Ictal Brain Hyperperfusion Contralateral to Seizure Onset: The SPECT Mirror Image


Address correspondence and reprint requests to Dr. G. Huberfeld at INSERM U739, Cortex & Epilepsie, Faculté de Médecine, Pitié-Salpêtrière, 105 Bd de l'Hôpital, 75003 Paris, France. E-mail: gilles.huberfeld@chups.jussieu.fr


Summary: Purpose: Ictal single-photon emission computed tomography (SPECT) may help localize the seizure-onset zone (SOZ) by detecting changes in regional cerebral blood flow induced by epileptic discharges. This imaging method also reveals hyperperfusions in areas of seizure propagation, including the hemisphere contralateral to the SOZ. We have studied the occurrence, the topography, and the clinical value of such contralateral ictal hyperperfusion areas (HPAs).

Methods: We examined data from presurgical evaluations of 36 consecutive patients with pharmacoresistant partial epilepsy of various localizations. Ictal and interictal SPECT examinations were made with 99mTc-ECD, and the scans were processed for coregistration, normalization, subtraction, and merging with MRI images.

Results: Contralateral HPAs were observed in 72% of the patients: 50% of mesiotemporal epilepsy cases with hippocampal sclerosis, 85.7% of the other mesiotemporal epilepsies, 85.7% of neocortical lateral temporal epilepsies, and 87.5% of extratemporal epilepsies. Contralateral HPAs were usually symmetrical to the SOZ, forming a mirror image, observed in 57.1% of the patients. They could be slightly asymmetrical in mesiotemporal epilepsies, perhaps because of the particular anatomic pathways linking temporal lobes. In neocortical epilepsies, they were located in the cortex homotopic to the SOZ.

Conclusions: We show that the symmetrical nature of the mirror image usually does not disturb SPECT interpretation. It can confirm the location of the SOZ (11 patients) and even occasionally improve the precision of its definition (nine patients) by restraining several potential SOZ-related HPAs to a single one or by permitting a restricted localization of the SOZ in a large HPA.

Focal epileptic discharges are generated in limited cerebral areas and propagate to both adjacent and distant cortical regions (1). Surgery of pharmacoresistant focal epilepsies aims to remove the cortical area responsible for seizure generation. Presurgical evaluation is therefore crucial to determine the seizure-onset zone (SOZ) and to distinguish it from zones of seizure propagation (1). Such propagation of epileptic activity can be local and slow, mediated by synaptic and nonsynaptic mechanisms, or long-range and fast (2), depending on myelinated fibers emerging within the SOZ or locally activated zones and projecting to distant ipsilateral and contralateral sites.

The tools used to identify the SOZ include clinical observation (1) with simultaneous surface (3,4) or intracranial EEG recordings of interictal and ictal semiology (5), positron emission tomography (PET) (6,7), magnetic resonance imaging (MRI), and ictal single-photon emission computed tomography (SPECT).

SPECT studies permit visualization, after injection of a radioactive tracer, of areas with an increased regional cerebral blood flow (RCBF) associated with ictal activity (8,9). The tracer, injected shortly after seizure onset, reaches the brain, is extracted, and is retained by neurons according to the RCBF. Because of the time delays involved in these processes, ictal SPECT measurements cannot always discriminate between the SOZ and areas receiving propagated activity. Hyperperfused areas (HPAs) will correspond both to persistently activated SOZ and to propagation areas. Although it has a poor temporal resolution, ictal SPECT is a powerful presurgical tool, because it provides a snapshot of the whole brain with a good spatial resolution. It may be particularly useful when the SOZ is difficult to localize, when structural brain abnormalities are not apparent, or when investigation tools do not agree. Its sensitivity (ability to localize the SOZ correctly) in temporal lobe epilepsies varies between 75 and 97%, and its specificity, between 71 and 100% (8–22).

Ictal SPECT studies have occasionally revealed HPAs in the hemisphere contralateral to the SOZ (20,23–25). However, although clinical observation and scalp or intracranial EEG records show that seizures propagate to the contralateral hemisphere (26), no SPECT study has specifically focused on contralateral HPAs.

This ictal SPECT study was therefore carried out to analyze ictal HPAs contralateral to the SOZ. We found that ictal SPECT often revealed contralateral HPAs at sites that mirrored the SOZ. We describe the occurrence of those mirror images in several types of partial epilepsy, their topography, their relation to the type of seizure, and their influence on SOZ localization in presurgical evaluation.



All patients who had received both interictal and ictal SPECT examinations for presurgical evaluation in the Epileptology Unit of the Pitié-Salpêtrière hospital, between January 1, 1998, and December 31, 2000, were reviewed for inclusion in this retrospective study. They were included only when the SOZ location could be precisely determined from investigations, other than the SPECT examination, and from the postoperative outcome. Ictal SPECT studies were performed routinely in 13 patients whose SOZ was unequivocally located by usual or noninvasive means. Twenty-three other patients required further explorations, comprising PET (n = 14) or intracranial recordings (n = 15) or both. We excluded nine patients in whom the SOZ could not be satisfactorily localized.

Thirty-six patients (14 women and 22 men) were included (see Table 1 for clinical data). Twenty-eight patients were diagnosed with temporal lobe epilepsy (TLE). The SOZ was localized to lateral, basal, or polar cortex (latTLE) in seven cases and to mesial structures (mTLE) in 21 cases. In the mTLE subgroup, hippocampal sclerosis (HS) was present in 14 cases. Eight patients had extratemporal epilepsies, including three frontal lobe, two central sulcus, and three temporoparietooccipital junction cases.

Table 1. Demographic, clinical, and SPECT examination data for patients studied
  1. *Intracranial recording. DNET, dysembryoplastic neuroepithelial tumor; SOZ, seizure-onset zone; HPA, hyperperfusion area; HS, hippocampal sclerosis; latTLE, lateral TLE; mTLE, mesial TLE; TLE, temporal lobe epilepsy; L, left; R, right.

 1R mTLEHSIIB50/141Temporal pole+Temporal poleLobeStructure+Improvement
 2R mTLEHSIC62/83Temporal basal and lateral+Temporal basolateral, frontal cingular gyrusLobeStructure+Neutral
 3L mTLE*HS16/3400 
 4L mTLEHS18/29Temporal antero-internal0 
 5L mTLEHSIA (9 month)29/106Temporal antero internal+Temporal internal (parahippocampal gyrus)StructureLobe+Improvement
 6R mTLEHSIA10/88Temporal internal+ 
 7L mTLEHSIA (12 month)37/57Temporal internal and pole+ 
 8L mTLEHSIB59/104Temporal pole+ 
 9L mTLEHSIA30/68Temporal internal and pole+ 
10R mTLEHSIA (7 month)45/91Temporal internal, pole and lateral+Temporal poleLobeStructure+Confirmation
11R mTLE*HSIII A12/98Temporal internal, pole and lateral+Temporal global, frontal orbitarStructureStructure+Misleading
12R mTLEHSIA50/56Temporal internal and pole+ 
13L mTLEHS31/100Temporal internal0Temporal poleLobeLobe Confirmation
14bilateral mTLE*Unilateral HSIV B67/69Temporal internal0Temporal amygdalaStructureLobe+Confirmation
15R mTLETumorIA (16 month)15/119Temporal internal0Temporal amygdalaLobeLobe Confirmation
16R mTLETumorIB (18 month)31/45Temporal pole0Temporal pole, frontal orbitar posteriorLobeStructure+Confirmation
17R mTLE*DNETIA55/70Temporal pole+Temporal amygdalaStructureStructure+Confirmation
18L mTLE*DNETIID (9 month)27/95Temporal internal+ 
19L mTLEDysplasiaIIA13/89Temporal antero-internal0Temporal antero-internalLobeStructure+Confirmation
20R mTLE0IA49/70Temporal antero-internal0Temporal amygdalaStructureStructure+Confirmation
21R mTLE*015/37Temporal pole+Temporal poleLobeStructure+Improvement
22L larTLE (pole)DysplasiaIA (10 month)32/93Temporal antero-internal0Temporal antero-internalStructureStructure+Confirmation
23R latTLE (pole)cavernomaIB39/65Temporal pole+ 
24R latTLE (pole)DNETIA (7 month)54/108Temporal pole+Temporal poleStructureStructure+Improvement
25R latTLE (pole)*0IA (6 month)12/63Temporal pole (basal)+Temporal amygdalaLobeLobe Confirmation
26R latTLE (pole)*0IV B8/87Temporal anterior+Temporal pole and amygdalaStructureStructure+Improvement
27R latTLECavernomaIC20/105Temporal lateral and basal0Temporal basal and lateral, frontal anterior cingular gyrusLobeLobe Misleading
28R latTLE (posterior)*0IVA33/50Temporoparieto-frontal0Parietal (angular gyrus), insulaStructureStructure+Improvement
29L Frontal lobe: orbitar posterior*0IA19/84Frontal orbitar posterior+Temporal pole, frontal orbitar posteriorStructureStructure+Neutral
30R Frontal lobe: orbitar internal*CavernomaIC32/80Frontal orbitar posterior+Frontal orbitar posterior, insulaLobeLobe Neutral
31L Frontal lobe: polar internal*0IA33/37Frontal polar internal0Frontal polar internalStructureStructure+Improvement
32L Central sulcus*Neonatal strokeIII B26/33Central sulcus+Insula, frontal orbitar lateral and posterior, frontal cingular gyrus00 Neutral
33L Central sulcus*08/28Central sulci (ascending frontal and parietal gyrus)+Central sulcus, insula, frontal operculaStructureStructure+Improvement
34R Temporo-parieto-occipital junction*DysplasiaIC18/32Parietotemporal+Temporooccipital internal and posteriorStructureLobe+Improvement
35L Temporo-parieto-occipital junction022/78Temporoparieto-occipital junction+Temporal basal, parietalStructureStructure+Confirmation
36R Parieto-fronto-temporalSubcortical Heterotopia20/2100 

Video and EEG recordings of all seizures explored with SPECT were examined to define the type of seizure, its duration, the EEG pattern, the topography and spread of the ictal activity, and the delay of injection of the SPECT tracer.

Data acquisition

Ictal SPECT was performed with video-EEG monitoring and was followed, 48 h later, by interictal SPECT, as previously described (27). In both examinations, 740–925 MBq (20–25 mCi) of 99mTc-ECD was injected intravenously. Images were acquired after 1 h, using either a double-head gamma-camera through high-resolution fan-beam collimators, or a three-headed gamma-camera equipped with parallel high-resolution collimators. No correction for Compton scatter was performed, and projections were reconstructed by using filtered backprojection (low-pass filter) yielding a 128 × 128 × 128 reconstructed volume. A postreconstruction uniform attenuation correction (Chang) was applied.

SPECT data were combined with an anatomic background derived from MRI scans obtained from the same patient, consisting of an inversion-recuperation fast spoiled gradient recall at the steady state (IR-F SPGR) sequence (1.5-Tesla Signa scanner; GE Medical Systems, Milwaukee, WI, U.S.A.).

Data processing

Ictal SPECT images were coregistered with interictal images by using AIR (automated image registration) software (28). Both images were then coregistered with 3D-MRI by using MPItool software (29).

SPECT data were corrected for variable global CBF after excluding voxels influenced by seizure-induced hyperperfusion in the ictal study, and a positive difference image (ictal minus interictal SPECT) was calculated with statistical thresholding. The processing included the following steps:

  • 1For each voxel, calculation of a relative difference (C–I)/I, where I and C represent the values of interictal and ictal studies, respectively.
  • 2For the whole brain, calculation of the mean value (m1) and the standard deviation (σ1) of the relative difference.
  • 3Exclusion of voxels in the ictal SPECT whose value exceeded the interval [(m1–σ1);(m11)] and calculation of a new mean value (m2) of the relative difference with the remaining voxels.
  • 4Normalization of the interictal study to the level of the ictal one by calculating new values (IN) for each voxel: IN= I + (m2× I).
  • 5Calculation of a subtracted volume (normalized ictal SPECT minus interictal SPECT) with a voxel value of S =[(C – IN)/IN]/σN, where σN is the standard deviation of the new relative difference calculated with the normalized values of the pixels of the interictal scan.

The normalized subtracted SPECT and MRI volumes were merged for visual analysis. A value of 2 standard deviations (SD) was used as the threshold of significance of an HPA.

Localization of ictal perfusion changes

SPECT results were interpreted (by M.O.H. and G.H.) blinded to the location of the SOZ defined from the other explorations. The main HPA was visually selected as the one with the highest intensity or the widest spatial extent or both, according to the interpreters' experience. Its location was secondarily compared with the SOZ location. In cases in which the initially defined main HPA and the SOZ did not match, we designated the main HPA as that corresponding to the SOZ. Such a method let us study the localizing value of the blindly defined main HPA and the effect on interpretation of the other hyperperfusions. Accessory HPAs were defined as separate zones in the same hemisphere as the main HPA. Contralateral HPAs were located in the opposite hemisphere. The symmetry of the contralateral HPAs was defined with respect to the site of the SOZ and to the location of the ipsilateral HPAs. Two levels of symmetry were considered: a lobar symmetry and a more precise structural symmetry (the opposite gyrus or nucleus for the amygdala). A mirror image was defined as a contralateral HPA located precisely in the structure symmetrical to the main HPA or the SOZ.

We also asked whether the existence of a mirror image could help identify the SOZ from SPECT data. Contralateral HPAs confirmed the location of the focus when they were symmetrical either to the SOZ or to a restricted main HPA, corresponding to realistic surgical extent. They improved the analysis of the examination when, by using an inverse symmetry, they let us discriminate between several putative main HPAs or restrict the location to a part of an extended main HPA. Contralateral HPAs were neutral in SPECT assessment when they were multiple, correlated with both accessory and the main HPAs. Finally, they were misleading when their intensity or extent was larger than that of the main HPA (false lateralization), or when a single contralateral HPA pointed to an accessory HPA rather than reflecting the SOZ (false location).

Statistical analysis

The χ2 test was used to compare two qualitative variables (corrected for 2.5 ≤ n < 5), the t test to compare means between two groups, and linear regression to compare two quantitative variables. Statistical significance was reached with p < 0.05.


Seizures explored with SPECT

The mean duration of the seizures explored by SPECT was 75.6 s (range, 21–141 s). They were simple partial seizures in eight cases, complex partial in 28 cases (with aura in 13 cases), and secondarily generalized in six cases. The mean tracer injection delay after the start of the seizure was 30.5 s (range, 8–67 s).

Main HPA

The main HPA contained the SOZ in 66.7% of patients (Table 1). A moderate spatial disparity could occur between the location of the SOZ and that of the main HPA in a few patients, particularly in mTLE (nine of 21) and rarely in other locations [latTLE (one of seven) and extratemporal epilepsies (one of eight)]. Indeed, in mTLE, the main HPA could be located either in the amygdalohippocampal formation and the parahippocampal gyrus (i.e., the SOZ itself) or in the pole or in its mesial part (anteromesial pattern) (30). This lack of overlap was as common in cases of mTLE with HS as in other mTLE etiologies (six of 14 and four of seven, respectively; χ2, p = 0.54). In patients with neocortical epilepsies, the main HPA included the SOZ in 13 (86.6%) of 15 cases. In extratemporal epilepsies, the hyperperfusion was larger than the SOZ in one third of frontal lobe cases, half of central sulcus cases, and two thirds of temporoparietooccipital junction cases.

The main HPA location was concordant with the SOZ, whether strictly or with a mild admissible mismatch in mTLE cases, in 18 of 21 mTLE cases, six of seven latTLE cases, and seven of eight extratemporal epilepsies. Further, the three misinterpreted cases in mTLE corresponded to one unilateralization, one false lateralization, and one false location (patients 3, 11, and 13, respectively).

Accessory HPAs

Accessory HPAs were present in 23 (64%) cases, suggesting a propagation from the SOZ to ipsilateral sites, with a single accessory HPA in 13 cases and multiple HPAs in 10 others. Ipsilateral propagation was inferred in 13 of 21 mTLE cases, four of seven latTLE cases, and six of eight extratemporal SOZ cases.

Accessory HPAs were often located in the orbitofrontal cortex (seven of 13) and the insula (eight of 13) in cases of mTLE. Propagation to the polar part of the frontal lobe (two of four) and the insula (two of four) was detected for lateral temporal foci. The sites of accessory HPAs in extratemporal epilepsies were temporal or insular in frontal lobe epilepsies, multiple and large in frontal and temporal areas, or restricted to the insula in central sulcus epilepsies, and either regional or distant in the frontal lobe in two cases of temporoparietooccipital junction epilepsy.

Contralateral HPAs

Contralateral HPAs were observed in 26 (72.2%) cases. In 16 cases, a single HPA was detected, and in 10 cases, multiples HPAs (Table 1) were found. They occurred in seven (50%) of 14 mTLE cases with HS, in six (85.7%) of seven mTLE cases without HS, in six (85.7%) of seven latTLE cases, and in seven (87.5%) of eight extratemporal epilepsies. The difference in frequency between the HS group and the rest of the population is close to significance (χ2, p = 0.068). The occurrence of either single or multiple contralateral HPAs was unrelated to the location of the SOZ.

In 25 of the 26 cases, the contralateral HPA, or at least one when several were present, was symmetrical to the lobe containing the main HPA, and 20 were mirror images.

In mTLE cases with HS, a mirror image was observed in six of the seven cases of contralateral HPA, mainly in the anteromesial area, corresponding globally to 35.7% of all HS cases. In cases of mTLE without HS, contralateral HPAs were located almost exclusively in the mesial part of the temporal lobe corresponding to a mirror image in five of six cases. The proportion of patients with a mirror image (71.4%) was particularly high in this subgroup. For patients with latTLE foci, mirror images were observed in four of seven cases, representing 57.1% of the population. When the SOZ was extratemporal, contralateral HPAs were seen in seven of eight cases (multiple in six cases) and included a mirror image in five of seven cases, or 62.5% of the eight extratemporal epilepsies. In all frontal lobe epilepsies, contralateral HPAs were present in the opposite lobe and formed a mirror image in two of three patients. They were present in two cases with central sulcus focuses but were asymmetrical in one case. A mirror image was observed in two of three temporoparietooccipital junction epilepsies.

Nonsymmetrical contralateral HPAs (cases 2, 11, and 16 of mTLE, cases 27 and 28 of latTLE, cases 29, 30, 32, 33, and 35 of extratemporal epilepsy) were never unique and were associated with a lobar symmetrical contralateral HPA in all cases but one (patient 32), and with a mirror image in all mTLEs, one of two latTLEs (patient 28), and three of five extratemporal cases (patients 29, 33, and 35).

A contralateral HPA was sometimes symmetrical to an accessory HPA rather than the main HPA (n = 3). In one latTLE (patient 27), it was located symmetrically in the contralateral lobe. In two other cases, one mTLE with HS (patient 2) and one frontal lobe epilepsy (patient 29), the contralateral HPA was located outside the lobe homotopic to the SOZ, whereas a mirror image existed among the other contralateral HPAs.

Correlation with electroclinical parameters

We found no correlation between the occurrence of contralateral HPAs and multiple factors including the injection delay, whether absolute or relative to the seizure duration (t test, p = 0.98 and 0.24, respectively), the seizure duration (t test, p = 0.18), the seizure type (aura or loss of contact or secondary generalization), or the duration of the epilepsy (t test, p = 0.10). Further, contralateral HPAs were not correlated with the existence of an accessory ipsilateral HPA (χ2, p = 0.93) or the intensity of the main ipsilateral HPA (t test, p = 0.11). The existence of such a contralateral HPA did not significantly influence the postoperative outcome. Contralateral HPAs were detected in 15 of the 21 operated patients that were free from disabling seizures (class I of Engel's classification) after surgery. In the eight patients that were not seizure free, contralateral HPAs were observed in seven cases (corrected χ2, p = 0.68; NS).

No correlation existed between the occurrence of contralateral HPAs and a contralateral seizure discharge in the scalp EEG during the 30 s after tracer injection. The correct lateralization of the seizure explored with SPECT was confirmed by EEG. Intracranial records were made from the cortical area symmetrical to the SOZ in a single patient in whom a bitemporal epilepsy was suspected (patient 11). In this case, ictal activity consistently propagated to the contralateral mesiotemporal region within the 30 s after seizure onset, and a mirror image was evident in the ictal SPECT. Records of ictal activity did not reveal cases of multiple SOZ, whether corresponding to secondary epileptogenesis or not, except in one case. The sole bitemporal case of this series (patient 14) consisted of a focus associated with a unilateral HS and a polar asymmetrical temporal contralateral focus. No chronologic information suggested a secondary epileptogenic focus. The ictal SPECT of this patient (exploring a seizure that originated from the side containing the HS according to scalp EEG recording) revealed a mirror image.

Influence of mirror images on ictal SPECT interpretation

We next asked whether mirror images might help interpret the ictal SPECT.

  • 1Mirror images confirmed the location of the focus in 11 cases, mainly mTLE cases (in Fig. 1: patients 5 and 22).
  • 2Focus location was improved for nine patients, in two situations. First, when several potential main HPAs existed, the presence of a mirror image facilitated the identification of the true main HPA (patient 24). Second, when the volume of the main HPA was very large, including more than one lobe, a smaller mirror image could restrict SOZ location to a portion of the main HPA (patients 28 and 31). In these cases, reference to the mirror image permitted a more precise location of the SOZ.
  • 3Contralateral HPAs did not influence SPECT assessment in four patients.
  • 4Finally, contralateral HPAs were misleading twice. In one mTLE case with HS (patient 11 in Table 1), a contralateral HPA was clearly higher and wider than the main HPA and could have led to a false lateralization. In a latTLE case, the symmetrical contralateral HPA pointed to an accessory HPA located in the same lobe as the SOZ but distant from it (patient 27 in Table 1).
Figure 1.

Individual cases. Ictal single-photon emission computed tomography (SPECT) examinations are shown in the SISCOM (Substracted Ictal SPECT COregistred to MRI) format. The color scale indicates the intensity of ictal hyperperfusions expressed in standard deviations. The white arrow shows the plan of the sagittal view. Patient 5: mesial temporal lobe epilepsy (mTLE) with left hippocampal sclerosis (HS). A mirror image in the right parahippocampal gyrus confirmed the seizure-onset zone (SOZ) location in the left medial temporal lobe. Patient 22: mesial left temporopolar dysplasia. A mirror image in the mesial part of the contralateral right temporal lobe confirmed SOZ location. Patient 24: right temporal pole dysembryoplastic neuroepithelial tumor (DNET). Multiple ipsilateral hyperperfusion areas (HPAs) in the orbitofrontal region (upper scans) and the temporal lobe (lower scans). A mirror image localized in the left temporal pole improved SPECT analysis, permitting the temporal lobe to be retained as the SOZ. Patient 28: childhood encephalitis, no anatomic lesion. A restricted mirror image in the posterior part of the left superior temporal gyrus improved SPECT interpretation, restraining the SOZ to the symmetrical part of the broad frontotemporoparietal main HPA. Stereo-EEG confirmed the location. Patient 31: no lesion, left frontal lobe epilepsy. A mirror image improved SPECT location of the SOZ in the extensive frontal main HPA.


In this retrospective study, we explored the occurrence and spatial distribution of HPAs contralateral to the SOZ in a wide range of partial epilepsies. These perfusion changes that demonstrate contralateral propagation occurred in >70% of cases studied. Our data suggest that they usually did not disturb SPECT interpretation and that, mirroring the SOZ, they could occasionally contribute to localize it.

Our results were obtained from a population of patients with severe pharmacoresistant partial epilepsies that were hard to localize. We retained only conclusive cases in this study (i.e., patients with SOZ location defined by presurgical investigations but also according to surgical outcome). We included some patients who were not operated on because, although the SOZ was precisely localized, it was situated in a zone that could not be excised for functional reasons. Patients with a focus that could not be localized (n = 9) were excluded. Although several patients experienced some persistent seizures after surgery, we retained them on the basis of sufficiently localizing presurgical data. Reasons for their postoperative failure were not ascertained, but could be a limited resection of a too-complex and wide epileptogenic network or the occurrence of a secondary epileptogenic focus hidden by the main one (31).

One of our findings is the existence, almost exclusively in mTLE cases (nine of 21), of a minor mismatch between the SOZ and its related hyperperfusion. Our anatomic resolution, thanks to SISCOM, permitted detection of this discrepancy, which might otherwise have been classified as a classic more functional “anteromesial” hyperperfusion (23,25,30). This mismatch might result from an imperfect correlation between the SPECT signal and ictal discharges, especially in the mesial part of the temporal lobe. Alternatively, especially extended epileptogenic networks, as shown by intracranial EEG recordings (32) or surgery failures (33), may be implicated.

The most striking result consists of the evidence of a hyperperfusion pattern contralateral to the lobe of seizure onset in partial epilepsies. Contralateral HPAs were visually identified, blinded to the SOZ location in our study as in routine practice, and their extent allowed us to distinguish them from the main HPA, related to the SOZ. Contralateral HPAs are often symmetrical to the main HPA at a structural (53.8%) or lobar (96.1%) level. This situation of a contralateral HPA strictly symmetrical to the SOZ or to the main HPA leads to the novel concept of a HPA mirror image and may lead to improving the localizing value of ictal SPECT. SPECT mirror images occurred less frequently in mTLE with HS (37.5%) than in mTLE without HS (71.4%) and latTLE (57.1%) or extratemporal epilepsies (66.7%).

Clinical contribution of mirror images

Ictal SPECT studies struggle to distinguish the SOZ from propagations, specifically when hyperperfusions are of large volume or when they are multiple. SPECT mirror images, which presumably reflect contralateral ictal propagations, are located in a homotopic fashion in the contralateral hemisphere. In nine cases, they improved SOZ location, permitting by inverse symmetry the restriction of the site of the SOZ to a part of a larger HPA, surgically realistic, or to one of several HPAs. Mirror images usually confirmed SOZ location for mTLE patients, with or without HS, but could miss it or, in one case, mislateralize. For latTLE patients, the mirror image improved SPECT analysis in half of the cases, narrowing the SOZ location, but was misleading once. In extratemporal epilepsies, its occurrence aided SOZ location in three cases, especially when several or wide areas were candidates for seizure generation. Further, contralateral HPAs are not associated with a poorer postsurgical outcome. Contralateral HPAs, improving SOZ location and therefore presurgical workup, may even lead to a clinical benefit for patients.

Occurrence of contralateral HPAs

Although our study reports frequent contralateral hyperperfusions, they have not previously been systematically characterized, and although mirror-image HPAs have been described, they have not been studied in detail. Most previous work has focused on the hyperperfusion related to the SOZ, and contralateral HPAs have been discarded as “unlateralized examinations” without providing an exact location. Table 2 summarizes reports of contralateral HPAs in ictal SPECT studies. Technical advances may partly explain the high frequency with which we observed contralateral HPAs. New-generation high-resolution gamma cameras were used, and the SISCOM method used to process our data may be more efficient than side-by-side comparison of ictal and interictal scans (17). With comparable techniques, contralateral symmetrical HPAs have already been reported, called hyperperfusion-plus, in 18.6% of TLE (56), a lower frequency than in our study. Such variation might be explained by the fact that we systematically visually searched for a contralateral HPA in the cortex homotopic to the SOZ. Potentially, in other studies, quantitative analyses of scans that compare symmetrical regions of interest to detect lateralized hyperperfusions, generating asymmetry indexes, may mask weaker contralateral HPAs.

Table 2. Previous descriptions of contralateral hyperperfusions
SOZ locationProportion of hyperperfusion contralateral to the main HPARemarkReference
  1. SOZ, seizure onset zone; HPA, hyperperfusion area; HS, hippocampal sclerosis; LatTLE, Lateral TLE, mTLE, mesial TLE; TLE, temporal lobe epilepsy; SISCOM, substracted ictal SPECT Coregistred to MRI.

TLE1/28Ictal and postictal SPECT(11)
TLE2/151 patient with nonlesion and 1 patient with bilateral HS(15)
No bitemporal hyperperfusion mentioned in 119 TLE cases 
TLE2/19 bilateral (55)
6/19 contralateral only 
TLE7/466/6 in bitemporal epilepsies(22)
TLE2/10 mTLE with HS (30)
1/8 mTLE with lesion 
5/7 latTLE with lesion 
TLE5/40 mTLE with HS63 patients after surgery(23)
3/10 mTLE with lesion 
1/13 TLE without lesionEquivalent outcome between unilateral and bilateral HPA 
4/16 latTLE 
TLE11/537 bitemporal HPA and 4 contralateral HPA(25)
TLE4/55No contralateral HPA in 11 extratemporal epilepsies(21)
TLEAura only: 2/5TLE with good postoperative outcome(57)
Motionless staring: 3/7SISCOM 
Automatism: 15/17 
Dystonic posturing: 9/13 
Secondary generalization: 6/8 
Frontal lobe epilepsy: Supplementary motor area4/8Asymmetrical projection(58)
Frontal lobe epilepsy3 orbitary, 1 mesial, 1 dorsolateralSymmetry unknown(59)
Posterolateral epilepsies2/5Symmetrical to ZE(60)
Neocortical extratemporal epilepsies3/42 contralateral HPA without Main HPA in the SOZNo data on propagations in cases of matching main HPA and SOZ(61)
Neocortical epilepsies4/27SISCOM(62)
Neocortical epilepsies1/64 (TLE patient)SISCOM(63)
Diverse focus epilepsies1/53 (location of the SOZ unknown)SISCOM(50)

Anatomic substrate of mirror images

The high proportion of contralateral HPAs agrees with the observation that ictal SPECT tends to report events during a period of 10 to 20 s, including seizure onset and initial propagation. Seizure-propagation patterns depend on anatomic connectivity, which differs between mesial temporal cortex and neocortical areas.

Homotopic cortical areas other than the temporal lobe are connected directly via the corpus callosum (34). Commissural interconnections between temporal lobes are more various and less straightforward for temporolimbic structures in humans. They include the presumably weak direct dorsal hippocampal commissure (35–38) and the stronger indirect anterior commissure and corpus callosum (35), which relay in entorhinal and perirhinal cortices.

This indirect pathway for contralateral transmission from the hippocampus proper might explain the weaker symmetry of contralateral HPAs associated with seizures in patients with mTLE. In half of the seizures examined, contralateral HPAs were located in the amygdalohippocampal formation and in the polar or anteromesial cortex for the other half. In contrast, the mirror image was often strictly symmetrical to the SOZ in nonhippocampal epilepsies including latTLE and extratemporal cases. We assume that direct commissural fibers mediate contralateral propagation of these seizures.

Furthermore, interhemispheric propagation of neocortical partial epileptic seizures occurs very quickly (39), whereas seizures of mTLE origin associated with HS spread more slowly to the contralateral hemisphere (40), with a mean delay of 29 s. This difference might explain the low occurrence of contralateral HPAs in mTLE with HS (50%) compared with that of other partial epilepsies (87%).

Physiological basis of mirror images

Both clinical and physiological data suggest that contralateral cortex is recruited during ictal discharges in focal epilepsies.

In animal experiments, ictal discharges propagate preferentially to the contralateral homotopic cortex (41,42). More recent techniques, including optical imaging of the whole brain (43) and a novel in vitro preparation comprising both hippocampi and their linking commissure (44), confirm this propagation pattern. Human ictal discharges regularly spread to the contralateral hemisphere (5,26,45,46), and contralateral hyperperfusion during seizures has been detected with other methods, such as 15O-H2O PET (47) and a thermal diffusion device (48).

We could not assess the physiological basis of contralateral HPAs in detail, because intracranial recordings were made from the cortical area symmetrical to the SOZ in only two cases (patients 11 and 14). Both cases revealed a propagation of the ictal activity and a SPECT hyperperfusion to the region homotopic to the SOZ. Because ictal SPECT studies associated with simultaneous intracranial recordings (49,50) have shown that focal hyperperfusions occur specifically in areas in which an epileptic fast discharge was recorded, mirror images seem likely to reflect the spread of ictal activity.

Contralateral spread of ictal activity was largely restricted to a region homotopic to the SOZ. We can make a parallel with the mirror focus (24,51), which has the same spatial pattern (52). As for the mirror focus, surround inhibition (43), consisting of the activation of local interneurons to form a “ring” of decreased activity (53), could act to restrict seizure spread away from the mirror site in contralateral cortex (54). Although mirror images and mirror focuses are both spatially symmetrical, they differ in that a mirror image reveals propagation of epileptic discharges, whereas a mirror focus represents a novel epileptogenic zone. The frequency of mirror foci is difficult to assess in humans and varies between studies (24,52), but is certainly lower than that of contralateral HPAs. None of the patients in our study had a clear mirror focus.

In conclusion, this study characterized ictal SPECT contralateral mirror images. They presumably derive from changes in blood flow related to seizure propagation. Knowledge of their spatial characteristics appears useful to interpret SPECT examinations in that, although rarely misleading, they may permit, by a reverse symmetry, a more precise localization of the SOZ.


Acknowledgment:  We are grateful to Dr. Richard Miles, Dr. Franck Semah, and Marion Noulhiane for helpful suggestions and to Dr. Nathalie Kubis for her assistance in statistical analysis.