode first (Figure 4(d), SD-proximal stimulation). In cases where the SD arrived at the stimulating electrode near-simultaneously with SD at the recording electrode (Figure 4(d), SD-iso stimulation), or after SD at the recording electrode (Figure 4(d), SD-distal stimulation), DCRs were only depressed when the SD arrived at the recording electrode. Also, DCR depression lasted longer in the cases of SD-proximal and SD-distal stimulation compared to SD-iso stimulation (Figure 4(d)). Notably, delta activity was suppressed upon the SD’s arrival at the recording site, regardless of the stimulating electrode’s position (Figure 4(d)).

OPEN IN VIEWER

Figure 4.

Dependence of DCR depression on the position of the stimulating electrodes along the SD trajectory: (a) scheme of ECoG recordings of DCRs evoked by stimulating electrodes placed at different sites along the SD trajectory (SD propagated from left to right through the ECoG array): upstream, SD-proximal stimulation site (left of the ECoG array) and downstream, SD-distal stimulation site (right of the ECoG array), (b) the SD onset at the stimulation site (green point) was estimated from the linear approximation (red line) of the SD onsets at ECoG electrodes along the SD trajectory (black circles), (c) DC-ECoG LFP at electrodes f3 (top) and e3 (bottom, black traces), delta-power (gray), and DCR amplitude (top, green; bottom, red) for stimulating electrode positions proximal (top) and distal (bottom) from the SD induction site, as indicated in panel (a). The green and red circles with vertical lines indicate the estimated time (t = 0) of SD onset at the stimulating electrode, as described on panel (b), and (d) average SD (DC-ECoG LFP, top), DCR amplitude (middle), and delta power (bottom), aligned for SDs occurring at the stimulating electrode with different delays relative to the recording site: SD at the stimulating electrode occurred 20 s before (left), 5 s after (middle), or 28 s after (right) the local SD at the recording electrode. Colored circles and vertical dashed lines indicate the SD onset at the stimulating electrode relative to the onset at the recording electrode. Pooled data from 48 SDs recorded from 11 rats.

Blanket inhibition and wave-like patterns of DCR depression

We next elaborated on these findings by exploring DCR depression through the entire ECoG array. To this end, we plotted DCR amplitude maps on the ECoG electrodes during SD propagation. The SD wavefronts were constructed from the onsets of DC-shifts on ECoG electrodes as shown in Figure 2(a). For the regions outside the array, the SD wavefronts were approximated using the procedure shown in Figure 4(b). The resulting DCR maps at different times of SD propagation are presented for the cases of the stimulating electrode positioned upstream (SD-proximal, Figure 5(a)) and downstream (SD-distal, Figure 5(b)) relative to the SD propagation. In the case of SD-proximal stimulation, the DCRs remained unchanged before the SD approached the stimulation site (Figure 5, −25 s). However, as soon as the SD arrived at the stimulation site, DCRs were simultaneously depressed (−15 s) and shortly afterward were completely blocked (−10 s) on all electrodes in the array. The SD entered the ECoG field against a background of complete DCR depression. We propose referring to this pattern of DCR depression, in which all cortical sites simultaneously stop responding to the stimulus, as “blanket inhibition.” This form of DCR depression is likely due to the inhibition at the origin of response, where SD causes depolarization block of neurons and neural fibers, rendering them unresponsive to the stimulus.^1,18 Notably, the recipient cortex remains active until the arrival of the SD wave, as evidenced by the persistence of delta activity and SEPs during the prelude phase of depression between the SD arrival at the stimulation site and its arrival at the recordings site (see Figure 6).

OPEN IN VIEWER

Figure 5.

Blanket inhibition and wave-like patterns of DCR depression during stimulation upstream and downstream of SD. Examples of N1 DCR amplitude maps at different time points of SD propagation through the ECoG array in cases where the stimulating electrode (black dot) was placed: (a) upstream of SD propagation (SD-proximal) and (b) downstream of SD propagation (SD-distal). Color-coded N1 DCR amplitudes are normalized to the control (pre-SD) values. The SD wave front is indicated by a red line. The top-left inset in each panel indicates the time relative to the SD onset on the ECoG electrodes outlined by white circles. Note the “blanket inhibition” pattern on DCR depression across the entire array in the case of SD-proximal stimulation, which onset coincides with the SD’s arrival at the stimulation site (a); in contrast, note the “wave” pattern on DCR depression that follows the SD as it propagates through the ECoG array in the case of an SD-distal stimulation (b).

OPEN IN VIEWER

Figure 6.

DCR depression as a function of the delay between SD on the recording and stimulating electrodes: (a) parameters of DCR depression as a function of the delay between SD at the recording and stimulating electrodes. See text for details, (b) parameters of delta oscillation power depression as a function of the delay between SD at the recording and stimulating electrodes, and (c) group data statistics on the onset, offset, and total duration of depression for DCR amplitude, delta power, and SEP amplitude. DCR1, DCR2, and DCR3 correspond to the time fragments indicated in panels (a) and (b). Brackets indicate significant differences between groups (p < 0.05).

A quite different, more orthodox wave-like DCR depression pattern characterized the dynamics of DCR depression evoked by stimulation downstream of the SD propagation (SD-distal, Figure 5(b)). In this case, the wave of DCR depression strictly followed the SD wave along its propagation through the ECoG array. Here, DCR depression was likely due to the loss of responsiveness in the cortical region invaded by the SD, while the stimulation site remained functional and continued to generate DCRs in regions not yet invaded by the SD until the SD reached the stimulation site.

We further quantified various parameters of DCR depression as a function of the delay between the SD onset at the stimulating and recording electrodes (Figure 6(a) and (c)). Overall, these data could be grouped into three temporal domains: (1) SD at the stimulation site preceding SD at the recording site (fragment 1 (−30 to −10 s), corresponding to SD-proximal stimulating electrode locations), (2) SD occurring nearly simultaneously at the stimulating and recording electrodes (fragment 2 (3–20 s), corresponding to the isochronic stimulating electrode locations); and (3) SD at the stimulation site following SD at the recording site (fragment 1 (25 to −35 s), corresponding to SD-distal stimulating electrode locations). The DCR depression onset significantly preceded the SD onset at the recording site in Fragment 1, consistent with the primary role of early DCR depression caused by SD at the stimulation site during SD-proximal stimulations. In contrast, the DCR depression offset in the Fragment 3 was delayed compared to Fragments 1 and 2, in line with the prolongation of DCR depression by SD at the stimulation site in cases of SD-distal stimulation. Consequently, the shortest total duration of DCR depression was observed in Fragment 2, where SD-related depression at the stimulation and recording sites largely overlapped in time. As expected, SD-related depression of delta activity was strictly linked to the local SD at the recording site and therefore showed no dependence on the delay between SD at the stimulation and recording sites (Figure 6(b) and (c)).

DCR depression dynamics for assessing SD propagation

Next, we asked whether the spatial-temporal dynamics of DCR depression could be used to determine the direction of SD propagation across the ECoG array. To this end, we first calculated the average SD direction vectors from the delays in SD-related DC shifts in randomly selected pairs of ECoG electrodes (Figure 7(a) and (b)). These vectors were then compared with the vectors of DCR depressions. Note that in DCR vector analysis, delays less that interstimulus interval of <3 s were excluded from the analysis. Representative examples of circular plots of DCR depression onset, offset and merged onset and offset vectors in SD cases displaying wave-like (left) and blanket inhibition (right) DCR depression patterns, are shown in Figure 7(c) below the corresponding DC shift vector plots (Figure 7(a)). DCR depression onset vectors matched the direction of DC shift vectors in the vast majority of SDs (52 out of 66 SDs) that displayed a wave-like DCR depression pattern on at least part of the electrodes. On average, the DCR depression onset vector deviated from the DC shift vector by 2.1° (−2.3° to 5.6°; n = 52 SDs; Figure 7(d)). These included not only the full wave pattern, but also the mixed wave/blanket inhibition pattern of DCR depression, in which the SD first caused a wave-like DCR depression pattern in some of the electrodes, before arriving at the stimulation site and causing a blanket depression pattern in the rest of the ECoG electrodes. In cases with a pure blanket DCR depression pattern (14 out of 66 SDs), none of the ECoG site pairs showed delays in DCR depression onset of more than 3 s. Recovery from SD-induced depression of DCRs also reflected the direction of SD propagation, as evidenced by DCR depression offset vector plots (Figure 7(c)). Although this metric was less precise than DCR depression onset, with an angle of deviation of 2.4° (−19.0° to 39.8°) from the direction of the DC shift (n = 61 SDs; Figure 7(d)), it was more ubiquitous (61 out of 66 SDs). Cases that “escaped” SD direction detection from DCR offsets included those with stimulation sites located furthest downstream of SD propagation, where DCR recovery was primarily determined by recovery at the stimulation site. Finally, combining DCR depression onset and offset vectors (Figure 7(c), bottom plots) enabled estimation of SD direction in 64 out of 66 SDs with decent precision, with an average deviation of 1.6° (−1.9° to −18.7°) from the DC-shift direction (n = 64 SDs; Figure 7(d)). In only two SD cases where the stimulus site was close to the ECoG array, upstream of the SD propagation, the blanket inhibition by the SD at the stimulus site largely overlapped in time with the SD in the recipient cortex, and DCR vectors could not be obtained for the DCR depression onsets and offsets. Nevertheless, the spatial-temporal features of DCR depression were informative for SD direction in the vast majority of SDs and the algorithm proposed here could be useful for assessing SD propagation during multisite ECoG recordings.

OPEN IN VIEWER

Figure 7.

SD direction assessment from DCR depression: (a) circular plot of SD direction vectors for the onset delays of DC-shifts in pairs of ECoG electrodes as shown on panel (b). Each individual point corresponds to an SD vector in a pair of randomly selected electrodes (n), average vector is shown by gray arrow, and average vector angle is shown by gray dashed line and black circle, 0° = right of the ECoG array, high-potassium was applied on the left from ECoG array. The plots represent two SD cases with stimulation upstream and downstream of SD with wave (left) and blanket inhibition (right) DCR depression patterns, respectively, (b) three example SD vectors (black arrow) show delay (yellow box) of DC shits in a pair of electrodes and angle against the right of the ECoG array (red), (c) corresponding circular plots of SD direction vectors calculated from delays of DCR depression onset (top), DCR offset (middle), and DCR onset and offset (bottom) in pairs of electrodes. Note that vectors of <3 s delay values were excluded, and that DCR depression onset plot in case of blanket inhibition is empty because DCRs were simultaneously and completely depressed during SD, and (d) group data on a deviation angle in SD direction estimation from DCR depression onset (top), offset (middle), and onset and offset (bottom) relatively SD direction estimated from DC shifts. Right insets show circular histograms of the angle error. Top boxplots show median (red line), Q1–Q3 quartile, confidence intervals (whiskers).