Brain Diffusion after Single Seizures


Address correspondence and reprint requests to Prof. Dr. A. Hufnagel at Department of Neurology, University of Essen, Hufelandstr. 55, D-45122 Essen, Germany. E-mail:


Summary:  Purpose: Diffusion-weighted magnetic resonance imaging (DWI) after focal status epilepticus has demonstrated focal alterations of the apparent diffusion coefficient (ADC) in the epileptogenic zone. We hypothesized that localized dynamic alterations of brain diffusion during the immediate postictal state will be detectable by serial DWI and correlate with the epileptogenic zone.

Methods: Nine adult patients (four men, five women) with medically intractable epilepsy were prospectively examined with a total of 25 DWI scans taken 2–210 min after a seizure.

Results: The interictal ADC was significantly (p < 0.05) elevated in the ictogenic hippocampus in all patients with temporal lobe epilepsy. The following postictal changes of the ADC were seen: (a) decreases by maximally 25–31%, which were most pronounced in the epileptogenic zone (n = 2); (b) generalized ADC changes after generalized seizures (n = 1) or prolonged complex partial seizures (n = 2); (c) no major changes after short-lived seizures or if the time to first DWI scan was >15 min or both (n = 3); and (d) widespread bilateral ADC increases after a flumazenil-induced seizure (n = 1).

Conclusions: ADC changes seen during serial postictal DWI are complex and appear to reflect origin and spread of the preceding seizure. A delineation of the epileptogenic zone appears to be possible only in complex-partial seizures of >60 s duration that do not secondarily generalize.

Transient postictal phenomena attributed to the epileptogenic zone have been observed clinically (e.g., hemiparesis), by EEG/electrocorticogram (ECoG) recordings (slow foci and attenuation), and by comparison of interictal with ictal single-photon emission computed tomography (SPECT; hypoperfusion vs. hyperperfusion) (1–4). They provide good to excellent data as to the localization or lateralization of the epileptogenic zone.

Descriptions of focal postictal alterations in structural magnetic resonance imaging (MRI) or CT are limited to a few patients and have mainly been based on T2 imaging sequences (5–10). They appeared to rely on long-lasting seizure activity such as focal status epilepticus (6–8,11). In addition, local postictal hyperperfusion was seen by Penfield during epilepsy surgery and documented by angiography (12,13).

Diffusion-weighted MR imaging (DWI) is a noninvasive tool for the early detection of acute ischemic lesions in humans and in animal models of focal status epilepticus (14–17). Furthermore, data from animal experiments in epilepsy suggest that brain-diffusion changes after focal status epilepticus may persist for hours and even days. Experiences in humans have targeted investigation of focal status epilepticus and are limited to a small number of patients (11,18–21). In these patients, postictal decrease of the apparent diffusion coefficient (ADC) and increase of the DWI signal were seen (22,23).

Latest-technology DWI imaging now allows high-resolution, serial measurements of diffusion changes in small brain volumes without the need of anatomic overlay techniques. Acquisition of a diffusion series (60 images) can be completed within <40 s, thereby minimizing the risk of gross motion artifacts of the patient, particularly during the state of postictal confusion.

We therefore hypothesized that brain-diffusion changes after single seizures may be detected by serial postictal DWI. The aim of the study was to correlate regional and temporal postictal diffusion changes to the epileptogenic zone and to seizure type and duration. Special attention was directed to the comparison of postictal DWI changes with interictal DWI of the same patient and to the comparison with healthy controls.



Ten healthy volunteers (five men and five women) at ages 20–52 years served as controls. Brain diffusion was evaluated at complete rest, and regions of interest (ROIs) were determined as in patients.


We prospectively selected nine patients (four men, five women) 23–48 years old with temporal lobe epilepsy (TLE; n = 5), temporal + extratemporal lobe epilepsy (TLE + ETE; n = 2), and extratemporal lobe epilepsy (ETE; n = 2) from a cohort of 16 patients undergoing evaluation during the same period (Table 1). All patients had long-standing, medically intractable epilepsy with complex partial seizures (CPSs; n = 9), simple partial seizures (SPSs; n = 5), and secondarily generalized seizures (GMs; n = 6). Initially, all patients received complete presurgical evaluation including structural MRI, noninvasive interictal and ictal EEG monitoring, and neuropsychological evaluation. Patients were thereafter selected to receive additional investigation of interictal and postictal DWI. To increase chances of capturing a postictal scan, patients had to have more than five seizures per month. Of the seven patients excluded from the study, three had a seizure frequency of five or fewer seizures/month, one had a high frequency of GMs, and three objected to DWI scanning. Patients with different locations of the epileptogenic zone were chosen from the epilepsy surgery program: patients with a strictly unilateral TLE with unilateral high-degree hippocampal sclerosis (patients 1–3); a patient with unilateral (right) ictal onset TLE and bilateral hippocampal sclerosis after encephalitis during infancy (patient 4); a patient with nonlesional right unilateral TLE, as evidenced by ictal and interictal EEG and temporal hypometabolism in positron emission tomography (PET; patient 5); two patients with left TLE and extratemporal lesions (one suprasylvian and one parietal and parietooccipital), as evidenced by structural MR involvement and EEG (patients 6 and 7); and two patients with epilepsy of extratemporal onset (one frontal and one parietal). Patients 1–3, who received a selective amygdalohippocampectomy, have been seizure free for 18–31 months (apart from one GM seizure in patient 3). In addition, patient 8 has remained seizure free after frontoorbital resection of a vascular malformation and the epileptogenic zone in its vicinity for 18 months. Histopathologic examination confirmed a high degree of hippocampus sclerosis in patients 1–3 and a vascular malformation in patient 8. Surgery was denied in patient 4 because of a high degree of impairment of memory and learning capabilities. Type of seizure, duration, and time to first DWI scan were monitored by video-EEG and direct observation of medical staff permanently observing the patient. If no spontaneous seizure occurred, patients received 1 mg of flumazenil, i.v. (applied in two fractions of 0.5 mg), to activate the epileptogenic zone or induce a seizure (24). As a precaution and to prevent any hazard, a physician was present next to the patient and the scanner after application of flumazenil.

Table 1.  Patient characteristics
SexAge (yr)Seizure typesSeizure
onset in EEG
duration (s)
Structural MRI abnormalities
  1. SPS, simple partial seizure; CPS, complex partial seizure; GM, secondarily generalized seizure; HS, hippocampal sclerosis; MRI, magnetic resonance imaging.

1F30CPS, GMTemporal r60HS r
2F23CPS, GMTemporal r30HS r
3M47SPS, CPSTemporal l20HS l
4M48SPS, CPS, GMTemporal r120Bilateral HS r>l
5F25SPS, CPSTemporal r30None
6M31CPS, GMTemporal l90Left lateral temporal + suprasylvian
 posttrauma atrophy
7F35SPS, CPS, GMTemporal l300–420Left temporal + parietal atrophy
8F31SPS, CPSFrontal l20Left frontobasal hemangioma
9M37CPS, GMParietal r300None

All patients gave informed consent for the study. Approval of the study was obtained from the local ethics committee. Approval included induction of seizures during the process of presurgical evaluation.

DWI acquisition

Baseline DWI scans were obtained in all patients with a minimum time lag of 12 h since the last seizure. No secondarily generalized seizure had occurred during the last 48 h before DWI scanning. Scans were performed at complete rest, with the patients being awake. A total of 25 postictal DWI scans was obtained in the nine patients within 2 to 210 min after seizure onset. All monitored seizures were CPSs, except in patient 4, who had a secondarily generalized seizure. However, CPS seizures in patients 7 and 9 were prolonged (>5 min). Latencies between seizure onset and the first postictal DWI scan varied because patients 1–3 and 6 waited next to the MRI scanner and could be scanned within 16 min after seizure onset, and patients 4, 5, and 7–9 were waiting on a patient ward when the seizure occurred and had to be transferred to the MRI department. In patient 2, a seizure was induced by i.v. injection of 1 mg of flumazenil 7 min after the injection while the patient was in the MRI scanner. DWI scanning was started during the CPS. Four postictal DWI scan seizures could not be evaluated because of motional artifacts.

MRI technique of diffusion imaging

MRI examinations were performed on a 1.5-T MR system (Sonata; Siemens, Erlangen, Germany) equipped with high-power gradients (amplitude, 40 mT/m; slew rate, 200 (T/m)/s). The slice orientation was oblique coronal, angulated by 90 degrees in relation to the parahippocampal gyrus.

Isotropic diffusion-weighted spin-echo imaging was performed with an echoplanar imaging (EPI) diffusion-weighted sequence. The acquisition parameters for the DWI studies were TR, 5,000 ms; TE, 83 ms; FOV, 220 mm; acquisition matrix, 128 × 128; slice thickness, 5 mm; interslice gap, 0.5 mm; and 20 slices, resulting in a total acquisition time of 40 s. ADC maps were calculated during the measurement on a pixel-by-pixel basis, resulting in a pixel size of 1.72 × 1.72 × 5 mm = 14.792 mm3.

The diffusion-sensitizing gradients were applied along each of the three orthogonal spatial directions with b = 0.500 and 1,000 s/mm2.

Typical EPI-induced eddy current–induced artifacts near the skull base and the paranasal sinuses were minimized by ultrashort echo and repetition times, resulting in high spatial and temporal resolution. However, slight distortions of the lateral temporal lobe were accepted to preserve coronal planes for evaluation.

T2 images (TE, 119 ms; TR, 2,800 ms; acquisition matrix, 512 × 512; slice thickness, 5 mm; gap, 0.5 mm) were acquired after every DWI with the same FOV, slice thickness, and position to ease the orientation for ROI definition. We performed no computer-based co-registration because EPI intrinsic warping of the DWIs prohibits direct transmission of ROI drawn in the anatomic images (T2 images) on the ADC maps. DWI scans were not cardiac triggered.

Evaluation of regions of interest

In all patients, cylindrical voxels were placed bilaterally in the following structures: Mid-hippocampal (AH), parahippocampal gyrus (PHG), thalamus (Thal), central white matter (WM), and cortex of the convexity remote from the epileptogenic zone (Cortex). Voxel volumes were generally symmetric and ranged from 25 μl (AH) to 50 μl (other ROIs). Detection of ROIs was based on T2 images co-registered with each series of DWI.

ROIs were directly defined in the ADC maps, thereby carefully avoiding ADC outliers (ADC values > 150 × 10–5 cm2/s) for the purpose of minimizing the risk of partial-volume effects (e.g., voxels containing CSF).

Statistical methods

We used SPSS to calculate averages and standard deviations (SDs). Group comparison between patients (as a whole) and controls could not be performed because of small sample size and diversity of the patient group. Paired t test was applied for comparison of right- versus left-sided measurement in controls and for baseline versus postictal measurements in patients. Wilcoxon's test was used for comparison of ictogenic versus nonictogenic side in patients with TLE, because the group distribution was unknown.


DWI in healthy subjects

Average ADC values ±1 SD are shown in Table 2. No significant differences were found when comparing analogous ROIs of the right and left hemisphere (paired t test). Mean ADC values ±2 SD were regarded as normal.

Table 2.  Apparent diffusion coefficient values
Regions of interestRight ± 1 SDLeft ± 1 SDMean ± 1 SD
  1. ADC values (10−5 cm2/s) of healthy subjects (n = 10) ± 1 SD (standard deviation).

Hippocampus89.2 ± 5.2585.8 ± 4.5287.5 ± 4.26
Parahippocampal gyrus78.5 ± 6.4776.2 ± 7.9377.4 ± 6.25
Thalamus82 ± 3.6581.6 ± 4.1781.8 ± 3.62
Cortex89 ± 6.3492.78 ± 6.9291.6 ± 4.27
White matter85.7 ± 2.1187.5 ± 4.4586.6 ± 2.77

Interictal (baseline) DWI

In the seven patients with TLE (including two patients with TLE + ETE), the interictal ADC was significantly (p = 0.018, Wilcoxon's test) elevated in the hippocampus on the ictogenic side by an average of 17.1% (mean, 110.9 × 10–5 cm2/s; range, 98 × 10-5 cm2/s to 125 × 10-5 cm2/s) as compared with the nonictogenic side (mean, 91.9 × 10-5 cm2/s; range, 77–110.9 × 10-5 cm2/s).

In these seven patients, the ADC in the ictogenic AH was above the ±2 SD margin of controls (margin controls right, 99.7 × 10-5 cm2/s; margin controls left, 94.84 × 10-5 cm2/s). The ADC on the nonictogenic side was not elevated and virtually unchanged when compared with controls in five of the seven patients. The two patients with elevation of the ADC on the nonictogenic side were patient 4, who had had encephalitis, and patient 7, who had a porencephalic cyst and perilesional atrophy.

In the two patients with ETE, the interictal ADC of the hippocampus was within normal limits on both sides.

The ADC of the PHG was within normal limits on either side, except in patient 6 (who had zoster encephalitis), in whom it was slightly elevated to 93 × 10-5 cm2/s on the nonictogenic right side, and patient 7, who had a porencephalic cyst, where it was elevated to 101 × 10-5 cm2/s just above the +3 SD limit. The ADC of the Thal, WM, and (unaffected) cortex were within the ±2 SD limits of controls in all patients.

Postictal versus baseline (interictal) measurements

Because the seizures showed great variations in type and duration, all patients are discussed and analyzed on an individual basis. ADC values of all measurements are shown in Table 3. Statistical differences between baseline and postictal measurements were considered significant on a p < 0.05 level.

Table 3.  Apparent diffusion coefficient values
Patient no.
  1. ADC values (×10−5 cm2/s) of baseline and postictal measurement of patients
    (n = 9).

  2. min, minutes after seizure onset; focus, side of seizure origin in ictal EEG;r, right; l, left; t, temporal; f, frontal; p, parietal; AH, hippocampus; PHG, parahippocampal gyrus; Thal, thalamus; Cor, cortex (unaffected); WM, white matter; b, baseline measurement; SD, standard deviation; n.c., not calculable.

r-tSD (b)961211571212911
 SD (16)89129810131388
 SD (37)10951077913108
 SD (11)111111119714878
r-tSD (b)124991071412119
 SD (2)861310991517129
 SD (15)19133759815139
 SD (30)14151191216991311
 SD (62)1091091271291010
l-tSD (b)510101178141588
 SD (16)81510148101412813
 SD (21)981413141317141012
 SD (27)941315121213151112
 SD (210)9131311991313108
r>l-tSD (b)161312261292316812
 SD (44)1610232489191498
r-tSD (b)12612379111279
 SD (50)1476135691064
 SD (75)1166676121578
l-tSD (b)1220241612101616108
 SD (13)15961212128101112
 SD (18)1012151513111918610
 SD (34)1291821108131388
 SD (66)121017141112131189
 SD (92)152111149111116813
l-tSD (b)813101799161058
 SD (60)n.c.n.c.141013121115912
l-fSD (b)81114107111410129
 SD (25)91012131099999
 SD (40)8101213111410678
r-pSD (b)13131013101113768
 SD (55)111071389111388
 SD (77)129611129131789
 SD (184)911151310811141110

Patient 1


This 30-year-old woman had had clear-cut right TLE since age 14 years, with unilateral high-degree hippocampal sclerosis. Interictal epileptiform activity was sparse and strictly ipsilateral temporal. After selective amygdalohippocampectomy, she has remained completely seizure free for 6 months;16 min before DWI, she had a CPS lasting ∼1 min.

ADC measurements

Sixteen minutes after seizure onset, a significant decline of the ADC by 24.8% was seen in the epileptogenic AH. The ADC remained significantly declined after 37 (–17.6%) and 63 min (–16%), and a partial return to baseline could be observed. The ADC of the contralateral AH remained unchanged. The ADC of the right cortex increased by 12.1% after 16 min and significantly by 19.7% after 37 min. A significant decline of the cortex of the nonictogenic left side was seen after 63 min. The ADC of the PHG, WM, and Thal remained unchanged on both sides.

Patient 2


This 23-year-old woman had had right TLE since age 17 years. MRI depicted high-degree right hippocampal sclerosis. The EEG demonstrated clear unilateral right-side ictal onset and persistent right temporal interictal epileptiform activity. Before DWI, she had an epigastric aura and short impairment of consciousness induced by 1 mg flumazenil given i.v. 7 min earlier. The seizure lasted ∼30 s and was identical to her habitual seizures. After selective amygdalohippocampectomy, she has remained completely seizure free for 8 months.

ADC measurements

An increase of the ADC by 14.7–20.5% in the epileptogenic (high-degree sclerotic) AH was seen 15–30 min after the seizure (significant after 30 min). Complete return to baseline had occurred 62 min postictally. A significant increase of the ADC of maximally 22% was observed in the contralateral (nonepileptogenic) AH after 2 and 15 min, but return to baseline was faster. The ADC of the cortex on the epileptogenic (right) side demonstrated a significant and continuous increase reaching 39% at 62 min postictally. The cortical ADC on the nonepileptogenic (left) side remained unchanged for the first 30 min and then significantly decreased by 16.7%. The ADC of the PHG significantly declined by 21% on the nonepileptogenic side 15 min after the seizure. The ADC of the Thal and WM remained unchanged bilaterally.

Patient 3


A 47-year-old man had unilateral left TLE with marked hippocampal sclerosis since age 13 years. Ictal and interictal epileptiform activity in the EEG was strictly lateralized to the left. Sixteen minutes before DWI, he had a short CPS of 20-s duration. After selective amygdalohippocampectomy, he has remained completely seizure free for 8 months, with the exception of one GM seizure after 4 months.

ADC measurements

No relevant alterations of the ADC were seen in the epileptogenic zone (AH, PHG), adjacent regions, or the analogous contralateral ROI. The ADC of the cortex on the nonictogenic right side significantly increased after 16 and 27 (19%) min. The ADC of the WM on the nonictogenic side significantly decreased by (maximum 14.4%) 16, 21, and 210 min postictally.

Patient 4


The 48-year-old man had TLE with right ictal onset and bilateral hippocampal sclerosis after encephalitis 5 years before. He had a secondarily generalized tonic–clonic seizure of ∼120-s duration with cyanosis of the lips and postictal confusion for 40 min. Surgery was denied because of impaired learning and memory capabilities.

ADC measurements

Forty-four minutes after this severe seizure, the ADC was elevated in almost all ROIs. The following elevations were found: AH: right, 8%; left: 11.3%; PHG: right, 22.7%; left, 12.9%; Thal: right, 17.8%; left, 13.5%; Cortex: left, 18.3%; right, no change. In contrast, the ADC of the WM decreased by right, 9.8%; and left, 9.1%. The elevations in the left AH and thalamus on both sides were significant.

Patient 5


A 25-year-old woman who had TLE including SPSs (epigastric auras) and CPSs since age 14 years. Structural MRI was normal. Ictal and interictal EEG demonstrated consistent right temporal lobe epileptiform activity. A PET scan depicted right temporal hypometabolism. Before DWI, she had a CPS of ∼30-s duration.

ADC measurements

At 50 and 75 min postictally, a small (11.7%) but significant decrease of the ADC in the nonepileptogenic AH was seen.

Patient 6


A 31-year-old man who had CPSs of temporal type and secondarily generalized tonic seizures since age 13 years after a zoster encephalitis. MRI depicted a left temporolateral atrophy with slight involvement of the suprasylvian and angular gyri. Interictal epileptiform activity was localized predominantly over F7, T3, and T5. Before DWI, he had a CPS of ∼90-s duration.

ADC measurements

Significant declines of the ADC of the hippocampus on the ictogenic left side were seen after 13, 18, 34, and 66 min. The ADC decrease reached a maximum of 31% after 66 min and partially returned to baseline after 92 min. The ADC of the hippocampus on the nonictogenic side did not change at the same time. The ADC of the PHG remained unchanged on both sides. A significant decline of the ADC of the Thal was seen on the ictogenic side after 66 min (–15%). The ADC of the WM demonstrated a significant decline on both sides after 18, 34, 66, and 92 min.

Patient 7


This 35-year-old woman had SPSs, CPSs, and secondarily generalized seizures since her first year of life. The MRI demonstrated a porencephalic cyst and perilesional atrophy, which was marked in left temporolateral, temporooccipital, and left temporoparietal areas. Before DWI, she had a prolonged CPS with automatisms of the face and arms for ∼5- to 7-min duration.

ADC measurements

The ADC declined significantly in both AH and both PHG 60 min postictally. The following decreases were seen: AH right, –25%; AH left, –30%; PHG right, –18%; PHG left, –27%. Decreases had a tendency to be higher on the ictogenic left side.

Patient 8


This 32-year-old woman had CPSs with epigastric aura, arrest, and short-lived tonic posturing since age 14 years. The MRI demonstrated an intracerebral left frontoorbital vascular malformation that, according to the surrounding hemosiderin, had bled in the past without clinical manifestation (other than seizures). A tailored frontoorbital resection of the lesion and 1–1.5 cm of surrounding cortex was performed. The patient has remained seizure free for 8 months. Before DWI, she had a CPS of 20-s duration.

ADC measurements:

A fairly large susceptibility artifact in the epileptogenic zone, caused by ethmoidal sinuses and hemosiderin, precluded direct ADC measurements in the frontoorbital epileptogenic zone. No significant changes were seen at any ROI on either side 25 and 40 min after the seizure.

Patient 9


This 37-year-old man had frequent CPSs and GMs after early childhood meningoencephalitis. He had a right parietal epileptogenic zone defined by ictal and interictal EEG. The structural MRI demonstrated no abnormality. Before DWI, he had a long-lasting CPS of ∼5-min duration.

ADC measurements

The ADC increased significantly by 18.5% in the AH on the ictogenic right side. It also significantly increased after 55, 77, and 184 min in the WM and after 77 min in the cortex of the ictogenic side. A significant ADC decrease of 17.6% was seen in the PHG of the nonictogenic left side after 55, 77, and 184 min and a decrease of 16% in the PHG of the ictogenic side.

Summary of individual postictal ADC measurements

A complex spatiotemporal pattern of ADC-changes emerged. The observed patterns may be categorized as follows:

  • 1The epileptogenic zone (AH) clearly demonstrated dynamic reversible, postictal decreases of the ADC, with complete or partial return to baseline in patients 1 and 6. In both patients, dynamic changes of minor extent also were seen in other ROIs, but changes were maximally in the epileptogenic zone.
  • 2Widespread bilateral increases of the ADC were seen in patient 4, 44 min after a GM seizure.
  • 3Bilateral substantial decreases of the ADC in the AH and PHG were seen in patient 7, 60 min after a long-lasting CPS.
  • 4Patient 9 had widespread bilateral ADC changes 55–184 min after a long-lasting CPS. ADC increases lateralizing to the ictogenic hemisphere were seen in the AH, the cortex, and the WM of the ictogenic side. In contrast, significant decreases were seen in the PHG of the nonioctogenic hemisphere.
  • 5No changes of the ADC in the ictogenic AH were seen in patient 3, who had a short seizure of only 20-s duration. He demonstrated increases of the ADC in Cortex and decreases in WM during the period 16–210 min after the seizure.
  • 6Significant decreases in the nonictogenic ADC were seen in patient 5, who had a short-lived CPS of 30-s duration and was scanned 50 and 75 min postictally.
  • 7No relevant changes of the ADC were seen in patient 8, who had a short-lived CPS and was first scanned 25 min postictally. The epileptogenic zone itself could not be explored because of susceptibility artifacts.
  • 8In patient 2, the seizure was induced by flumazenil. Even though the patient declared the seizure to be identical with spontaneous seizures, pharmacologicinduced effects may have to be considered.


Technical considerations

These preliminary findings demonstrate the feasibility of the method. It was possible to acquire DWI scans repetitively in the postictal period. Only four scans had to be repeated because motional artifacts of the patient had degraded the image quality. However, only a few patients tolerated being in the scanner for >2 h. In addition, susceptibility artifacts at the temporopolar and frontoorbital regions further limit use of this method. Furthermore, ROI analysis may be limited by volume-averaging artifacts; especially, inclusion of CSF in analyzed voxels must be avoided Further improvement may be achieved by using segmented EPI data acquisition to reduce image distortions and enhance spatial resolution.

Interictal ADC measurements

In this study, the interictal ADC was significantly increased (p = 0.018) in the epileptogenic hippocampus (AH) when compared with the nonictogenic AH. Elevations above the +2 SD margin of controls coincided with the side of seizure onset. Furthermore, the ADC in the AH was bilaterally elevated in a patient with bilateral hippocampal sclerosis and not elevated in patients with ETE. This fairly robust finding is in keeping with previously published experimental data and studies in humans. In the study of Wieshmann et al. (25) and that of Hugg et al. (26), a higher ADC and a lower anisotropy index was demonstrated in the atrophic hippocampus in 14 and eight patients with hippocampal sclerosis and TLE, respectively. Yoo et al. (27) reported that 18 of 18 patients with intractable TLE had elevated interictal ADC on the epileptogenic side. The most plausible hypothesis for this observation is enlargement of the interictal space as a consequence of degenerative neuronal loss. An elevated ADC may thus be used as a surrogate marker to lateralize the side of seizure origin in TLE.

DWI after status epilepticus

Postictal DWI studies in animal experiments have so far focused on observations during or after the kainic acid– or pilocarpine-induced status epilepticus in the rat (15,16,28,29). In the kainate-induced status epilepticus in the rat, the ADC decreased by 7–30%, and MR-visible sodium contents increased by 12–90% during a period 5–24 h postictally. Maximal changes were seen in the piriform cortex and the amygdala (16). During the early phase, little evidence for structural damage was detected on T2-weighted MR images. These findings are consistent with the hypothesis that sequential seizures lead to an intracellular influx of sodium ions. Failure of the Na+/K+-ATPase may lead to subsequent cytotoxic edema, thereby reducing water diffusion. Excessive release of excitatory amino acids, such as glutamate (30,31), and increased membrane ion permeability may contribute to edema during status epilepticus, which eventually may evolve into cell necrosis or apoptosis. Affirmatively, swelling of dendrites and astrocytes has been described histologically in animal experimental studies (31). The ADC changes were closely correlated with the assumed epileptogenic brain areas and the tissue damage such as neuronal pyknosis and neuropile vacuolation shown on histopathologic examination (15).

The decrease of the ADC in the hippocampus, amygdala, and piriform cortex after the kainic acid–induced status epilepticus may last 1–3 days and completely resolve after 9 days (16,28,29).

The return of the ADC to baseline values and greater during the days to follow may be explained by subsequent neuronal degeneration, leading to an increase of the extracellular space. Neuronal and glial cell death after status epilepticus has been documented in animal experiments (32–35) and by MR-detected brain atrophy in humans (10).

In contrast to experimental data, evaluations of DWI in humans are based on a few patients, mainly with epileptogenic brain lesions (18–20,36). Furthermore, in most reports, the effects of an underlying neurologic disease such as cortical vein thrombosis (20), intraparenchymal hemorrhages, and infection (11) are difficult to distinguish from seizure activity. Lansberg et al. (11) found a decrease of the cortical ADC in circumscribed areas of the cortex in three patients with focal status epilepticus. Diehl et al. (19) found decreases of the ADC in only one of six patients after a short seizure. However, her scanning DWI scans were only 45–150 min after the seizure, and short-lived effects may have vanished. In a patient with a nonconvulsive status epilepticus, an 18% decrease of the ADC and a 28% increase of regional cerebral blood flow was found within the affected left temporoparietal cortex when compared with the unaffected side (21). All changes normalized within 1 month.

DWI after single seizures

In contrast to DWI measurements after status epilepticus, dynamic ADC changes after single seizures have not been reported. Hypothetically, postictal changes may rely on many factors such as seizure duration, type, and propagation or the underlying lesion. Furthermore, the time lag since seizure termination may be crucial.

In this study, we hypothesized that ADC alterations may be assumed to differ from those after status epilepticus, because single seizure–induced cell membrane changes and ion imbalances are quickly transient, and morphologic damage usually does not occur, or only to a minor degree.

In keeping with this hypothesis, Zhong et al. (37) found an ADC reduction of 4% after a single 10-pulse cortical electrical-stimulation train that lasted 0.1 s. The ADC decrease was marked (–7 to 8%) if shocks were repeated once a minute. Based on this experience in animal experiments, one would expect the ADC to decline postictally and return to baseline within minutes to a few hours in humans. Changes resembling this hypothesis were seen in patients 1 and 6, in whom an ADC decline of maximally 25–31% occurred, and a complete (patient 1) or partial (patient 6) return to baseline was seen in the epileptogenic zone. Additional changes of minor extents were seen in both patients additional changes outside the epileptogenic zones.

Hypothetically, postictal ADC changes may be quickly transient and related to the seizure severity (duration). They may not be detected at all if the seizure was short-lived or if the time lag until the first DWI scan was too long. Affirmatively, no major ADC changes were seen in patients 3, 5, and 8, who had seizures of ≤30-s duration and were scanned 16–75 min after the seizure. From the postictal ECoG via subdural electrodes, we know that no slow focal activity is detectable if a seizure lasted <30 s (2). Moreover, in ictal SPECT, injections of the tracer have to be performed within a short period of ∼2 min to enable depiction of hyperperfusion in the epileptogenic zone.1 These observations point out that changes of the ADC values after a single seizure are possible but not obligatory.

In contrast to purely focal seizure activity, generalized suppression of brain electrical activity is known to occur after generalized seizures and may last for hours and days. However, whereas generalized increases of ADC values were seen in all regions other than WM in patient 4 after a GM seizure and in AH, Cortex, and WM of patient 9 after a prolonged CPS, a generalized decrease of ADC values was seen in patient 7, 60 min after a prolonged CPS. There are two possible explanations for this divergence. The first hypothesis is the cyclic course of the interictal–ictal–postictal–interictal transition. The underlying cyclic changes of brain electrical activity are known to correlate with changes of cellular processes at all levels and vascular supply. In particular, reactive hyperpolarizations and hyperperfusion are known to occur (12,13). In SPECT, an interictal hypoperfusion is followed by an ictal hyperperfusion and a marked postictal hypoperfusion. On the cellular level, interictal epileptiform discharge represents synchronous depolarization in a small neuronal pool. This turns into highly repetitive discharges during the seizure and usually sparse epileptiform discharge postictally. The ictal depolarizations may lead mainly to massive Na+ and Ca2+ influx followed by hydrate water and K+ efflux. This possibly results in a temporary breakdown of the Na+/K+ pump, and K+ overload in the extracellular space and glia (38). Because the intracellular compartment is more restrictive to water motion, the ADC usually decreases. During the postictal recovery period, all changes are usually reversed to the interictal baseline but may be overreactive during a transitory phase. The ADC value measured may hence depend on the exact timing and may possibly correlate with phases of reactive hyperperfusion or reactive hyperpolarization (39).

Alternatively, the epileptogenic process itself may cause differences in diffusion changes. Whereas ADC increases have been observed in the rat hippocampus 24 h after a pilocarpine-induced status epilepticus (17), ADC decreases have been found after kainate-induced status epilepticus (16). Whereas kainic acid is a glutamatergic, excitatory amino acid that has a high affinity to CA1 and CA3 hippocampal pyramidal neurons and causes excitotoxic neuronal loss, pilocarpine is a cholinergic agonist that primarily activates cholinergic afferents to the granule cells of the dentate gyrus. Granule cells are known to act as filters to excitatory activation of the secondary pyramidal cells and die apoptotically rather than necrotically (40,41). Both processes may diminish or protract neuronal loss and so cause differences in seizure-induced ADC changes. The specific pathomechanisms of the underlying epilepsy syndrome may lead to differences in postictal diffusion changes.

Overall, because ofo the heterogeneous ADC changes, the small sample size, and different types of seizures, no characteristic correlations could be observed. In contrast to the generally decreased ADC values after focal status epilepticus, ADC changes after a single seizure appear to be complex.

In this respect, serial DWI scans, starting as early as possible after the seizure, are necessary to depict fully the spatiotemporal pattern of DWI changes in the epileptogenic area and its vicinity.

In conclusion, DWI proved to be a sensitive diagnostic tool for the in vivo depiction of functional postictal brain diffusion alterations after single seizures. The spatiotemporal pattern of ADC changes appears to be complex and may be indicative of the underlying seizure activity. More data from larger DWI series are needed to decide whether DWI may be used for functional delineation of the epileptogenic zone in the course of presurgical evaluation (42–44).