• fMRI;
  • Language;
  • Children;
  • Epilepsy;
  • Cerebral lesions


  1. Top of page
  2. Abstract
  6. Acknowledgments

Summary: Purpose: Lateralization of language function is crucial to the planning of surgery in children with frontal or temporal lobe lesions. We examined the utility of functional magnetic resonance imaging (fMRI) as a determinant of lateralization of expressive language in children with cerebral lesions.

Methods: fMRI language lateralization was attempted in 35 children (29 with epilepsy) aged 8–18 years with frontal or temporal lobe lesions (28 left hemisphere, five right hemisphere, two bilateral). Axial and coronal fMRI scans through the frontal and temporal lobes were acquired at 1.5 Tesla by using a block-design, covert word-generation paradigm. Activation maps were lateralized by blinded visual inspection and quantitative asymmetry indices (hemispheric and inferior frontal regions of interest, at p < 0.001 uncorrected and p < 0.05 Bonferroni corrected).

Results: Thirty children showed significant activation in the inferior frontal gyrus. Lateralization by visual inspection was left in 21, right in six, and bilateral in three, and concordant with hemispheric and inferior frontal quantitative lateralization in 93% of cases. Developmental tumors and dysplasias involving the inferior left frontal lobe had activation overlying or abutting the lesion in five of six cases. fMRI language lateralization was corroborated in six children by frontal cortex stimulation or intracarotid amytal testing and indirectly supported by aphasiology in a further six cases. In two children, fMRI language lateralization was bilateral, and corroborative methods of language lateralization were left. Neither lesion lateralization, patient handedness, nor developmental versus acquired nature of the lesion was associated with language lateralization. Involvement of the left inferior or middle frontal gyri increased the likelihood of atypical language lateralization.

Conclusions: fMRI lateralizes language in children with cerebral lesions, although caution is needed in interpretation of individual results.

Standard methods of language lateralization, including intracarotid amytal (ICA) testing and cortical stimulation, are often difficult to perform in children, and their invasive nature limits repeated assessments and study of normal children. In contrast, functional magnetic resonance imaging (fMRI) is relatively noninvasive and, in spite of several challenges in its application to children, has been used to lateralize language in children by several groups (1–13). Word-generation paradigms used to study expressive language in normal children produce consistent left-lateralized activation, primarily in the inferior frontal gyrus and middle frontal gyrus (2,3,5,6,10–12).

Lateralization of language function is crucial to the planning of epilepsy, tumor, and vascular surgery and to the understanding of patients' baseline and postoperative language deficits. Cortical organization of language function is more likely to be aberrant in children with epilepsy and brain lesions (14,15). The application of fMRI to pediatric patients with neurologic conditions has not been extensively studied. Contributing reasons include the practical difficulties inherent in the use of fMRI in children, neurobehavioral disabilities prevalent in pediatric neurology patients, and organizational hurdles encountered by clinical departments associated with implementation of protocols developed largely for research. Reports of fMRI of language in pediatric patients are presently limited to case studies (16–20) and series of seven to 14 patients (1,4,9,13,21,22). These studies demonstrate that fMRI is feasible in children and valid as a measure of language lateralization when assessed against the gold standards of cortical stimulation or ICA testing (1,4,9,13,18,20), irrespective of the specific language paradigm, imaging techniques, or quantitative assessments used. One of the larger series considered issues of cerebral plasticity and reported that lateralization cannot be inferred from the proximity of lesions to language cortex, highlighting the complexity inherent in predicting language lateralization (4). Case studies are helpful in understanding concepts of plasticity and critical periods for language acquisition (17), as well as questioning the meaning of fMRI activation in clinical subjects (16–18).

Given the relatively limited clinical experience with fMRI in children with cerebral lesions, we aimed to examine language lateralization by using fMRI in a large sample of children with cerebral lesions and to analyze the results from a clinical perspective. Secondary aims were to evaluate the results in relation to corroborative evidence from other modalities, to compare visual and quantitative methods of lateralization, and to examine the need for Bonferroni correction in data analysis.


  1. Top of page
  2. Abstract
  6. Acknowledgments


Thirty-five English-speaking children with lesions on structural MRI involving the frontal or temporal lobes, who had been cooperative with anatomic MRI without need for general anesthesia, were recruited from the Royal Children's Hospital, Melbourne, Australia, mostly from the Children's Epilepsy Program. The project had ethics approval from the Royal Children's Hospital Ethics in Human Research Committee.

Clinical information was obtained from medical records and epilepsy case conference notes. Anatomic MRI scans were reviewed for lesion characteristics and location. WISC-III (23) Full Scale IQ data were obtained from the most recent neuropsychological assessment. The Peabody Picture Vocabulary Test–Revised (PPVRT – R) (24) was administered as a measure of receptive vocabulary. Handedness was assessed by using the Edinburgh Handedness Inventory (25). Before fMRI scanning, subjects performed the standardized version of the Controlled Oral Word Association Test (COWAT) (26) with verbal responses. Raw scores of the COWAT are reported here, as these provide an out-of-scanner estimate of performance on the fMRI activation paradigm (see later). Normative data for children's COWAT performance are available (27).

Activation paradigm

The activation task was an orthographically cued lexical retrieval task based on the COWAT. A block design with six alternating task–rest cycles was used, beginning with a rest phase. During the task, subjects were required to think of words beginning with a given letter, presented aurally through headphones, while lying motionless in the MRI scanner with their eyes closed. During rest phases, they were instructed to keep their eyes closed and not to think about anything. Cycles were 36 s in duration with two letters presented per task cycle.

Subject preparation

Subjects were familiar with the MRI facility, having previously undergone anatomic MRI for clinical purposes. Before fMRI, subjects were refamiliarized with the MRI environment. All children were given training in the activation paradigm outside the magnet immediately before scanning. Each demonstrated an understanding of the task requirements as a condition of inclusion in the study. During scanning, the head was immobilized within the head coil by using foam rubber pads and individually molded thermoplastic masks held in place with Velcro straps.

Image acquisition

Functional, anatomic, and angiographic MRI was performed at 1.5 Tesla on a GE Signa LX Echospeed system (software version 5.6; General Electric Medical Systems, Milwaukee, WI, U.S.A.) by using a standard birdcage quadrature head-coil. Single-shot gradient-recalled echo-planar imaging was used (TR, 5 s; TE, 60 ms; flip, 30 degrees; interscan delay, 1 s; 128 × 128 matrix; bandwidth, ±100 kHz, 1 NEX; 250 × 180-mm field of view, reconstructed to an in-plane pixel size, 1.95 × 1.95 mm). An automated shim was performed to maximize field homogeneity before each experimental session. Slice-number limitations at the commencement of this research precluded full brain coverage. Data acquisition was therefore focused on the frontal lobes, given the expressive nature of the activation task, and previous findings reported by our group (5,28) and others (4,6,11,13,29) using verbal-fluency paradigms. Coverage was typical of that reported by others examining expressive language function in children (11,13). Six coronal and six axial slices (4-mm slice thickness; 3- to 6-mm gap) were acquired (four patients had four slice studies, and axial slices were not acquired in two patients).


iBrain (30) was used to analyze the reconstructed MR images. The first image of each task and rest epoch was excluded from analysis, allowing the hemodynamic response to reach a peak level. Scans in each image set were masked by using an iterative intensity-based technique to exclude most out-of-brain pixels. Global intensity normalization was performed to achieve a constant within-brain mean. An estimation of motion between slices was based on a measure of within-brain center of mass. Images in each series that deviated from the median by more than one third of a pixel were excluded from further analysis. A gaussian smoothing filter (full width at half maximum, 2 pixels = 3.9 mm) was applied to each image to reduce the effects of residual subject motion, to improve the signal-to-noise ratio, and to ensure an approximately gaussian noise profile. A statistical parametric map for each slice in each child was obtained by performing a one-tailed Student's t test at each pixel location comparing signal intensity between task and rest states.

Visual lateralization

Lateralization by visual inspection was conducted by a neurologist and a neuropsychologist blind to each child's identity. Activation maps (thresholded at p < 0.005 but not corrected for multiple comparisons) displayed in pseudo-color and superimposed on the anatomy of the raw echo-planar images were reviewed at one sitting. Only regions with clusters of pixels displayed in “warm” colors (corresponding to a one-tailed significance of p < 0.0001, uncorrected) and surrounded by activation significant at p < 0.005 were considered. For the purposes of this study, data sets in which activation was not present in the inferior frontal gyrus were considered unsuccessful. Visual-inspection lateralization was based on the degree and spatial extent of activation primarily in the region of the inferior frontal gyrus and secondarily in the middle and superior frontal gyral regions. Coronal slices were favored for fMRI lateralization, as the gyral and sulcal anatomy in the frontal and temporal lobes was better represented than that in the axial slices. Lesions were apparent on echoplanar images on 11 studies exposing identity in some cases.

Quantitative lateralization

Rectangular “hemispheric” regions of interest were drawn on the raw echoplanar coronal images without the statistical data superimposed. These were sized to individual anatomy on all coronal slices in each data set. The medial boundary was drawn parallel to the midline at the level of the grey–white matter junction at the depth of the cingulate sulcus, thus excluding medial frontal areas where accurate lateralization can be difficult (Fig. 1). An inferior frontal region of interest was drawn from the sylvian fissure at the lateral surface, extending medially to middle of the insular cortex and then superiorly to the apex of the lateral ventricle and through the midpoint between the inferior and middle frontal gyrus grey matter to the lateral surface (Fig. 1). The process of drawing this inferior frontal region of interest on the echoplanar images was more cumbersome, particularly in children, where lesions distorted the anatomy and consequently was prone to being less reliable than the “hemispheric” region of interest. Activated pixels within each region of interest were counted by using iBrain (30). Asymmetry indices (AIs) were calculated using the formula:

  • image

where inline image is the sum of activated left-side pixels, and inline image is the sum of activated right-side pixels. A positive AI corresponded to a left-predominant activation, and a negative AI indicated right-predominant activation. Asymmetry indices between −0.25 and +0.25 were considered to reflect bilateral language activation (29). AIs were calculated in this manner for statistical data thresholded at p < 0.001 uncorrected and at the more conservative level of p < 0.05, Bonferroni corrected for multiple comparisons.


Figure 1. Hemispheric region of interest (thick rectangle shown on left) and inferior frontal region of interest (thin polygon shown on right), drawn on the average of the raw echoplanar imaging without activation data overlaid.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgments

The demographic and lesion characteristics of the 35 recruited children (19 boys) age 8–18 years (mean age, 12 years) are summarized in Table 1. Although a specific language concern was not an inclusion criterion, the source of patients resulted in a bias toward patients with seizures, left-hemisphere lesions, and right-hemisphere lesions associated with left-handedness. Twenty-nine (83%) children had epilepsy, and two had a single seizure. Twenty-eight (80%) children had lesions confined to the left hemisphere, five had right-hemisphere lesions, and two children had lesions with considerably greater involvement on the left. Twenty children had developmental tumors or dysplasias, and 15 had acquired lesions with a variety of antecedents including febrile convulsions, stroke, and encephalitis. Seven children underwent frontal lobe cortical stimulation or ICA testing, and 19 children underwent epilepsy surgery. The children varied considerably in general intelligence (mean IQ, 86; range, 59–117), receptive vocabulary (mean PPVT-R standard score, 83; range, 46–124), and ability to perform the standardized version of the COWAT (mean number of words, 20; range, 5–40).

Table 1. Patient demographic and lesion data
Patient dataLesion data
PtAge SexHandedEpilepsyFSIQPPVTCOWATMRI findings (pathology of resected tissue)Etiology (antecedent) [age acquired in yr]
  1. Pt, patient; FSIQ, Full Scale Intelligence Quotient (standard score); PPVT, Peabody Picture Vocabulary Test (standard score); COWAT, Controlled Oral Word Association Test (raw score); M, male; F, female; FCD, focal cortical dysplasia; HS, hippocampal sclerosis; IFG, inferior frontal gyrus; MFG, middle frontal gyrus; STG, superior temporal gyrus; MCF, middle cranial fossa; ITG, inferior temporal gyrus; MCA, middle cerebral artery; HSVE, herpes simplex virus encephalitis.

Successful fMRI studies
  113, MRY868328R gyrus rectus FCD (dysplasia)Developmental
  210, MLY849218R temporoparietooccipital atrophy/gliosis & HSAcquired (chemotherapy) [3.5]
  314, MLY786719L HSAcquired (nil antecedent) [?<4]
  414, MR1 seizure90104 21L IFG + MFG calcified tumorDevelopmental
  511, FLY917912R HS (HS)Acquired (febrile seizures) [1]
  614, FRY108 9425L temporal FCD 2° to tuberous sclerosis (dysplasia)Developmental
  7 8, FLY8152 6L temporal encephalomalacia/gliosis & HS (no pathology)Acquired (HSVE) [0.5]
  818, MLY94n/a26L hemisphere atrophy & HS (HS)Acquired (nil antecedent)
  910, FRY92100 22L STG tumor (dysembryoplastic neuroepithelial tumor)Developmental
 10 8, FRY9088 5L IFG tumor (ganglioglioma, later malignant transformation)Developmental
 1114, MRN7178 5L temporal tumor with hemispheric edema (glioblastoma)Acquired (tumor growth) [13]
 1214, FRY998040L parahippocampal FCD (dysplasia) & hypothalamic hamartomaDevelopmental
 1312, FL1 seizure859920L IFG porencephalic cyst & small hemisphereAcquired (nil antecedent)
 1412, MLY889021L MFG FCD (dysplasia)Developmental
 1510, FRY594725L SFG FCDDevelopmental
 1610, MLY695810L STG tumor ?dysembryoplastic neuroepithelial tumorDevelopmental
 17 9, MLN113 105 26L MCA territory atrophy/gliosisAcquired (stroke) [9]
 18 9, MRY879214L hippocampal & parahippocampal FCD (dysplasia)Developmental
 1913, MRY979821L IFG tumor (dysembryoplastic neuroepithelial tumor)Developmental
 2018, FRY947736L occipitotemporal atrophy/gliosis & HS (HS, atrophy)Acquired (nil antecedent)
 21 9, MRY847815L hemisphere atrophy/gliosis & HS (encephalomalacia)Acquired (HSVE) [4]
 2213, MRY769018L SFG + MFG FCD (dysplasia)Developmental
 2313, FLY117 12430L inferior temporal tumor (ganglioglioma)Developmental
 2415, MLY838019L perisylvian (insula, STG, IFG, MFG) FCD (dysplasia)Developmental
 2516, MLY715418L > R HS & white-matter abnormalitiesAcquired (head injury) [4]
 2614, FRY849817R posterior temporal angiomas (cavernoma)Developmental
 27 9, MR/MixedN87113 13R MCF arachnoid cyst and R temporal hypoplasiaDevelopmental
 28 8, MRY107 9025L amygdala & STG hyperintensity (chronic encephalitis)Acquired (Rasmussen) [8]
 2914, FRN106 9726L inferior frontal atrophy/gliosisAcquired (stroke) [13]
 3014, FRY644628L mesial occipitotemporal FCD (dysplasia)Developmental
Unsuccessful fMRI studies
 3115, FRY777123L HS (HS)Acquired (febrile seizures) [1.4]
 3211, MLY628012L HS (HS)Acquired (meningitis) [0.5]
 3318, FRY836519L ITG tumor (ganglioglioma)Developmental
 3410, MRY858917L > R multiple FCD 2° to tuberous sclerosisDevelopmental
 3514, FLY836211L inferior temporal granuloma (meningioangiomatosis)Developmental

fMRI language lateralization

The results of fMRI are shown in Table 2. Significant activation (p < 0.0001, uncorrected) was seen in all but one patient (patient 33) on coronal examination. Three studies with activation were also considered unsuccessful, as no activation was noted in the inferior frontal region (patients 31, 32, 34). Thus 30 (86%) of the 35 studies were considered successful in terms of activating inferior frontal regions. Activation remained significant at the p < 0.05 Bonferroni corrected level in all but one (patient 26) of the 30 successful studies.

Table 2. Functional MRI language lateralization and corroborative data
PtVisual inspection Side Gyri activated (p < 0.0001, warm colors)Asymmetry indicesBest corroborative evidence of language lateralization
 LIFGRIFGLMFGRMFGLSFGRSFG(p < 0.001) hemisphere inf. frontalp < 0.05 Bonferroni hemisphere inf. frontal
  1. Pt, patient; L/RIFG, left/right inferior frontal gyrus; L/RMFG, left/right middle frontal gyrus; L/RSFG, left/right superior frontal gyrus; hemisph, hemispheric; inf. frontal, inferior frontal; L, left; R, right; B, bilateral; ICA, intracarotid amobarbital testing; MCA, middle cerebral artery.

Successful fMRI studies
 1L+ + 0.820.911   1    
 2L+++ + 0.520.370.610.5 
 3L+ +++ 0.430.610.710.96 
 4L+ + + 0.610.351   1    
 5L+ + ++0.880.850.750.75 
 6L+ ++++0.790.900.640.79L frontal & temporal stimulation: aphasic
 7L+++ ++0.530.540.650.88L frontal & temporal stimulation: aphasic
 8B++++++−0.19 −0.17 −0.15 −0.19 L frontal & temporal stimulation: not aphasic
 9L+ ++0.870.940.971   Ictal aphasia
 10B++++++   Ictal aphasia
 11L+ ++++0.390.600.570.63Aphasia with tumor enlargement
 12L++++++0.640.680.9 0.87L frontal & temporal stimulation: aphasic
 13R + +−0.50 −1    −1    −1     
 14L++++++0.280.310.490.47L frontal stimulation: aphasic
 15L++++++0.350.440.310.68Ictal aphasia
 16R++++ +−0.64 −0.59 −0.62 −0.27  
 17R + + −0.33 −0.38 −1    −1    Not aphasic with acute L MCA stroke
 18L+ + 0.861   1   1    
 19B+++ ++0   0.310.080.72L frontal stimulation: aphasic
 20L+ + + 0.390.8 1   1   Ictal aphasia
 21L+ + ++0.920.930.980.98 
 22L+ + 0.780.871   1    
 23R + + +−0.94 −0.93 −1    −1     
 24R + +++−0.88 −1    −0.78 −1    L ICA testing: not aphasic
 25L+ ++++0.720.800.820.80 
 26L+ + 0.400.71n/a n/a  
 27L+ + ++0.570.471   1    
 28L+++ + 0.640.400.8 0.77Ictal aphasia
 29R++++++−0.22 −0.02 −0.67 −0.06 Aphasic with L MCA stroke
 30L+ 0.640.641   1    
Unsuccessful fMRI studies
 31 + Ictal aphasia
 32 + 
 33 + 

Visual inspection revealed that inferior frontal activation was unilateral in 17 children and prominently asymmetrical in 10 children, leading to lateralized language in 27 children (21 left, six right). Three studies showed an equivalent degree of inferior frontal activation and were judged to have bilateral language. Activation was also present in the region of the middle frontal gyrus (three children), superior frontal gyrus (four children), or both (22 children). In the 27 lateralized studies, middle frontal and superior frontal activation was often bilateral, but where unilateral, activation was ipsilateral to the greater inferior frontal activation. Less-frequent regions of activation were the anterior cingulate (nine children) and insular cortex (six children). Cerebellar activation (seven children) and inferior or lateral temporal activation (eight children) also were seen, but these regions were not adequately imaged with the limited number of slices acquired. Figure 2 shows examples of typical lateralized and bilateral activation.


Figure 2. Typical examples of left-lateralized (patient 14), bilateral (patient 10), and right-lateralized (patient 23) frontal language activation. Activation thresholded at p < 0.005 (uncorrected probability of null hypothesis) and overlaid onto the average of the raw echoplanar imaging. Activity at p < 0.0001 is shown as bright white, whereas 0.0001 < p < 0.005 is shown in a dark shade of grey.

Download figure to PowerPoint

Quantitative lateralization at each statistical threshold and for both hemispheric and inferior frontal regions of interest was concordant with lateralization from visual inspection in all but two children (patients 19 and 29; 93% concordant). Visual-inspection lateralization and hemispheric region-of-interest quantitative lateralization at both thresholds were bilateral for patient 19 but left lateralized based on the inferior frontal region of interest. Inferior frontal region of interest lateralization and hemispheric lateralization at p < 0.001 were bilateral for patient 29, whereas visual inspection and hemispheric lateralization at p < 0.05 Bonferroni corrected were right lateralized. No right–left discordance was seen.

Corroborative evidence of fMRI lateralization

Seven patients underwent language testing with ICA injection or electrical stimulation of the frontal cortex with results concordant with fMRI in six patients (patients 6, 7, 8, 12, 14, 24). Aphasia during left frontal cortical stimulation was potentially discordant with bilateral language activation on visual inspection and the hemispheric asymmetry index in patient 19 but concordant with left lateralization on the inferior frontal AI. Language deficits during seizures or in the course of disease were apparent in a further seven children and were concordant with fMRI lateralization in five (patients 9, 11, 15, 20, 28) and discordant in patients 10 and 29. In patient 17, the absence of a language deficit after an acute left middle cerebral artery (MCA) stroke was concordant with fMRI language lateralization to the right hemisphere. None of the 19 operated-on patients had a language deficit.

An apparent discordance between fMRI lateralization and that from these other modalities was therefore present in three children (patients 10, 19, and 29). Timing of fMRI assessment in relation to an acquired lesion offered a plausible explanation in patient 29, who was aphasic after a left MCA territory stroke, suggesting preexisting left hemisphere dominance. fMRI showing bilateral (inferior frontal AI) or right hemisphere language lateralization (visual inspection and hemispheric AI) was conducted 6 months after the stroke when language function had recovered and may reflect reorganization in the right hemisphere. No plausible explanation was apparent for two patients who showed bilateral activation on visual inspection and hemispheric AI but had corroborative evidence of left hemisphere language dominance (cortical stimulation in patient 19, and ictal aphasia in patient 10). These children were both right-handed and had epileptogenic developmental tumors in the left inferior frontal gyrus. Repeated fMRI studies 2 years after the first examination (not analyzed in this series) in patient 19 showed left-hemisphere dominance consistent with the cortical-stimulation data, which were acquired at that time and left lateralization by inferior frontal asymmetry index in the present study. In patient 10, repeated fMRI studies (not analyzed in this series) showed right lateralized language, despite a deterioration of language function as the tumor increased in size and seizure control worsened (16).

Thus corroborative evidence from standard methods of language lateralization or aphasiology available in 15 patients was clearly consistent with fMRI language lateralization in 12 patients. A plausible explanation was apparent for the discrepancy in a further case. Discordance between fMRI and other methods of language lateralization could not be resolved in two patients (10 and 19).

fMRI language activation in relation to lesion characteristics

Classification of language lateralization as either typical (left) or atypical (right or bilateral) was based on visual inspection for the purpose of statistical calculations. Atypical lateralization was not associated with lesion lateralization, nature of lesion (developmental versus acquired), or timing of acquired lesions (Table 3). Atypical language lateralization was associated with lesion involvement of the left inferior or left middle frontal gyrus.

Table 3. Relation between lesion and patient variables and lateralization of language by fMRI (visual inspection)
 No.“Typical” language (left) (left) (n = 21)“Atypical” language (right or bilateral) (n = 9)p Value
  1. IFG, inferior frontal gyrus; MFG, middle frontal gyrus; FSIQ, Full Scale Intelligence Quotient; PPVT, Peabody Picture Vocabulary Test; COWAT, Controlled Oral Word Association Test.

  2. aFisher's exact.

  3. bStudent's t test.

Left-sided lesion2516 90.3a
Left IFG/MFG lesion specifically 9 3 60.01a
Developmental lesion1712 51a   
Late acquired (>6 yr) lesion 4 2 20.5a
Mean FSIQ (standard score)3085950.07b
Mean COWAT (raw score)3020201b   
Mean PPVT (standard score)2981940.1b
Mean age of patient (yr)3012120.8b
Right-handed1815 30.1a

Nine children had lesions in the left inferior frontal or middle frontal gyrus. fMRI language lateralization was left for three (patients 4, 14, 22), right for four (patients 13, 17, 24, 29), and bilateral for two (patients 10, 19). Six children had developmental lesions, three showed left lateralized activation (patients 4, 14, 22), and two showed bilateral activation (patients 10, 19) with activation overlying or abutting the lesion for all these cases (Figs. 3 and 4). One left-handed patient (patient 24) with a large dysplastic lesion involving the entire left perisylvian region, including frontal and temporal prototypic language regions, was right language lateralized. All three children with acquired lesions showed right lateralized language activation on fMRI (patients 13, 17, 29). These patients differed in terms of the timing of their lesions. Patient 13 had a presumed prenatal incident and was left-handed. Patients 17 and 29 had late acquired lesions, both having left MCA territory strokes. Patient 17 was left-handed and had no aphasia as a result of the stroke at age 9 years, whereas patient 29, as noted earlier, was right-handed and experienced aphasia after a stroke at age 13 years, with recovery of language function more than 6 months before fMRI investigation (Fig. 5).


Figure 3. Three axial anatomic MRI scans (top) and fMRI scans (bottom) showing language activation in left frontal cortex abutting an area of focal cortical dysplasia (arrows) (patient 14). Activation thresholded at p < 0.005 (uncorrected probability of null hypothesis) and overlaid onto the average of the raw echoplanar imaging. Activity at p < 0.0001 is shown as bright white, whereas 0.0001 < p < 0.005 is shown in a dark shade of grey.

Download figure to PowerPoint


Figure 4. Two coronal fMRI scans showing bilateral language activation (insula, inferior frontal gyrus, superior frontal gyrus) in relation to a left inferior frontal developmental tumor (patient 19). Activation thresholded at p < 0.005 (uncorrected probability of null hypothesis) and overlaid onto the average of the raw echo-planar imaging. Activity at p < 0.0001 is shown as bright white, whereas 0.0001 < p < 0.005 is shown in a dark shade of grey.

Download figure to PowerPoint


Figure 5. Two coronal fMRI scans showing right-lateralized language activation (insula, inferior frontal gyrus, superior frontal gyrus) in relation to a region of left inferior frontal atrophy and gliosis due to a left middle cerebral artery territory stroke (patient 29). Activation thresholded at p < 0.005 (uncorrected probability of null hypothesis) and overlaid onto the average of the raw echo-planar imaging. Activity at p < 0.0001 is shown as bright white, whilst 0.0001 < p < 0.005 is shown in a dark shade of grey.

Download figure to PowerPoint

fMRI language activation in relation to neuropsychological and patient variables

Patients who showed successful language activation on visual inspection did not differ from those whose activation was unsuccessful in terms of their age [t(33) =−1.006; p = 0.322], ability to perform the COWAT [t(33) = 1.001; p = 0.320], intellect [IQ, t(33) = 1.516; p = 0.139], or language abilities [PPVT-R, t(32) = 1.244; p = 0.111].

Correlation between COWAT performance and Bonferroni-corrected hemispheric asymmetry index was not significant [r(27) =−0.037; p = 0.851], suggesting that language lateralization in fMRI scans showing activation was not related to level of proficiency on the activation task. Correlation between age and Bonferroni-corrected hemispheric asymmetry index was also not significant (r=−0.033; p = 0.866).

No significant differences between those whose language was left lateralized on visual inspection and those who showed atypical language lateralization were apparent in terms of IQ, COWAT performance, PPVT-R standard score or age (Table 3). However, there was a trend for IQ and PPVT-R scores to be higher in patients with atypical language lateralization, the lesions in these patients all being on the left. Language lateralization was not associated with handedness (Table 3).


  1. Top of page
  2. Abstract
  6. Acknowledgments

Consistent with previous studies, we have shown that it is possible to demonstrate language activation in children by using fMRI at 1.5 T in a clinical setting. Our study reports fMRI language lateralization in the largest group of pediatric patients to date. Patients were relatively heterogeneous with respect to the nature of their epilepsy, lesions, and neurobehavioral profiles, typical of pediatric clinical practice. The simple but robust paradigm successfully revealed language activation in children as young as 8 years of age, including those with intellectual disability and language impairment. Activation was consistently seen in the inferior frontal gyrus, as well as the middle and superior frontal gyri, consistent with previous reports of language fMRI using verbal fluency paradigms (4–6,11,13,28,29,31). As was the case in these studies, a degree of bilateral activation was present in about 60% of our patients. Nevertheless, lateralization could be easily determined by visual inspection. Visually rated lateralization was highly concordant with the quantitative AIs, regardless of region of interest used or whether the data were statistically corrected for multiple comparisons. In the majority of patients for whom data were available, fMRI lateralization was also concordant with lateralization determined on the basis of cortical stimulation, ICA testing, or aphasiology. Only two patients had seemingly unexplained discordance between visual-inspection lateralization (bilateral) and corroborative data (left), they being two children with developmental lesions in the left inferior frontal gyrus in whom different fMRI lateralization on follow-up imaging suggested possible influence of age, tumor change, and perhaps seizures.

The majority of children in the study were left lateralized for language despite the predominance of left-hemisphere lesions. Consistent with the epilepsy literature (14,32,33), however, the percentage of children who showed atypical language lateralization (20% right lateralized, 10% bilateral) was higher than for the normal population and indeed twice that found in a normative study of children by our group by using a visually presented version of the same orthographic lexical retrieval paradigm (2% right lateralized, 13% bilateral) in which lateralization was not related to subject age, gender, task proficiency, or handedness (5). Our clinical sample also included a higher incidence of left-handedness (41%) than the normal population. However, our sample could not be regarded as an unselected epilepsy sample, and both left-handedness and atypical language are probably overrepresented. Yet neither left-handedness nor the presence of a lesion in the left inferior or middle frontal gyrus, which was statistically associated with a greater likelihood of atypical language lateralization, was absolutely predictive of right-hemisphere language involvement, and clearly such indicators in isolation do not reliably predict language lateralization in patients.

Whereas our data could be considered from a number of perspectives, our interest was primarily in pragmatic clinical application of this technology in children with neurologic problems. Two subgroups of our sample were of particular clinical interest in this regard: first, the patients with lesions involving putative language cortex (inferior left frontal lobe), and second, the patients found to have atypical language lateralization.

An understanding of how lesions in language areas affect cortical organization and lateralization of language function may be important for the planning of surgery in children with tumors or refractory partial epilepsy. Destructive injury to language cortex before the end of a critical period for language development [traditionally regarded as around 5 to 6 years of age (32,34,35)] may result in interhemispheric shift in language dominance, whereas later injury is more likely to cause reorganization of language within the hemisphere (32,34–37). The only patient in our series with an early-acquired lesion involving the inferior left frontal lobe (patient 13) showed right-lateralized language consistent with this premise. Developmental lesions, even those causing intractable epilepsy, do not necessarily result in transfer of language function; rather language often develops in proximity to the lesion in the dominant hemisphere (4,34,38), perhaps due to the functional integrity of some dysplastic cortex (39). Only one patient in the previously reported pediatric fMRI series is reported to have had a developmental lesion (cortical dysplasia) specifically in the inferior left frontal lobe (9), language being left lateralized. In five of the six children in our series with developmental lesions involving the left inferior or middle frontal cortex, fMRI showed left or bilateral language activation overlying or abutting the lesion, consistent with the previous observations (4,34). Patient 24, with an extensive left perisylvian dysplasia and right lateralized language, suggests that developmental lesions may also lead to interhemispheric language reorganization when involvement of the left hemisphere is so widespread that it precludes intrahemispheric reorganization.

Right dominance is rare in dextrals (33) and is seen in only 10% of normal left-handers when assessed with fMRI (29). Innate right-hemisphere language may account for three cases of right-lateralized activation seen in our series. It is the most likely explanation in the left-handed child with a late acquired left-hemisphere stroke that did not disturb language (patient 17). It may also be the explanation in the two left-handed children with small, discrete developmental tumors in the left temporal lobe in whom associated seizures did not produce aphasia (patients 16, 23). As discussed earlier, right-hemisphere language may result from a focal lesion in the left frontal language cortex if the lesion is acquired early in development (patient 13), or from a large developmental lesion in the left hemisphere (patient 24). Patient 29 is also a likely example of interhemispheric transfer of language, given that her language recovered over the 6 months after her left hemisphere stroke at age 13 years, and was associated with right or bilateral language activation. Cases of language reorganization beyond the age of 6 years are reported with progressive cerebral pathology (17,40,41) and challenge the traditional limit of the critical period for language development. A further case of aberrant language lateralization was that of patient 8, who also had an extensive left-hemisphere acquired lesion and whose fMRI was bilateral and consistent with the failure of left frontal cortical stimulation to elicit aphasia (although right dominance cannot be excluded on the basis of the cortical stimulation evidence alone). This may reflect reorganization of language function but again, the underlying mechanism is uncertain. It is interesting to speculate that the trend to higher receptive language and intellectual scores in the nine patients with language in the right hemisphere, contralateral to their lesions, was a function of effective language development in a normal hemisphere, either innately or as a result of “transfer.”

The three cases of potential discordance between language lateralization on fMRI highlight the need for caution in interpretation of fMRI findings in pediatric patients. It is noteworthy that the two patients who remained discordant despite careful clinical consideration were the only children in the sample with epileptogenic lesions in the inferior frontal gyrus. These suggest that the interpretation of fMRI activation in children with epileptogenic lesions in inferior frontal language cortex is not straightforward, and might produce false lateralization or the impression of bilaterality in some patients. Lesions in the inferior frontal gyrus may distort activation and lead to misinterpretation of visual images or incorrect asymmetry indices, particularly when automated quantitation is used. The effect of seizures in the case of epileptogenic lesions and the meaning of activation contralateral to expected language are not well understood (16–18).

An appreciation of the impact of neurologic conditions on the organization of language is essential in planning for neurosurgery in the dominant hemisphere. Determination of language lateralization by clinical prediction (e.g., from handedness, lesion location, or nature of lesion in terms of developmental versus acquired and age at lesion) is problematic and should certainly not be the sole basis of surgical planning. fMRI is a useful, noninvasive tool for lateralization of language function in children with cerebral lesions, particularly as data are increasingly available in normal children (5,6,10–12,42) and indicate that word-generation paradigms produce a reliable pattern of mostly lateralized frontal activation. fMRI language lateralization can be performed at 1.5 T in a clinical setting, is well tolerated by neurologically afflicted children, and generally leads to robust activation in a consistent pattern that can be readily interpreted without the need for quantitative analysis, Bonferroni correction of data, or coregistration with anatomic images. The neurobiologic basis or explanation of specific fMRI findings in individual patients often requires consideration of a number of factors, including handedness, lesion location and type, innate factors, and the presence of seizures. Consequently, individual interpretation is often required. The issues of plasticity and transfer are hypothetical to all practical intents and purposes. At a clinical level, corroboration from other modalities, and the favorable language outcome after surgery in our series, suggest that fMRI of language is likely to replace modalities such as ICA for language lateralization in pediatric patients. Now many groups internationally report fMRI language lateralization in children, and many more use this technology clinically. A need exists for consensus in acceptable paradigms, image-acquisition parameters, and analytic approaches to allow application of reported data to clinical practice. Currently we believe the greatest challenge in pediatric fMRI is the understanding of activation and absence of activation in terms of “plasticity” and its clinical implications.


  1. Top of page
  2. Abstract
  6. Acknowledgments

Acknowledgment:  This research was conducted at the Royal Children's Hospital, Melbourne, with salary support for Ms. Anderson and Dr. Harvey from the Murdoch Children's Research Institute, Melbourne. This work was also supported in part by NHMRC grant number 226100.


  1. Top of page
  2. Abstract
  6. Acknowledgments
  • 1
    Fernandes MA, Smith ML, Logan W, et al. Comparing language lateralization determined by dichotic listening and fMRI activation in frontal and temporal lobes in children with epilepsy. Brain Lang 2006;96:10614.
  • 2
    Szaflarski JP, Holland SK, Schmithorst VJ, et al. fMRI study of language lateralization in children and adults. Hum Brain Mapp 2005 [Epub ahead of print]
  • 3
    Wilke M, Lidzba K, Staudt M, et al. Comprehensive language mapping in children, using functional magnetic resonance imaging: what's missing counts. Neuroreport 2005;16:9159.
  • 4
    Liegeois F, Connelly A, Cross JH, et al. Language reorganization in children with early onset lesions of the left hemisphere: an fMRI study. Brain 2004;127:122936.
  • 5
    Wood A, Harvey AS, Wellard M, et al. Language cortex activation in normal children. Neurology 2004;63:103544.
  • 6
    Gaillard WD, Sachs BC, Whitnah JR, et al. Developmental aspects of language processing: fMRI of verbal fluency in children and adults. Hum Brain Mapp 2003;18:17685.
  • 7
    Sabbah P, Chassoux F, Leveque C, et al. Functional MR imaging in assessment of language dominance in epileptic patients. Neuroimage 2003;18:4607.
  • 8
    Balsamo L, Xu B, Grandin CB, et al. A functional magnetic resonance imaging study of left hemisphere language dominance in children. Arch Neurol 2002;59:116874.
  • 9
    Liegeois F, Connelly A, Salmond CH, et al. A direct test for lateralization of language activation using fMRI: comparison with invasive assessments in children with epilepsy. Neuroimage 2002;17:18617.
  • 10
    Holland SK, Plante E, Weber Byars AW, et al. Normal fMRI brain activation patterns in children performing a verb generation task. Neuroimage 2001;14:83743.
  • 11
    Gaillard WD, Hertz-Pannier L, Mott SH, et al. Functional anatomy of cognitive development: fMRI of verbal fluency in children and adults. Neurology 2000;54:1805.
  • 12
    Lee BC, Kuppusamy K, Grueneich R, et al. Hemispheric language dominance in children demonstrated by functional magnetic resonance imaging. J Child Neurol 1999;14:7882.
  • 13
    Hertz-Pannier L, Gaillard WD, Mott SH, et al. Noninvasive assessment of language dominance in children and adolescents with functional MRI: a preliminary study. Neurology 1997;48:100312.
  • 14
    Woods RP, Dodrill CB, Ojemann GA. Brain injury, handedness, and speech lateralization in a series of amobarbital studies. Ann Neurol 1988;23:5108.
  • 15
    Devinsky O, Perrine K, Llinas R, et al. Anterior temporal language areas in patients with early onset of temporal lobe epilepsy. Ann Neurol 1993;34:72732.
  • 16
    Anderson DP, Harvey AS, Saling MM, et al. Differential functional magnetic resonance imaging language activation in twins discordant for a left frontal tumor. J Child Neurol 2002;17:7669.
  • 17
    Hertz-Pannier L, Chiron C, Jambaque I, et al. Late plasticity for language in a child's non-dominant hemisphere: a pre- and post-surgery fMRI study. Brain 2002;125:36172.
  • 18
    Jayakar P, Bernal B, Santiago Medina L, et al. False lateralization of language cortex on functional MRI after a cluster of focal seizures. Neurology 2002;58:4902.
  • 19
    Benson RR, Logan WJ, Cosgrove GR, et al. Functional MRI localization of language in a 9-year-old child. Can J Neurol Sci 1996;23:2139.
  • 20
    Ries ML, Boop FA, Griebel ML, et al. Functional MRI and Wada determination of language lateralization: a case of crossed dominance. Epilepsia 2004;45:859.
  • 21
    Booth JR, MacWhinney B, Thulborn KR, et al. Developmental and lesion effects in brain activation during sentence comprehension and mental rotation. Dev Neuropsychol 2000;18:13969.
  • 22
    Stapleton SR, Kiriakopoulos E, Mikulis D, et al. Combined utility of functional MRI, cortical mapping, and frameless stereotaxy in the resection of lesions in eloquent areas of brain in children. Pediatr Neurosurg 1997;26:6882.
  • 23
    Wechsler D. WISC-III Wechsler Intelligence Scale for Children, 3rd ed. San Antonio : Psychological Corporation, 1991.
  • 24
    Dunn LM, Dunn L. Manual for the Peabody Picture Vocabulary Test, revised. Circle Pines : American Guidance Service, 1981.
  • 25
    Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971;9:97113.
  • 26
    Benton AL. Differential behavioural effects in frontal lobe disease. Neuropsychologia 1968;6:5360.
  • 27
    Anderson V, LaJoie G, Bell R. Neuropsychological assessment of the school-aged child. Melbourne : V Anderson, 1995.
  • 28
    Wood AG, Saling MM, Abbott DF, et al. A neurocognitive account of frontal lobe involvement in orthographic lexical retrieval: an fMRI study. Neuroimage 2001;14:1629.
  • 29
    Pujol J, Deus J, Losilla JM, et al. Cerebral lateralization of language in normal left-handed people studied by functional MRI. Neurology 1999;52:103843.
  • 30
    Abbott D, Jackson G. iBrain: software for analysis and visualisation of functional MR images [Abstract]. Neuroimage 2001;13:S59.
  • 31
    Van Der Kallen BF, Morris GL, Yetkin FZ, et al. Hemispheric language dominance studied with functional MR: preliminary study in healthy volunteers and patients with epilepsy. AJNR Am J Neuroradiol 1998;19:737.
  • 32
    Rasmussen T, Milner B. The role of early left-brain injury in determining lateralization of cerebral speech functions. Ann N Y Acad Sci 1977;299:35569.
  • 33
    Springer JA, Binder JR, Hammeke TA, et al. Language dominance in neurologically normal and epilepsy subjects: a functional MRI study. Brain 1999;122:203346.
  • 34
    Duchowny M, Jayakar P, Harvey AS, et al. Language cortex representation: effects of developmental versus acquired pathology. Ann Neurol 1996;40:318.
  • 35
    Vargha-Khadem F, O'Gorman AM, Watters GV. Aphasia and handedness in relation to hemispheric side, age at injury and severity of cerebral lesion during childhood. Brain 1985;108:67796.
  • 36
    Satz P, Strauss E, Wada J, et al. Some correlates of intra- and interhemispheric speech organization after left focal brain injury. Neuropsychologia 1988;26:34550.
  • 37
    Ojemann G, Ojemann J, Lettich E, et al. Cortical language localization in left, dominant hemisphere: an electrical stimulation mapping investigation in 117 patients. J Neurosurg 1989;71:31626.
  • 38
    DeVos K, Wyllie E, Geckler C, et al. Language dominance in patients with early childhood tumors near left hemisphere language areas. Neurology 1995;45:34956.
  • 39
    Kirschstein T, Fernandez G, Grunwald T, et al. Hetertopias, cortical dysplasias and glioneural tumors participate in cognitive processing in patients with temporal lobe epilepsy. Neurosci Lett 2003;338:23741.
  • 40
    Vargha-Khadem F, Carr LJ, Issacs E, et al. Onset of speech after left hemispherectomy in a nine-year-old boy. Brain 1997;120:15982.
  • 41
    Vining EP, Freeman JM, Pillas DJ, et al. Why would you remove half a brain? The outcome of 58 children after hemispherectomy: the Johns Hopkins experience: 1968 to 1996. Pediatrics 1997;100:6371.
  • 42
    Schlaggar BL, Brown TT, Lugar HM, et al. Functional neuroanatomical differences between adults and school-aged children in the processing of single words. Science 2002;296:14769.