Address correspondence and reprint requests to Dr. W. Van Paesschen at Department of Neurology, UZ Gasthuisberg, 49 Herestraat, 3000 Leuven, Belgium. E-mail: Wim.vanpaesschen@uz.kuleuven.ac.be


Summary:  The localizing value of ictal single-photon emission computed tomography (SPECT) performed with cerebral blood flow agents in patients with epilepsy is based on cerebral metabolic and perfusion coupling. Ictal hyperperfusion is used to localize the epileptogenic zone noninvasively, and is particularly useful in magnetic resonance (MR)-negative partial epilepsy and focal cortical dysplasias. Subtraction ictal SPECT coregistered with MRI (SISCOM) improves the localization of the area of hyperperfusion. Ictal SPECT should always be interpreted in the context of a full presurgical evaluation. Early ictal SPECT injections minimize the problem of seizure propagation and of nonlocalization due to an early switch from ictal hyperperfusion to postictal hypoperfusion during brief extratemporal seizures. The degree of thresholding of SISCOM images affects the sensitivity and specificity of ictal SPECT. Ictal hypoperfusion may reflect ictal inhibition or deactivation. Postictal and interictal SPECT studies are less useful to localize the ictal-onset zone. Statistical parametric mapping analysis of groups of selected ictal–interictal difference images has the potential to demonstrate the evolution of cortical, subcortical, and cerebellar perfusion changes during a particular seizure type, to study seizure-gating mechanisms, and to provide new insights into the pathophysiology of seizures.

Ictal single-photon emission computed tomography (SPECT) has the potential to localize the ictal-onset zone accurately in a noninvasive manner. Reliably to deliver early ictal SPECT injections, detailed attention should be paid to the logistics of ictal SPECT setup. Ictal SPECT injections should be performed in the video-EEG suite, with the nursing and review station close to the rooms of the patients. Medical personnel should be educated in handling of radioactive materials and be familiar with the electroclinical features of epileptic seizures. The brain-perfusion agent should be available in the room, and the injection system should allow fast ictal injections (1,2). High-resolution SPECT and magnetic resonance imaging (MRI) scanner should be available. Excellent cooperation between the neurology and nuclear medicine department is of crucial importance (3). If the implementation of ictal SPECT is too difficult, referral of selected patients for ictal SPECT should be considered.

Brain-perfusion tracers are lipophilic substances that cross the blood–brain barrier, have a long retention time in the brain, and are 99mtechnetium (99mTc)-labeled agents. Two commonly used tracers are 99mTc-hexamethylene propylene amine (99mTc-HMPAO) (Ceretec) and 99mTc-ethyl cysteinate dimer (99mTc-ECD) (Neurolite). The first-pass extraction for 99mTc-ECD is ∼60%, and for 99mTc-HMPAO, ∼85%. 99mTc-ECD is retained in the brain after an enzymatic conversion to ionized acid compounds and 99mTc-HMPAO after conversion to a nondiffusible hydrophilic compound after cell uptake. These different mechanisms of brain retention could explain the differences in cerebral distribution of the two tracers (4). 99mTc-ECD is stable 6 to 8 h, and the stabilized form of 99mTc-HMPAO, for 4 h. 99mTc-ECD is cleared from the body more rapidly than 99mTc-HMPAO, and gives a higher brain-to–soft tissue activity ratio, which improves image quality (5). Lee and colleagues (6) found 99mTc-HMPAO ictal SPECT superior to 99mTc-ECD ictal SPECT in localizing the epileptogenic zone. In their study, however, the number of patients treated with 99mTc-ECD was rather small, and the duration of the injected seizures was not given, which could potentially have influenced the results. Further studies will be required to settle the issue. Because early ictal injections give the best results, it is advisable to use a stabilized compound and to inject via an indwelling intravenous cannula in the arm that is involved less in the seizure. Smith and colleagues (7) reported ictal SPECT injections in 77% of 110 consecutive patients, with a medium injection time of 27 s. In 160 consecutive patients admitted to our video-EEG suite, we were able to inject 80% of patients, with a median injection time of 19 s, by using an ictal SPECT setup with indwelling intravenous cannula (2). In view of the decay of the tracer, the injected dose should be adjusted. Alternatively, if the full dose is given, the scanning time should be adjusted. In 20% of our patients, ictal SPECT injection was not performed because of seizures that occurred outside the hours of ictal SPECT (59%), because the patient did not have seizures (23%), or because the seizure was not noticed by the medical personnel (18%). A solution for these problems could be an on-call service for selected patients to allow ictal injections during the night, and a seizure-warning system.

Both interictal and ictal SPECT images should be obtained and coregistered. To adjust for differences in administered dose during interictal and ictal studies, the interictal and ictal SPECT scans are normalized. Difference images are obtained by subtraction of the interictal from the ictal SPECT. The difference images are thresholded, usually at 2 standard deviations (SDs), to highlight regions of hyperperfusion. Subtraction ictal SPECT coregistered to MRI (SISCOM) has improved localization and visualization of the region of hyperperfusion (8). O'Brien and colleagues (9) reported localization in 39% by using side-by-side visual inspection versus 88% localization by using SISCOM (Fig. 1).

Figure 1.

Improved localization of the area of ictal hyperperfusion with SISCOM. The area of ictal hyperperfusion was missed on side-by-side visual analysis of ictal (A) and interictal (B) SPECT images. Subtraction of interictal from ictal SPECT (threshold, +2 SD) clearly showed the area of ictal hyperperfusion (C). SISCOM allowed an accurate anatomic localization of the ictal hyperperfusion in the brain (D).

Statistical parametric mapping (SPM) between interictal and ictal SPECT scans has been reported to give results comparable with difference imaging (10). With SPM, the ictal SPECT scan can be compared with a normal brain SPECT database, without the need for an interictal SPECT scan (11).

The interpretation of ictal SPECT images should always be done in the context of a full presurgical evaluation. The neurologist/epileptologist has, therefore, an important role in the interpretation of the SPECT images. The injection time should be known, because early injections give the best results. The area of highest ictal hyperperfusion is usually the ictal-onset zone, unless the seizure has propagated. Several propagation patterns have been described. Propagation is often from posterior brain regions (parietooccipital lobes) to anterior brain regions (temporal and frontal lobe) (12). Noachtar and colleagues (13) reported propagation in 85% of parietooccipital epilepsy (Fig. 2). Another propagation pattern is from the temporal to the frontal lobe. In patients with a temporal lobe lesion on MRI and discordant frontal lobe seizures, ictal SPECT may show propagation from temporal to frontal lobe, obviating the need for invasive monitoring. Propagation from one temporal lobe to the contralateral temporal lobe has been reported in ∼1% of cases (14,15). Propagation of ictal activity can partly explain why a high SISCOM threshold has a lower sensitivity and higher specificity compared with a low threshold, which has a higher sensitivity but lower specificity (16) (Fig. 3). In clinical practice, using different SISCOM thresholds can help elucidate propagation patterns (see Fig. 2). The injected seizure type and ictal semiology should be known for a correct interpretation of ictal SPECT. In our hands, ictal SPECT during simple partial seizures gave no information in ∼40% of cases. For this reason, we have limited the use of self-injection ictal SPECT, because this was often during isolated simple partial seizures (15). Complex partial seizures (CPSs) give the best results, and secondarily generalized seizures may give multiple regions of hyperperfusion (17). The duration of the injected seizure is important in the interpretation of ictal SPECT studies. After injection in an arm vein, the tracer takes ∼30 s to reach the brain. The postictal switch (i.e., switch from ictal hyperperfusion to postictal hypoperfusion) occurs ∼1–2 min postictally in temporal lobe seizures (18), but is shorter in extratemporal seizures. It has been estimated that extratemporal seizures should last ≥10–15 s after ictal SPECT injection to give localizing information (16).

Figure 2.

Propagation of ictal activity. A: FLAIR-hyperintense focal cortical dysplasia (FCD) in the left posterior temporal lobe. Irritative zone and ictal-onset zone on EEG were at T5, concordant with the MR-visible FCD. B: Outline of FCD in green. C, D: SISCOM (threshold, +2 SD) showed an area of hyperperfusion partially overlapping the FCD. The injected seizure was a complex partial seizure that lasted 53 s. The injection was given 30 s after seizure onset. E: SISCOM (threshold, +2 SD) showed the highest hyperperfusion in the left temporal lobe. F: SISCOM (threshold, +1 SD) showed an area of hyperperfusion extending toward the FCD (white arrow) not visible at a threshold of +2 SD in E. G: SISCOM (threshold, +1 SD) at the level of the FCD showed an area of hyperperfusion extending from the FCD toward the temporal lobe (white arrow). These findings are consistent with an ictal onset in the FCD with ictal propagation toward the temporal lobe.

Figure 3.

Thresholding of SISCOM images and the sensitivity and specificity of ictal SPECT. FLAIR showed a hyperintense FCD in the left frontal lobe. The injected seizure was a partial seizure that lasted 15 s. The injection was given 3 s after seizure onset. The SISCOM thresholded at 1 SD gave the largest area of hyperperfusion, which most likely included the ictal-onset zone. The area of hyperperfusion on the 1 SD image extending in the right frontal lobe (white arrow) probably represented propagated seizure activity. Conversely, a high threshold of 4 SD gave a small region of hyperperfusion adjacent to the FCD. A high threshold is more specific (fewer false positives) in the detection of the ictal-onset zone, and a low threshold, more sensitive (fewer false negatives).

MR-negative partial epilepsy remains a difficult subgroup in terms of presurgical evaluation. Invasive EEG studies are usually indicated, and ictal SPECT findings may guide electrode placement. Pitfalls are that ictal SPECT may show propagated ictal activity and that ictal SPECT hyperperfusion does not exclude multifocal seizure onset (16). With ictal SPECT and invasive EEG studies, Siegel and colleagues (19) reported a good seizure outcome in 83% of patients with refractory MR-negative partial epilepsy. O'Brien and colleagues (20) reported an excellent outcome when SISCOM localization was concordant with surgical-resection site, but not when SISCOM and resection site were discordant in patients with nonlocalizing MRI findings. In MR-negative patients, SISCOM may be able to detect subtle focal cortical dysplasia (FCD) (21). Ictal SPECT in combination with 3-T MRI scanning appears particularly promising for the detection of subtle FCD (Fig. 4). Marusic and colleagues (22) and Boonyapisit and colleagues (23) reported FCD types with different functional characteristics. Focal cortical dysplasias with an increased fluid-attenuated inversion recovery (FLAIR) signal were characterized by balloon cells at the site of the increased signal and an overlying cortex that was not functional. The ictal-onset zone was not within the MR-visible FCD, but adjacent to it. Conversely, FCDs without increased FLAIR signal had the ictal-onset zone in the MRI-visible FCD, did not contain balloon cells, and the overlying cortex could be functional. Preliminary observations suggest that ictal SPECT may be able to demonstrate these characteristics noninvasively and may become a noninvasive marker of the epileptogenic zone in FCD.

Figure 4.

Detection of small FCDs by using ictal SPECT. The patient had refractory right parietal lobe epilepsy with sensory auras in the left arm. A 1.5-T MRI scan was normal. A 3-T MRI scan revealed a FCD at the place of ictal SPECT hyperperfusion. The injected seizure was a secondarily generalized tonic–clonic seizure that lasted 91 s. The injection was given 2 s after seizure onset. A: Subtracted ictal SPECT coregistered with FLAIR (threshold, +2 SD) showed hyperperfusion in the right parietal lobe. B: FLAIR without SPECT overlay showed a hyperintense lesion at the place of ictal SPECT hyperperfusion. C: T2-weighted MR image also showed the increased signal. D: A T1-weighted image showed blurring of the grey/white matter junction at the location of hyperperfusion. Surgery rendered the patient seizure free. Pathology confirmed the presence of a Taylor-type FCD with balloon cells.

Subtraction ictal SPECT coregistered to MRI usually highlights regions of hyperperfusion to detect the seizure-onset zone. When no threshold is applied in difference imaging, it is obvious that large areas of both hypo- and hyperperfusion are present (Fig. 5). To study these perfusion changes in a systematic way, SPM of ictal–interictal SPECT difference images of selected groups of patients can be used. Inclusion criteria such as seizure-onset zone, injected seizure type, epilepsy syndrome, and injection time of ictal SPECT should be used to obtain homogeneous groups. Interictal and ictal SPECT images are coregistered and transformed into the same reference space. A correction for differences in administered tracer dose is applied. Finally, SPM identifies regions with statistically significant increased or decreased perfusion during the ictal phase compared with the interictal phase. The aim of this type of study is to determine systematic perfusion changes during seizures and to understand better the pathophysiology of seizures.

Figure 5.

SISCOM image with and without thresholding. A: A SISCOM image thresholded at +3 SD of a patient with right temporal lobe epilepsy showed a region of hyperperfusion in the right temporal lobe. B: A SISCOM image without thresholding of the same patient showed a much more extensive region of hyperperfusion (in orange) surrounded by large regions of hypoperfusion (in blue).

With this method, CPSs in patients with unilateral hippocampal sclerosis have been studied (24,25) (Fig. 6). Ipsilateral temporal lobe hyperperfusion was present throughout the seizure but disappeared in the postictal period. Ipsilateral and also contralateral frontal lobe hypoperfusion was present both during the ictal and postictal period. Contralateral cerebellar hypoperfusion was present in the ictal period, and hyperperfusion in midline cerebellar structures, during the postictal period. Ipsilateral parietal lobe hypoperfusion was a late ictal phenomenon and was observed in ictal SPECTs with injection times ranging from 60 to 90 s. Bilateral medial thalamic hyperperfusion was observed postictally. Ictal frontal lobe hypoperfusion could represent an ictal surround inhibition. In favor of the latter hypothesis was the presence of a crossed cerebellar diaschisis, which has been shown to be due to deactivation of Purkinje cells caused by a decrease in excitatory input due to suppression of electrical activity in the contralateral frontal cortex (26). Further, these SPECT findings corroborate optical imaging experiments that showed a decrease in optical signal together with a decrease in neuronal activity in cortex surrounding an epileptic focus, consistent with ictal surround inhibition (27). The decrease of contralateral cerebellar perfusion in the ictal SPECT injection time window of 0–30 s was ∼10%. Gold and Lauritzen (26) showed that a major proportion of the basal cerebellar blood flow was independent of neuronal activity, and that a decrease of cerebellar perfusion on the order of 10–15% was associated with major suppression of electrical activity in the contralateral frontal lobe. Ictal surround inhibition is a defense mechanism against secondary generalization. Secondarily generalized tonic–clonic seizures show multilobar hyperfusion (17,28) and could represent failure of this ictal surround inhibition. Part of the ictal and postictal semiology of CPSs may be due to this ictal surround inhibition.

Figure 6.

Brain regions with significant ictal cerebral perfusion changes during CPS in 24 patients with hippocampal sclerosis (HS) are shown on a surface rendering of an MRI of the brain. A: Contralateral view. B: Ipsilateral view. Ictal hyperperfusion is in red, and ictal hypoperfusion is in blue. Ictal hyperperfusion was noted in the ipsilateral temporal lobe (1). Ictal hypoperfusion was present in the frontal lobes (ipsilateral>contralateral) (2) and contralateral cerebellum (3). The hypoperfusion of the ipsilateral frontal lobe probably represented an ictal surround inhibition, giving rise to a crossed cerebellar diaschisis in all patients (3). This figure was originally published in Brain (ref. 24).

Blumenfeld and colleagues (28) reported an SPM-ictal SPECT study of generalized tonic–clonic seizures during electroconvulsive therapy (ECT). Bilateral cerebellar and parietotemporal lobe hyperperfusion was observed during bilateral and right-sided ECT. Bilateral frontal hyperperfusion was present during bilateral ECT, and right frontotemporal hyperperfusion, during right-sided ECT. Bilateral cingulate hypoperfusion was present in bilateral ECT, and a left temporal lobe hypoperfusion, in right ECT. Blumenfeld and Taylor (29) postulated that abnormal increased activity in frontoparietal association cortices during secondarily generalized seizures and abnormal decreased activity in the same regions during CPSs may be the neural substrate of loss of consciousness.

In conclusion, ictal SPECT is able to demonstrate ictal neuronal activation and is a noninvasive marker of the ictal onset zone. The interpretation of ictal SPECT may be confounded by propagation of ictal activity or early switch from ictal hyperperfusion to postictal hypoperfusion during brief extratemporal seizures, which can be minimized by early ictal SPECT injections. Statistical parametric mapping analysis of ictal and interictal SPECT difference images of selected groups of patients is a promising method that highlights regions of significant hyper- and hypoperfusion and may provide new insights into the pathophysiology of seizures.


Acknowledgment:  This work was supported by Research Fund Katholieke Universiteit Leuven Interdisciplinair Onderzoeksprogramma (IDO 99/005). The following persons contributed to the present work: Patrick Dupont, Ph.D.; Ben Van Heerden, M.D.; Alex Maes, M.D.; Koen Van Laere, M.D.; Andre Palmini, M.D.; Hubert Van Billoen, Ph.D.; Guido Van Driel, R.N.; Beatrice Van Harck, R.N. Figures 2–5 were made by using MRIcro (http://www.psychology.nottingham.ac.uk/staff/cr1/mricro.html).