Address correspondence to Seung Bong Hong, M.D., Ph.D., Department of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Korea. E-mail: email@example.com, firstname.lastname@example.org
Purpose: Ictal single-photon emission computerized tomography (SPECT) is often nonlocalized in patients with partial epilepsy. We repeated ictal SPECT in patients with partial epilepsy whose first ictal SPECT was nonlocalized. We also performed subtraction ictal SPECT coregistered to magnetic resonance imaging (MRI) (SISCOM) to test the localizability of ictal SPECT.
Methods: We recruited 69 patients with partial epilepsy (33 male and 36 female, mean plus or minus standard deviation age 29.5 ± 12.2 years), who had a repeated ictal SPECT. Ictal-interictal SPECT subtractions were performed, and the subtracted SPECTs were coregistered with their brain MRI studies. SISCOM results were considered to be localizing when the results were concordant with the final location of the epileptic focus, as determined by the presurgical evaluation. We compared seizure duration, tracer injection time, interictal and ictal scalp electroencephalography (EEG) patterns, presence and time of secondary generalization, and epilepsy classification between the localized and nonlocalized SISCOM groups.
Key Findings: The SISCOM results of the second ictal SPECT were localized in 43 (62.3%) patients and nonlocalized in 26 (37.7%) patients. In the second ictal SPECT, the radiotracer injection time was significantly shorter in the localized group (25.1 ± 8.9 s), as compared to the nonlocalized group (49.2 ± 55.8 s) (p = 0.008). Furthermore, the radiotracer injection time of the second ictal SPECT was significantly shorter than the first ictal SPECT, only in the localized group (36.8 ± 23.8 s in the first and 25.1 ± 8.9 s in the second ictal SPECT in the localized group, p = 0.004). The percent injection time ([(tracer injection time−seizure onset time)/total seizure duration] × 100%) in the second SPECT was significantly shorter in the localized group, as compared to the nonlocalized group (37.9 ± 23.0% in the localized group and 72.3 ± 46.2% in the nonlocalized group, p < 0.001). The localized ictal EEG patterns at the time of injection were more frequent in the localized SISCOM group. The secondary generalization of seizures at the time of injection was more frequent in nonlocalized groups.
Significance: Repeated ictal SPECT with SISCOM analysis is helpful for localizing an epileptic focus in patients with partial epilepsy who have a nonlocalized first ictal SPECT. The most important factor for increasing the localizability of repeated ictal SPECT is early injection time and a localizing ictal EEG pattern at the time of radiotracer injection.
Brain single-photon emission computed tomography (SPECT) has been used for localizing seizure focus in patients with partial epilepsy (Weinand et al., 1997; Kaiboriboon et al., 2002; Pastor et al., 2008; Matsuda et al., 2009). Ictal SPECT is performed by injecting radiotracer intravenously during seizures, and has shown high sensitivities for the localization of epileptic focus. Ictal SPECT shows the snapshot of cerebral blood flow change during seizure. But the result of ictal SPECT is not always localizing in a considerable number of patients with partial epilepsy. Various factors affect the result of ictal SPECT. The early injection of radiotracer is important for the successful localization of epileptic focus (Lee et al., 2000c, 2006; Hong et al., 2008). Injection prior to secondary generalization also enhanced the localizability of ictal SPECT (Lee et al., 2006). Duration of seizures, propagation speed of seizures, and pathway of ictal discharge spreading are also important factors for the seizure localization of ictal SPECT (Lee et al., 2000c). Furthermore, locations of epileptic foci may affect ictal SPECT results. Propagation speeds of ictal discharges in lateral temporal lobe epilepsy (TLE) and occipital lobe epilepsy (OLE) are slower than those in frontal lobe epilepsy and parietal lobe epilepsy (FLE; PLE) (Lee et al., 2000a, 2006).
SISCOM compares ictal SPECT image with the same patient’s interictal SPECT image, and produces difference image between the two SPECTs. Theoretically SISCOM is expected to reveal cerebral perfusion changes during seizure more accurately than the visual inspection of ictal SPECT. Several studies reported that SISCOM significantly increased the sensitivity of ictal SPECT and provided more accurate anatomic localization of seizure with MRI (O’Brien et al., 2000; Kakisaka et al., 2009).
We reviewed retrospectively brain SPECT results in patients with partial epilepsy whose first ictal SPECT failed to localize epileptic focus and then had a repeated ictal SPECT. SISCOM was performed with the first and the second ictal SPECT images. We evaluated the sensitivity and usefulness of the second ictal SPECT by SISCOM analysis and investigated various factors that might have affected the results of the second ictal SPECT.
Subjects and Methods
We recruited 69 patients (33 male and 36 female; mean age 29.5 ± 12.2 years) with intractable partial epilepsy, who underwent a repeated ictal SPECT because the first ictal SPECT failed to localize an epileptic focus. We excluded patients who showed nonlocalized or multiregional ictal-onset zones in presurgical evaluation. All patients underwent presurgical evaluation at the Epilepsy Center of Samsung Medical Center in Seoul, Korea, including clinical examination, video–electroencephalography (EEG) monitoring, brain MRI, ictal and interictal SPECT, and 18F-fluorodeoxyglucose–positron emission tomography (FDG-PET). In addition, we investigated the patients’ histories of febrile convulsions and the clinical features of their seizures. Among them, 32 patients had an epilepsy surgery after the presurgical evaluation. Postsurgical outcome for more than one (1.6–9.2 years) year after surgery was evaluated according to the modified Engel classification (Engel et al., 1993) (Table 1).
Table 1. Demographic and clinical features of the patients
All (n = 69)
29.5 ± 12.2
Gender (ratio of men to women)
Epilepsy syndrome (number of patients)
Temporal lobe epilepsy
Hippocampal sclerosis (HS) (%)
Cavernous angioma (%)
Cortical dysplasia (%)
History of febrile convulsion (%)
Extratemporal lobe epilepsy (FLE/PLE)
Cortical dysplasia (%)
History of febrile convulsion (%)
Seizure analysis and EEG interpretation
The results of video-EEG monitoring were interpreted by two epileptologists (SBH and EYJ). The final conclusion of discordant interpretations was made by discussion and agreement of two doctors. Seizure onset was determined by identifying the earliest ictal EEG changes from baseline that were associated with the seizure or the earliest alteration of behavior or impaired awareness. The end of a seizure was defined as the cessation of ictal discharges or ictal movement and behavior. We also investigated the clinical features of the seizures and the presence of secondary generalization at the time of radiotracer injection.
Ictal EEG patterns were characterized as regional, ipsilateral hemispheric, contralateral hemispheric, bilateral, or diffuse. Ictal EEG patterns at seizure onset and at the time of radiotracer injection were investigated.
The MRI scans were performed with GE Signa 1.5 or 3.0 Tesla scanners (GE Medical Systems, Milwaukee, WI, U.S.A.). The spoiled gradient recalled (SPGR) MRI was scanned with the parameters of no gap, 1.6-mm thickness, TR (repetition time)/TE (echo time) 30/7, flip angle = 45 degrees, 1 NEX (number of excitation), coronal. The fluid-attenuated inversion recovery (FLAIR) MRI was scanned with the parameters of 1.0-mm gap, 4.0-mm thickness, TR /TE 10002/127.5, 1 NEX, oblique coronal. The T2-weighted MRI was scanned with the parameters of 0.3-mm gap, 3.0-mm thickness, TR /TE 5300/99, flip angle = 95 degrees, 3 NEX, oblique coronal.
Brain SPECT scans were performed within 30–60 min after radiotracer (99mTc-ECD∼25mCi) injection using a three-headed Triad XLT system (Trionix Research Laboratory, Twinsburg, OH, U.S.A.) equipped with high-resolution collimators. Radiotracer for ictal brain SPECT was injected during seizures, whereas interictal SPECT was performed when the patient had documented no seizure activity for at least 24 h.
The SISCOM was processed using offline SUN Ultra 1 Creator Workstation (Sun Microsystems, Santa Clara, CA, U.S.A.) with ANALYZE 7.5 (Biomedical Imaging Resource, Mayo Foundation, Rochester, MN, U.S.A.). All biomedical images were transferred from each scanner console to the UNIX workstation by a 4-mm DAT (digital audio tape) device. The SPECT subtraction was processed using the following sequence. The detailed description of each step was published for previous studies (Lee et al., 2000a; Shin et al., 2002; Joo et al., 2004).
1 Ictal-interictal SPECT registration.
2 Normalization of radioisotope uptake level.
3 Ictal transformed interictal SPECT subtraction.
4 Noise erasing.
5 Registration of MRI and subtracted SPECT.
Interpretation of SISCOM
Two epileptologists (SBH and EYJ), blinded to the patient’s clinical information and results of other presurgical evaluations, reviewed the SISCOM results. The two reviewers independently reviewed images. If the two reviewers disagreed, the final decision was made by discussion and agreement of two reviewers. Ictal hyperperfusion of subtracted SPECT was considered significant only when the regional cerebral blood flow (rCBF) difference in each pixel of the brain SPECT image between ictal and interictal states was >2 SD (standard deviations). We analyzed the presence of ictal hyperperfusion in the following anatomic regions: orbitofrontal, medial frontal, dorsolateral frontal, insular cortex, cingulate gyrus, basal ganglia (BG), thalamus, temporal lobe (anterior, mesial, basal, lateral, posterior), parietal lobe (superior, inferior), and occipital lobe (medial, lateral). SISCOM results were considered as localized when they were concordant with the final location of the epileptic focus, as determined by the presurgical evaluation (Fig. 1). In the event of two lobar patterns, if predominant hyperperfusion was located in the same region as epileptic focus, it was considered as localized.
Injection of radiotracer
Seizure onset was defined as the time when the earliest ictal EEG changes or ictal behaviors were observed. The time of the injection was defined as the time when the plunger on the syringe containing the radiotracer was fully depressed. To estimate how early radiotracer was injected compared to total seizure duration, the percent injection time was calculated (% injection time = [(radiotracer injection time−seizure onset time)/total seizure duration] × 100%). In addition, we compared the tracer injection time with the time of secondary generalization in both localized and nonlocalized SISCOM groups.
Statistical analysis was performed using SPSS version 13.0 for Windows (SPSS Inc., Chicago, IL, U.S.A.), and the significance level was set at p < 0.05. Fisher’s exact test was used to compare categorical data. Comparison of seizure duration and radiotracer injection time between the localized group and the nonlocalized group was assessed using the nonparametric, Mann-Whitney U-test. The Wilcoxon signed rank test was used to compare seizure duration and radiotracer injection time between the first and second ictal SPECTs of each group.
Localizability of SISCOM with the second ictal SPECT
Among the 69 patients whose first SISCOM failed to localize an epileptic focus, the second SISCOM was localized in 43 (62.3%) patients and nonlocalized in 26 patients (37.7%). Six patients (four TLE, two FLE) showed a false localization in the first SISCOM and three (two TLE, one FLE) in the second SISCOM results. The ictal EEG patterns of patients with a false localization at the time of radiotracer injection were bilateral or ipsilateral hemispheric.
Thirty-two patients had epilepsy surgery after the presurgical evaluation: 19 in the localized group and 13 in the nonlocalized group (Fig. 2). The localized group showed significantly better surgical outcomes compared to the nonlocalized group (p < 0.05; Engel class I = 14. Class II = 5 in the localized group versus class I = 7, class II = 1, class III = 1, and class IV = 4 in the nonlocalized group).
Clinical factors related to the localizability of repeated SISCOM
Clinical factors such as age, gender, presence of MRI lesion, and histopathology did not significantly affect the localizability of the second SISCOM. The seizure classifications between the localized and nonlocalized groups were not significantly different (Table 2). The presence of secondary generalization at the time of radiotracer injection were more frequent in the nonlocalized group [four (9.3%) secondary generalization in the localized group vs. six (23.1%) in the nonlocalized group]. However, seizure duration did not differ significantly between the localized and the nonlocalized groups (Fig. 3A).
Table 2. Location of epileptic focus of the localized and the nonlocalized groups
Location of epileptic focus
Localized (n = 43)
Nonlocalized (n = 26)
There was no statistically significant difference in locations between the two groups (p = 0.57).
Radiotracer injection was completed before the termination of seizure activity in all patients. The radiotracer injection time during the second ictal SPECT was significantly shorter in the localized group (25.1 ± 8.9 s), as compared to the nonlocalized group (49.2 ± 55.8 s) (p = 0.008). Furthermore, the radiotracer injection time during the second ictal SPECT was significantly shorter than the time during the first ictal SPECT, only in the localized group (36.8 ± 23.8 s during the first and 25.1 ± 8.9 s during the second ictal SPECT in the localized group, p = 0.004) (Fig. 3B). The percent injection time during the second SPECT was also shorter in the localized group than in the nonlocalized group (37.9 ± 23.0% in the localized group vs. 72.3 ± 46.2% in the nonlocalized group, p < 0.001) (Fig. 3C). The injection time in the second ictal SPECT from secondary generalization (injection time−secondary generalization time) was shorter in localized group than in nonlocalized group (−2.0 ± 0.7 s in localized group; 17.4 ± 13.3 s in the nonlocalized group, p = 0.025).
Ictal EEG pattern
The ictal EEG patterns at the time of injection were more frequently localizing in the localized SISCOM group (63.1% in the localized group vs. 42.3% in the nonlocalized group, p < 0.01) (Table 3).
Table 3. The ictal EEG patterns at the time of radiotracer injection. Localized ictal EEG patterns were more frequently observed in the localized group than in the nonlocalized group (p < 0.01) in the 2nd SISCOM
Ictal EEG pattern at the time of injection
SISCOM, subtracted ictal SPECT coregistered to MRI.
Bilateral or diffuse
Noninvasive functional neuroimaging, including SPECT and PET, provides important information for the identification of epileptogenic foci in patients with intractable partial epilepsy (Lee et al., 2006). Brain SPECT reveals cerebral perfusion changes in the ictal and interictal states. Ictal SPECT compared to icterictal SPECT typically demonstrates areas of regional hyperperfusion, which reflects increased neuronal activity in parts of the brain involved in seizure activity (Lee et al., 2000b, 2002; Pastor et al., 2008; Wehner & Luders, 2008). Because these modalities are not dependent on the electrical activity of the brain, they may be useful means for the localization of epileptic foci when EEG-based methods fail to localize epileptic foci (la Fougere et al., 2009). Previously the clinical utility of ictal-interictal subtraction SPECT for the localization of epileptic foci has been reported (Lee et al., 2000a). Visual comparison of interictal and ictal SPECT without subtraction may be difficult because of different slice levels, different orientations, and the low-resolution power of SPECT. In this respect, SISCOM may be superior to the visual inspection of ictal SPECT. However, one must use caution when interpreting ictal SPECT and SISCOM. The radiotracer is injected after a seizure starts and takes time to reach the brain. Despite efforts to shorten the time delay, periictal propagation of seizure activity complicates the interpretation of results of ictal SPECT (Ho et al., 1996; Lee et al., 2002). Consequently, SISCOM shows increases in rCBF that are associated with both the epileptic foci and the propagation areas. If the time delay between seizure onset and radiotracer injection is too long, false lateralization or localization of SISCOM due to the postictal switch phenomenon or the excitation of propagated brain region can occur (Pastor et al., 2008). Furthermore, ictal SPECT may fail to show an area of significant hyperperfusion, giving rise to false-negative results (Newton et al., 1992). Therefore, early injection of radiotracer is a very important factor for the accurate localization of epileptic foci.
We did not find a significant difference in the seizure localization of SISCOM according to the location of epileptic foci. In other studies, the localizability of SISCOM is higher in TLE and occipital lobe epilepsy (OLE), because the propagation of ictal discharges are slower in TLE and OLE than in frontal lobe epilepsy and parietal lobe epilepsy (FLE and PLE) (Lee et al., 2006). However, our present study included only those patients with TLE whose first ictal SPECT was nonlocalized. Because of its low temporal resolution, ictal SPECT shows characteristic ictal hyperperfusion patterns or often contains both the ictal-onset zone and propagation pathways. These patterns often have a multilobulated “hourglass” appearance, i.e., the two lobules are connected by a trail of hyperperfusion, propagation trail (Dupont et al., 2006; Van Paesschen et al., 2007). This spatial feature is the signature of ictal propagation. Ictal SPECT of seizure with faster propagation shows longer propagation trail. Furthermore, seizure with faster propagation speed may show no hyperperfusion within or immediately neighboring, but rather at a distance from the epileptogenic lesion.
We considered ictal hyperperfusion of subtracted SPECT as significant only when the rCBF difference in each pixel of the brain SPECT image between ictal and interictal states was >2 SD. One study (Kaiboriboon et al., 2002) reported that differences in optimal thresholding of subtraction images are likely related to the technical factors of SPECT acquisition and the experience of individual interpreter. Use of a triple-headed SPECT camera is capable of yielding a higher resolution images and may allow interpretation of SISCOM images at lower thresholding values.
Although the presence of secondary generalization during ictal SPECT did not significantly affect the result of SISCOM in our study, the start time of secondary generalization minus the injection time was significantly shorter in the localized group. Secondary generalization usually results in widespread ictal discharges over the entire brain and diffuse perfusion changes. Injection prior to secondary generalization increases the localizability of SISCOM, as compared with injection during secondary generalization (Lee et al., 2006). However, given that it takes about 30–60 s for radiotracer to reach the brain after injection (Rowe et al., 1991; Newton et al., 1992; Chassagnon et al., 2006), injection prior to secondary generalization may still fail to localize the epileptogenic foci. The total duration of a seizure is also an important factor for the localization of epileptic foci. Consequently, seizure classification and connectivity between the epileptic foci and adjacent area may play an important role in the localizability of ictal SPECT and SISCOM.
In our study, ictal EEG patterns at the time of radiotracer injection are classified as regional, ipsilateral hemispheric, contralateral hemispheric, bilateral, or diffuse, according to the distribution of ictal EEG discharges. The localized SISCOM group had more frequent localized ictal EEG patterns, whereas the nonlocalized group showed more bilateral ictal EEG patterns at the time of radiotracer injection. Ictal SPECT reflects alterations of rCBF related to the electrical activity of the brain when the radiotracer reaches the brain. Therefore, more localized ictal EEG patterns at the injection time may predict higher localizability of ictal SPECT. But ictal SPECT and SISCOM should be interpreted with caution because it usually takes up to 60 s for the radiotracer to reach the brain (Lee et al., 2000a; O’Brien et al., 2000). Although injection times of <60 s from the onset of ictal discharges were found to be sufficient to catch hyperperfusion at the epileptogenic zones (Rowe et al., 1991; Newton et al., 1992; Lee et al., 2002), the propagation rate of ictal discharges to the adjacent brain area, location of ictal onset zone, intensity of ictal discharges, and metabolic status of the epileptogenic focus are all important factors that affect the localizability of ictal SPECT and SISCOM (Lee et al., 2000c).
This research was supported by a grant (#2010K000817) from Brain Research Center of the 21st Century Frontier Research Program funded by the Ministry of Education, Science and Technology, the Republic of Korea and by a grant (no. A110097) of the Good Health R&D Project, Ministry of Health & Welfare, Republic of Korea.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journalós position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.