Concordance of MRI lesions and EEG focal slowing in children with nonsyndromic epilepsy


  • Byoungho H. Noh,

    1. Department of Pediatrics, Kangwon National University Hospital, Kangwon National University, Chuncheon, Gangwon, Korea
    Search for more papers by this author
  • Anne T. Berg,

    1. Epilepsy Center, Anne & Robert H. Lurie Children's Hospital of Chicago, Northwestern University, Chicago, Illinois, U.S.A
    Search for more papers by this author
  • Douglas R. Nordli Jr

    Corresponding author
    1. Epilepsy Center, Anne & Robert H. Lurie Children's Hospital of Chicago, Northwestern University, Chicago, Illinois, U.S.A
    • Department of Pediatrics, Kangwon National University Hospital, Kangwon National University, Chuncheon, Gangwon, Korea
    Search for more papers by this author

Address correspondence to Douglas R. Nordli, Jr., 225 E. Chicago Avenue, Epilepsy Box 29, Chicago, IL 60611, U.S.A. E-mail:



To investigate the significance of electroencephalography (EEG) focal slowing in children with epilepsy and to determine the correlation between focal slowing and focal lesions on brain magnetic resonance imaging (MRI).


We reviewed 5,149 EEG and 22,543 MRI reports for children who visited our institution from 2000 to 2010. Patients with nonsyndromic epilepsy (n = 253) were divided into groups with: focal slowing without any interictal epileptiform discharge (IED) (group 1); focal IEDs without focal slowing (group 2); focal slowing and focal IED (group 3); and normal findings (group 4). Focal slowing and MRI lesions were categorized by location, side, and depth.

Key Findings

We found MRI abnormalities in 59% of subjects in group 1, 56% in group 2, 74% in group 3, and 27% in group 4 (p < 0.0001). Cortical malformation (CM) was the most common pathology in groups with focal slowing. Focal slowing often correlated with the laterality of the MRI lesion (61–70%), but the location was concordant in only 40%. The associated lesions rarely were exclusively confined to the centrum semiovale (18%).


Focal slowing in children with epilepsy is highly associated with focal structural lesions on MRI, most commonly CM, and usually involves multiple layers. Focal slowing, as well as focal interictal epileptiform activity, is an important and useful EEG indicator of a brain structural abnormality in children with nonsyndromic epilepsy.

For >75 years, electroencephalographers have appreciated that focal slowing is an ominous feature in adults. Indeed, electroencephalography (EEG) began its clinical application by localizing cerebral tumors (Walter, 1937). This occurred only 8 years after the first report of the human EEG and 1 year after Lord Adrian predicted the potential of its clinical application (Adrian, 1936). Walter coined the term, “delta” to refer to the slowing that accompanied various pathologic states of brain. In 1937, Walter reported the concordance of EEG and surgical findings in 12 patients, mostly adults. Numerous studies followed his seminal findings, and Goldensohn (2005) showed that EEG can localize lesions in 68% of patients with brain tumors (Niedermeyer & Silva, 2005).

By the end of the 1980s, researchers reported correlations between focal slowing and lesions on computerized tomography (CT) mostly in adults (Gilmore & Brenner, 1981; Schaul et al., 1986; Marshall et al., 1988). Over time, less interest was paid to focal slowing as attention shifted to advanced technologies such as positron emission tomography (PET), magnetoencephalography, and computerized analysis. This may have led to a reduced emphasis on slowing in EEG reports.

Today, imaging has largely replaced EEG for localizing tumors, and conventional EEG is used primarily to diagnose and manage patients with epilepsy. In children with epilepsy, magnetic resonance (MR) studies show a higher percentage of cortical malformations than brain tumors (King et al., 1998; Berg et al., 2009). This tendency is especially high in younger children (Hsieh et al., 2010). These observations led us to investigate the significance of EEG focal slowing in children with epilepsy and specifically to determine the correlation between focal slowing and focal lesions on brain magnetic resonance imaging (MRI). This information could inform the clinical neurophysiologist about the modern clinical significance of this important EEG finding.

Materials and Methods


EEG reports, EEG data, and MRI (1.5 or 3 T) reports were retrospectively reviewed for children with nonsyndromic epilepsies who visited our institution from January 2000 to December 2010. All EEG data were obtained by long-term video-EEG monitoring technique. EEG reports were generated by certified pediatric electroencephalographers. Each EEG was reviewed again for research purpose by a board eligible pediatric electroencephalographer (BH Noh) who was blinded to MRI findings. All MRI reports were generated by certified pediatric neuroradiologists.

Searching technique

All relevant studies were initially ascertained through a computer-programmed searching technique using a visual basic application (VBA) and a keyword list. After programmed keyword searching, data in groups 1, 2, 3, and 4 were manually reviewed. Some subjects in the programmed categorized groups were reassigned to different groups or excluded after manual review.

Inclusion and exclusion criteria

From January 2000 to December 2010, there were 5,149 patients who had a long-term video-EEG and 22,543 who had a brain 1.5 or 3.0 T MRI at our center (Fig. 1). Of these, 2,571 children had both video-EEG and brain MRI. We excluded patients who had undergone any kind of previous invasive procedures, including recordings through Burr holes, cortical resections, and shunt operations (N = 154). In order to have a sample that was not biased toward abnormal MRI studies, we also excluded patients who have been diagnosed and treated with specific electroclinical syndromes that are associated with structural abnormalities and high degree of background slowing such as Ohtahara, West, and Lennox-Gastaut syndromes (N = 448), as well as children with epilepsy syndrome, which are commonly known as nonlesional, specifically childhood absence epilepsy (CAE), benign epilepsy with centrotemporal spikes (BECTS), juvenile absence epilepsy (JAE), and juvenile myoclonic epilepsy (JME). Patients who were determined to have nonepileptic events such as psychogenic seizure were similarly excluded (N = 326). Among the patients with nonsyndromic epilepsy (N = 1,643), the patients who had abnormal findings but no focal slowing or no focal IED were excluded (N = 1,390).

Figure 1.

Diagram of inclusion/exclusion criteria.

The remaining 253 patients were divided into the following four groups: group1, patients with epilepsy who had focal slowing but without interictal epileptiform discharges (IEDs); group 2, focal IED but without focal slowing; group 3, patients with both EEG focal slowing and focal IED; group 4, patients with epilepsy but whose EEGs were normal during the video recording.

Data categorization

EEG electrodes were placed according to the 10–20 system. The location of MRI lesions was identified according to previously published data regarding MRI-EEG electrode matching (Okamoto et al., 2004; Koessler et al., 2009).

The localization of EEG findings and MRI lesions were categorized as anterior, temporal, posterior, and other (e.g., diffuse, multifocal, hemispheric, and diencephalon) (Fig. 2A). Focal slowing was similarly categorized as anterior, temporal, and posterior. The location of the MRI lesion and EEG focal slowing was determined according to the predominant focus and its field extension. For example, if there was T5 predominant focal slowing with extension to T3, it was considered to be temporal, whereas T5 predominant focal slowing with extension to O1 was considered to be posterior.

Figure 2.

Diagram of categorization for location and depth. Location of electrodes are placed as 10–20 system and grouped with three different regions (A) of anterior, temporal, and posterior region. Depth of lesion were categorized into nine groups (B): Cortical gray matter only (a); White matter only (b); Both of CGM and WM (c); Junction of CGM and WM (d); Multilayer (e); Hippocampi (f); Diencephalon (g). Subependymal and extracortical regions were not marked in this diagram. (BH Noh had modified sources downloaded through a website “Google 3D Warehouse” to draw the Fig. 1A using “Google Sketchup” program. The license to modify and publish the source was allowed by Google policy ( Figure 1B was drawn by BH Noh.

Side was assigned as follows: left only and left > right were considered left-sided; L = R was considered bilateral; right > left and right only were considered right-sided.

The depth of the lesions was designated as: cortical gray matter only (CGM); white matter only (WM); both of CGM and WM (GWM); the junction of CGM and WM (GWJ); hippocampi; diencephalon; subependymal; extracortical; and multilayer (Fig. 2B).


We examined the frequency of MRI-demonstrated structural abnormalities in patients with focal slowing on the EEG and the concordance between the EEG and MRI features. Chi-square and t-tests were used as appropriate to the data (SAS 9.2, SAS Institute, Cary, NC, U.S.A.).


Characteristics of the subjects

There were 253 subjects in our sample: 34 in group 1, 84 in group 2, 102 in group 3, and 33 in group 4. The average age at enrollment was similar among the four groups (Table 1).

Table 1. Characteristics of the study sample
 Overall N = 253Group 1 n = 34Group 2 n = 84Group 3 n = 102Group 4 n = 33p Value
Mean age at investigation7 years 1 month6 years 0 month6 years 7 months7 years 4 months8 years 4 months0.24
Mean interval between EEG and MRI8 months5 months11 months7 months8 months0.16
Abnormal MRI151 (60%)20 (59%)47 (56%)75 (74%)9 (27%)<0.0001

The interval between the EEG and MRI was less than a month in 81 patients (32%); 1–6 months in 92 patients (36%), and >6 months in 80 patients (32%). Among the 80 patients with a >6 month interval, 30 patients had normal MRIs, 16 had cortical malformation, and one had a hypothalamic hamartoma. The EEG-MRI intervals were similar among the groups.

Abnormal MRI findings

The presence of MRI abnormalities varied substantially among the four groups: underlying lesions were found in 20 (59%) of children in group 1, 47 (56%) in group 2, 75 (74%) in group 3, and 9 (27%) in group 4 (p < 0.0001). The proportion of children with an abnormal MRI was different in group 1 versus 4 (p = 0.0092); group 3 versus 4 (p < 0.0001); however, groups 1 and 3 did not differ significantly (p = 0.11).

MRI studies were performed with either 1.5 or 3 T magnets. Strength of magnets used was not significantly different among the four groups (p = 0.0009). The proportion of subjects with abnormal scan results was marginally higher in those who had 3 T scans: 23 (74%) of 31 of those who had a 3 T scan versus 128 (58%) of 222 who had a 1.5 T scan (p = 0.10).

Types of structural lesions

The most common type of structural lesion was a cortical malformation (CM) in 30% of cases and did not vary significantly among groups 1 (30%), 2 (21%), 3 (37%), and 4 (22%). Atrophy (26%), nonspecific gliosis (13%), stroke (12%), and mesial temporal sclerosis/hippocampal sclerosis (MTS/HS) (7%) were the next most common findings. Tumors were the least common lesions.

The depth of lesions

We found that focal slowing is not always restricted to lesions involving white matter (Table 2). Although one fifth of our subjects had lesions confined to white matter 17 (18%) of 95, four fifths (or 75 [79%] of 95) had lesions not only in white matter but also in other layers.

Table 2. Types and location of MRI lesions overall and by EEG-defined groups
 OverallGroup 1Group 2Group 3Group 4
  1. CGM, cortical gray matter only; WM, white matter only; GWM, both of CGM and WM; GWJ, junction of CGM and WM.

  2. a

    Cases include two tuberous sclerosis, two perinatal insult, one hypothalamic hamartoma, one metabolic origin, and one acute edema.

Types of MRI lesion1512047759
CM46 (30%)6 (30%)10 (21%)28 (37%)2 (22%)
Atrophy39 (26%)2 (10%)18 (38%)17 (23%)2 (22%)
Stroke18 (12%)4 (20%)5 (11%)7 (9%)2 (22%)
Nonspecific20 (13%)2 (10%)9 (19%)7 (9%)2 (22%)
MTS/HS10 (7%)3 (15%)1 (2%)6 (8%)0
Tumor6 (4%)3 (15%)02 (3%)1 (11%)
Othera12 (8%)04 (9%)8 (11%)0
Depth of MRI lesion     
CGM7 (5%)03 (6%)4 (5%)0
GWM34 (23%)5 (25%)3 (6%)23 (31%)3 (33%)
GWJ1 (1%)1 (5%) 00
WM30 (20%)5 (25%)11 (23%)12 (16%)2 (22%)
Diencephalon4 (3%)0 3 (4%)1 (11%)
Subependymal1 (1%)0 1 (1%)0
Multilayer61 (40%)2 (10%)29 (62%)27 (36%)3 (33%)
Extracortical3 (2%)3 (15%) 00
Hippocampal10 (7%)4 (20%)1 (2%)5 (7%)0

Concordance between EEG and MRI abnormalities

The side of the focal slowing and the MRI lesion were concordant in 70% of cases in group 1 and 61% of cases in group 3. Focal IEDs and MRI abnormalities were on the same side in 61% of cases.

The location of focal slowing and MRI abnormalities was concordant in 38 (40%) of 95 (Table 3); focal slowing and focal IEDs were in the same location in 75 (74%) of 102; focal IEDs and MRI abnormalities matched in 33 of (30%) 111 cases (Table 4). When there was a localized underlying lesion, the EEG focus was concordant with the lesion in 38 (67%) of patients with focal slowing and 33 (55%) with focal IEDs. Concordance with MRI location was somewhat less for posterior EEG focus (50%) than foci in the anterior (71%) or temporal (68%), but this did not attain statistical significance (p = 0.33). For example, patient 3 had left hippocampal lesion with focal slowing concordant in the same region (Fig. 3A,B). On the other hand, patient 71 had an anterior temporal lesion with focal slowing presenting posteriorly (Fig. 3C,D).

Table 3. Location of focal slowing and MRI abnormality
 Focal slowing location
 Total (%)AnteriorTemporalPosterior
  1. a

    Cases include diencephalon, hemispheric, diffuse lesion.

MRI location136474841
Normal41 (30%)15 (32%)17 (29%)9 (22%)
Anterior21 (15%)12 (26%)7 (15%)2 (5%)
Temporal29 (21%)5 (11%)17 (35%)7 (17%)
Posterior10 (7%)01 (2%)9 (22%)
Othera31 (23%)15 (32%)6 (13%)10 (24%)
Table 4. Location of focal IED and MRI abnormality
 Focal IED location
 Total (%)AnteriorTemporalPosterior
  1. a

    Cases include diencephalon, hemispheric, diffuse lesion.

MRI location172734158
Normal61 (35%)31 (42%)13 (32%)17 (29%)
Anterior18 (10%)11 (15%)4 (10%)3 (5%)
Temporal27 (16%)10 (14%)13 (32%)4 (7%)
Posterior12 (7%)1 (1%)2 (5%)9 (16%)
Othera54 (31%)20 (27%)9 (22%)25 (43%)
Figure 3.

Cases of MRI and EEG findings. MRI of patient 3 (A) showed mesial temporal sclerosis of the left hippocampus indicated by red arrow. There is dental brace artifact seen on right hippocampal region in axial cuts. MRI was performed at the age of 9 years 11 months with 1.5 T. Video-EEG (B) was performed at the age of 10 years 11 months. There were intermittent runs of polymorphic delta activity with left midtemporal predominance (longitudinal bipolar montage and sensitivity of 15 uV/mm). The patient had undergone the left temporal lobectomy at the age of 13 years. The pathology from anterior temporal tip showed focal cortical dysplasia type I. MRI of patient 71 (C) showed cortical malformation of the right anterior temporal lobe. There is blurring of the gray–white matter junction and slight cortical thickening. MRI was performed at the age of 9 years 9 months with 1.5 T. Video-EEG (D) was performed at the age of 9 years 4 months with a 3-month interval prior to the MRI. There were intermittent runs of polymorphic delta activity in the right posterior region (longitudinal bipolar montage and sensitivity of 20 uV/mm). It is remarkable that focal slowing presents posterior to the actual structural lesion.


The main finding of this study was that 70% of children with nonsyndromic epilepsy and focal slowing and 56% of children with focal IEDs on their EEG had a brain MRI abnormality. About one third of these structural abnormalities were diffuse; however, the other two thirds were associated with a clear focal structural abnormality. There was a very high concordance with the laterality of the slowing and the presence of lesion on MR imaging, but not location.

Overall, 20% of children in this tertiary referral center sample with epilepsy and no focal slowing had a focal structural lesion on brain MR. This percentage is undoubtedly higher than would be seen in the general population due to the referral biases of children with more complicated epilepsy to our center. However, the difference between the groups with or without focal slowing is still significant.

Consistent with current guidelines on pediatric neuroimaging in patients with epilepsy, we excluded those who had syndromes in which imaging is unlikely to be of benefit including CAE, BECTS, JAE, and JME (Hirtz et al., 2000; Gaillard et al., 2009). Had we included these patients, the proportions of normal scans would have been higher.

MRI findings in different pathologic settings may change over time, including the number of foci or the size of the lesions. In 68% of patients, the interval was <6 months. In the patients who had a longer interval, 59% had MRI findings, which would not expected to change over time. Therefore, only 13% of all patients may have had MRI findings altered by this delay—not a number to change the overall trends.

Higher-strength magnets provide better resolution and sensitivity for detecting focal structural lesions, although it is difficult to quantify this benefit precisely as studies show variable results (Knake et al., 2005; Phal et al., 2008; Strandberg et al., 2008). Because our study combined the use of 1.5 and 3 T imaging, it is plausible that the proportion of subjects with abnormal results would have been greater had all scans been performed at the higher signal strength.

Cortical malformations were seven times more common than tumors in children with MRI lesions. These findings are similar to other modern imaging studies showing a higher percentage of cortical malformations in younger children and a higher proportion of atrophy and tumors in older individuals (King et al., 1998; Berg et al., 2009; Gaillard et al., 2009).

In contrast to the classic teaching that focal delta is seen with lesions of the cortical white matter (Schaul et al., 1986), we found that focal slowing in children with epilepsy was often associated with lesions that covered multiple layers including the cortical gray mantle. In other words, we did not find slowing was restricted to lesions involving the cortical white matter only.

Our findings are completely consistent with the latest recommendations set forth in guidelines for imaging children with epilepsy (Hirtz et al., 2000; Gaillard et al., 2009): “Imaging is recommended when localization-related epilepsy is known or suspected, when the epilepsy classification is in doubt, or when an epilepsy syndrome with remote symptomatic cause is suspected.” Furthermore, the results of this study show that focal slowing, as well as focal interictal epileptiform activity, are important and useful EEG indicators of a brain structural abnormality in children with nonsyndromic epilepsy.


Dr. Noh reports no disclosures. Dr. Berg receives support from grant NINDS-R37-NS31146. She has received travel funding and honoraria from Eisai, the British Pediatric Neurological Association, and the Epilepsy Research Center (Melbourne); travel funding from UCB the American Epilepsy Society and the International League Against Epilepsy; BIAL, awards from the American Epilepsy Society and British Pediatric Neurological Association; and consulting fees from Dow Agro Science. She serves on the Editorial Boards of Epileptic Disorders, Epilepsy & Behavior, and Neurology. She is past Chair of the International League Against Epilepsy's (ILAE's) Commission on Classification and Terminology, Current Chair of the ILAE's Task Force on Classification-Diagnostic Manual, member of the American Epilepsy Society Psychiatric Task Force, Steward for the National Institute of Neurological Disorders and Stroke (NINDS) Benchmarks in Epilepsy Research. Dr. Nordli, Jr, has received funding as a co-investigator in NIH/NINDS 1-RO1-NS43209 and he is an associate editor for UpToDate. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.