Clinical functional MRI of the language domain in children with epilepsy

Clinical Speech fMRI in Children and Adolescents : Development of an Optimal Protocol and Analysis Algorithm

Purpose

In patients with drug-resistant focal epilepsy, surgical resection is often the only treatment option to achieve long-term seizure control. Prior to brain surgery involving potential language areas, identification of hemispheric language dominance is crucial. Our group developed and validated a functional magnetic resonance imaging (fMRI) battery of four pediatric language tasks. The present study aimed at optimizing fMRI data acquisition and analysis using these tasks.

Methods

We retrospectively analyzed speech fMRI examinations of 114 neuropediatric patients (age range 5.8–17.8 years) who were examined prior to possible epilepsy surgery. In order to evaluate hemispheric language dominance, 1–4 language tasks (vowel identification task VIT, word-chain task WCT, beep-story task BST, synonym task SYT) were measured.

Results

Language dominance was classified using fMRI activation in the 13 validly lateralizing ROIs (VLR) in frontal, temporal and parietal lobes and cerebellum of the recent validation study from our group: 47/114 patients were classified as left-dominant, 34/114 as bilateral and 6/114 as right-dominant. In an attempt to enlarge the set of VLR, we then compared for each task agreement of these ROI activations with the classified language dominance. We found four additional task-specific ROIs showing concordant activation and activation in ≥ 10 sessions, which we termed validly lateralizing (VLR_new). The new VLRs were: for VIT the temporal language area and for SYT the middle frontal gyrus, the intraparietal sulcus and cerebellum. Finally, in order to find the optimal sequence of measuring the different tasks, we analyzed the success rates of single tasks and all possible task combinations. The sequence 1) VIT 2) WCT 3) BST 4) SYT was identified as the optimal sequence, yielding the highest chance to obtain reliable results even when the fMRI examination has to be stopped, e.g., due to lack of cooperation.

Conclusion

Our suggested task order together with the enlarged set of VLR_new may contribute to optimize pediatric speech fMRI in a clinical setting.

Supplementary Information

The online version of this article (10.1007/s00062-021-01097-z) contains supplementary material, which is available to authorized users.

Presurgical Language fMRI in Children, Adolescents and Young Adults : A Validation Study

PURPOSE

To validate four established, child-friendly functional magnetic resonance imaging (fMRI) language tasks (word chain task [WCT], vowel identification task [VIT], synonym task [SYT] and beep story task [BST]) in a predominantly pediatric cohort.

METHODS

Intracarotid amobarbital procedures (IAP) (n = 17) and unchanged language after hemispherotomy (n = 6) were used as gold standards. The fMRI activations of nine regions of interest (ROI) in the frontal, temporal and parietal lobes as well as in the cerebellum were visually assessed in 23 fMRI examinations (in total 117 fMRI task sessions) of 23 patients (age range 10.0-23.0 years) with drug-refractory epilepsies.

RESULTS

The ROIs were considered valid when they showed activation in more than 25% of all sessions for the respective task and never showed false lateralization (in comparison to gold standards). Thus, 13 valid, task-specific ROIs were identified: 5 ROIs for the WCT (frontal operculum, inferior frontal gyrus, middle frontal gyrus, intraparietal sulcus, cerebellum), 3 ROIs for the VIT (frontal operculum, inferior frontal gyrus, middle frontal gyrus), 3 ROIs for the SYT (frontal operculum, inferior frontal gyrus, temporal language area) and 2 ROIs for the BST (inferior frontal gyrus, middle frontal gyrus).

CONCLUSION

Clinical fMRI using the battery of four tasks is a valid tool for lateralizing language in children, adolescents and young adults. Each task proved to be specifically useful, which confirms that applying different tasks increases the probability of diagnosing language dominance in presurgical candidates.

Functional magnetic resonance imaging

Among the range of methods available to assess neurodevelopmental disorders, functional MRI (fMRI) has been a preferred tool of choice. Indeed, fMRI can reveal functional alterations in brain networks, irrespective of their structural integrity. Yet, whether fMRI studies have provided unique added value and influenced the clinical care and assessments in children with these conditions remains controversial. This chapter aims to give an overview of the clinical use of task-based as well as resting-state fMRI in children with neurodevelopmental disorders, such as dyslexia, DLD, and epilepsy. We introduce analysis methods that appear promising (namely PPI and machine learning) and describe strengths and limitations of fMRI in the field of pediatrics. Altogether, we suggest that fMRI has provided us with a unique understanding of some developmental conditions. Indeed, findings from group studies have both informed neuroanatomical models and revealed compensation mechanisms. In addition, improvements have made fMRI an increasingly child-friendly method. Nevertheless, clinicians should be aware of limitations, including (1) lack of replication of results, (2) the limited specificity as a diagnostic tool, and (3) difficulties with interpretation of findings. The use of fMRI in the clinic currently remains restricted, with the exception of epilepsy surgery planning, where it is used routinely.

The role of task-based neural activation research in understanding cognitive deficits in pediatric epilepsy

Children with epilepsy can experience significant cognitive dysfunction that can lead to academic underachievement. Traditionally believed to be primarily due to the effects of factors such as the chronicity of epilepsy, medication effects, or the location of the primary epileptogenic lesion;, recent evidence has indicated that disruption of cognition-specific distributed neural networks may play a significant role as well. Specifically, over the last decade, researchers have begun to characterize the mechanisms underlying disrupted cognitive substrates by evaluating neural network abnormalities observed during specific cognitive tasks, using task-based functional magnetic resonance imaging (fMRI). This targeted review assesses the current literature investigating the relationship between neural network abnormalities and cognitive deficits in pediatric epilepsy. The findings indicate that there are indeed neural network abnormalities associated with deficits in executive function, language, processing speed, and memory. Overall, cognitive dysfunction in pediatric epilepsy is associated with a decrease in neural network activation/deactivation as well as increased recruitment of brain regions not typically related to the specific cognitive task under investigation. The research to date has focused primarily on children with focal epilepsy syndromes with small sample sizes and differing research protocols. More extensive research in children with a wider representation of epilepsy syndromes (including generalized epilepsy syndromes) is necessary to fully understand these relationships and begin to identify underlying cognitive phenotypes that may account for the variability observed across children with epilepsy. Furthermore, more uniformity in fMRI protocols and neuropsychological tasks would be ideal to advance this literature.

Quantifying the Confidence in fMRI-Based Language Lateralisation Through Laterality Index Deconstruction

In epilepsy patients, language lateralisation is an important part of the presurgical diagnostic process. Using task-based fMRI, language lateralisation can be determined by visual inspection of activity patterns or by quantifying the difference in left- and right-hemisphere activity using variations of a basic formula [(L-R)/(L+R)]. However, the values of this laterality index (LI) depend on the choice of activity thresholds and regions of interest. The diagnostic utility of the LI also depends on how its continuous values are translated into categorical decisions about a patient's language lateralisation. Here, we analysed fMRI data from 712 epilepsy patients who performed a verbal fluency task. Each fMRI data set was evaluated by a trained human rater as depicting left-sided, right-sided, or bilateral lateralisation or as being inconclusive. We used data-driven methods to define the activity thresholds and regions of interest used for LI computation and to define a classification scheme that allowed us to translate the LI values into categorical decisions. By deconstructing the LI into measures of laterality (L-R) and strength (L+R), we also modelled the relationship between activation strength and conclusiveness of a data set. In a held-out data set, predictions reached 91% correct when using only conclusive data and 82% when inconclusive data were included. Although only trained on human evaluations of fMRIs, the approach generalised to the prediction of language Wada test results, allowing for significant above-chance accuracies. Compared against different existing methods of LI-computation, our approach improved the identification and exclusion of inconclusive cases and ensured that decisions for the remaining data could be made with consistently high accuracies. We discuss how this approach can support clinicians in assessing fMRI data on a single-case level, deciding whether lateralisation can be determined with sufficient certainty or whether additional information is needed.

Assessing motor, visual and language function using a single 5-minute fMRI paradigm: three birds with one stone

Clinical functional Magnetic Resonance Imaging (fMRI) requires inferences on localization of major brain functions at the individual subject level. We hypothesized that a single "triple use" task would satisfy sensitivity and reliability requirements for successfully assessing the motor, visual and language domain in this context. This was tested here by the application in a group of healthy adults, assessing sensitivity and reliability at the individual subject level, separately for each domain.Our "triple use" task consisted of 2 conditions (condition 1, assessing motor and visual domain, and condition 2, assessing the language domain), serving mutually as active/control. We included 20 healthy adult subjects. Random effect analyses showed activation in primary motor, visual and language regions, as expected. Less expected regions were activated both for the motor and visual domains. Further, reliability of primary activation patterns was very high across individual subjects, with activation seen in 70-100% of subjects in primary motor, visual, and left-lateralized language regions.These findings suggest the "triple use" task to be reliable at the individual subject's level to assess motor, visual and language domains in the clinical fMRI context. Benefits of such an approach include shortening of acquisition time, simplicity of the task for each domain, and using a visual stimulus. Following establishment of reliability in adults, the task may also be a valuable addition in the pediatric clinical fMRI context, where each of these factors is of high relevance.

Lesion characteristics driving right-hemispheric language reorganization in congenital left-hemispheric brain damage

Pre- or perinatally acquired ("congenital") left-hemispheric brain lesions can be compensated for by reorganizing language into homotopic brain regions in the right hemisphere. Language comprehension may be hemispherically dissociated from language production. We investigated the lesion characteristics driving inter-hemispheric reorganization of language comprehension and language production in 19 patients (7-32years; eight females) with congenital left-hemispheric brain lesions (periventricular lesions [n=11] and middle cerebral artery infarctions [n=8]) by fMRI. 16/17 patients demonstrated reorganized language production, while 7/19 patients had reorganized language comprehension. Lesions to the insular cortex and the temporo-parietal junction (predominantly supramarginal gyrus) were significantly more common in patients in whom both, language production and comprehension were reorganized. These areas belong to the dorsal stream of the language network, participating in the auditory-motor integration of language. Our data suggest that the integrity of this stream might be crucial for a normal left-lateralized language development.

A "one size fits all" approach to language fMRI: increasing specificity and applicability by adding a self-paced component

We have previously established an fMRI task battery suitable for mapping the language processing network in children. Among the tasks used, the synonyms and the vowel identification task induced robust task-related activations in children with average language abilities; however, the fixed presentation time seems to be a drawback in participants with above- or below-average language abilities. This feasibility study in healthy adults (n = 20) was aimed at adapting these tasks to the individual level of each patient by implementing a self-paced stimulus presentation. The impact of using a block- versus an event-related statistical approach was also evaluated. The self-paced modification allowed our participants with above-average language abilities to process stimuli much faster than originally implemented, likely increasing task adherence. A higher specificity of the event-related analysis was confirmed by stronger left inferior frontal and crossed cerebellar activations. We suggest that self-paced paradigms and event-related analyses may both increase specificity and applicability.

Applications of blood-oxygen-level-dependent functional magnetic resonance imaging and diffusion tensor imaging in epilepsy

The lifetime prevalence of epilepsy ranges from 2.7 to 12.4 per 1000 in Western countries. Around 30% of patients with epilepsy remain refractory to antiepileptic drugs and continue to have seizures. Noninvasive imaging techniques such as functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI) have helped to better understand mechanisms of seizure generation and propagation, and to localize epileptic, eloquent, and cognitive networks. In this review, the clinical applications of fMRI and DTI are discussed, for mapping cognitive and epileptic networks and organization of white matter tracts in individuals with epilepsy.

Processing of non-canonical word-order: a case-series on lesion-induced reorganized language and age-effects in typical development

Complex grammatical structures are mastered late in language acquisition. We studied age-effects on performance in object topicalization in 48 typically developing German-speaking participants (aged 8-30years) and in five patients (children and adolescents) with lesion-induced atypical language representation. Production was tested by a sentence repetition task, comprehension by an acting out task. Three topicalized conditions with differing disambiguation (agreement, case, and case plus agreement) were contrasted with canonical control sentences. Children's (aged 8-13years) performance was significantly below that of adolescents and adults in all topicalized conditions. All participants made most mistakes in the agreement condition. Patients showed remarkable difficulties as compared with age-appropriate control groups in all topicalization conditions and across age-groups. Despite the small sample size, the consistency of these difficulties might hint to the importance of an intact typical neural language substrate for processing complex grammatical structures even in very early brain lesions.

Effects of dual pathology on cognitive outcome following left anterior temporal lobectomy for treatment of epilepsy

The objective of this retrospective study was to determine if dual pathology [DUAL - focal cortical dysplasia (FCD) and mesial temporal sclerosis (MTS)] in patients with left temporal lobe epilepsy is associated with greater risk for cognitive decline following temporal lobectomy than single pathology (MTS only). Sixty-three adults (Mage=36.5years, female: 52.4%) who underwent left anterior temporal lobectomy for treatment of epilepsy (MTS=28; DUAL=35) completed preoperative and postoperative neuropsychological evaluations. The base rate of dual pathology was 55.5%. Repeated measures ANOVAs yielded significant 2-way interactions (group×time) on most measures of language and memory with generally moderate effect sizes. Specifically, patients with MTS only demonstrated postoperative declines, while those with dual pathology remained unchanged or improved. Results suggest that dual pathology may be associated with better cognitive outcome following epilepsy surgery than MTS alone, possibly reflecting limited functionality of the resected tissue or intrahemispheric reorganization of function in the context of a developmental lesion.

Classification of fMRI patterns--a study of the language network segregation in pediatric localization related epilepsy

This article describes a pattern classification algorithm for pediatric epilepsy using fMRI language-related activation maps. 122 fMRI datasets from a control group (64) and localization related epilepsy patients (58) provided by five children's hospitals were used. Each subject performed an auditory description decision task. Using the artificial data as training data, incremental Principal Component Analysis was used in order to generate the feature space while overcoming memory requirements of large datasets. The nearest-neighbor classifier (NNC) and the distance-based fuzzy classifier (DFC) were used to perform group separation into left dominant, right dominant, bilateral, and others. The results show no effect of age, age at seizure onset, seizure duration, or seizure etiology on group separation. Two sets of parameters were significant for group separation, the patient vs. control populations and handedness. Of the 122 real datasets, 90 subjects gave the same classification results across all the methods (three raters, LI, bootstrap LI, NNC, and DFC). For the remaining datasets, 18 cases for the IPCA-NNC and 21 cases for the IPCA-DFC agreed with the majority of the five classification results (three visual ratings and two LI results). Kappa values vary from 0.59 to 0.73 for NNC and 0.61 to 0.75 for DFC, which indicate good agreement between NNC or DFC with traditional methods. The proposed method as designed can serve as an alternative method to corroborate existing LI and visual rating classification methods and to resolve some of the cases near the boundaries in between categories.

Functional MRI and structural MRI as tools for understanding comorbid conditions in children with epilepsy

Children with epilepsy are at risk for behavioral and cognitive comorbidities. Potential etiologies can be assessed in part by neuroimaging. Functional magnetic resonance imaging (MRI) has a major role in presurgical evaluation and prediction of postoperative outcome by mapping of language and memory. Structural MRI and functional MRI have shown changes in children and adolescents with attention deficit hyperactivity disorder and disruptive behavior, common comorbidities in children with epilepsy. Neuroimaging has the potential for significantly increasing understanding of the basis of cognitive and behavioral problems in children with epilepsy.

Clinical applications and future directions of functional MRI

First described for use in mapping the human visual cortex in 1991, functional magnetic resonance imaging (fMRI) is based on blood-oxygen level dependent (BOLD) changes in cortical regions that occur during specific tasks. Typically, an overabundance of oxygenated (arterial) blood is supplied during activation of brain areas. Consequently, the venous outflow from the activated areas contains a higher concentration of oxyhemoglobin, which changes the paramagnetic properties of the tissue that can be detected during a T2-star acquisition. fMRI data can be acquired in response to specific tasks or in the resting state. fMRI has been widely applied to studying physiologic and pathophysiologic diseases of the brain. This review will discuss the most common current clinical applications of fMRI as well as emerging directions.

An iterative jackknife approach for assessing reliability and power of FMRI group analyses

For functional magnetic resonance imaging (fMRI) group activation maps, so-called second-level random effect approaches are commonly used, which are intended to be generalizable to the population as a whole. However, reliability of a certain activation focus as a function of group composition or group size cannot directly be deduced from such maps. This question is of particular relevance when examining smaller groups (<20–27 subjects). The approach presented here tries to address this issue by iteratively excluding each subject from a group study and presenting the overlap of the resulting (reduced) second-level maps in a group percent overlap map. This allows to judge where activation is reliable even upon excluding one, two, or three (or more) subjects, thereby also demonstrating the inherent variability that is still present in second-level analyses. Moreover, when progressively decreasing group size, foci of activation will become smaller and/or disappear; hence, the group size at which a given activation disappears can be considered to reflect the power necessary to detect this particular activation. Systematically exploiting this effect allows to rank clusters according to their observable effect size. The approach is tested using different scenarios from a recent fMRI study (children performing a “dual-use” fMRI task, n = 39), and the implications of this approach are discussed.

Left-handedness and language lateralization in children

This fMRI study investigated the development of language lateralization in left- and righthanded children between 5 and 18 years of age. Twenty-seven left-handed children (17 boys, 10 girls) and 54 age- and gender-matched right-handed children were included. We used functional MRI at 3T and a verb generation task to measure hemispheric language dominance based on either frontal or temporo-parietal regions of interest (ROIs) defined for the entire group and applied on an individual basis. Based on the frontal ROI, in the left-handed group, 23 participants (85%) demonstrated left-hemispheric language lateralization, 3 (11%) demonstrated symmetric activation, and 1 (4%) demonstrated right-hemispheric lateralization. In contrast, 50 (93%) of the right-handed children showed left-hemispheric lateralization and 3 (6%) demonstrated a symmetric activation pattern, while one (2%) demonstrated a right-hemispheric lateralization. The corresponding values for the temporo-parietal ROI for the left-handed children were 18 (67%) left-dominant, 6 (22%) symmetric, 3 (11%) right-dominant and for the right-handed children 49 (91%), 4 (7%), 1 (2%), respectively. Left-hemispheric language lateralization increased with age in both groups but somewhat different lateralization trajectories were observed in girls when compared to boys. The incidence of atypical language lateralization in left-handed children in this study was similar to that reported in adults. We also found similar rates of increase in left-hemispheric language lateralization with age between groups (i.e., independent of handedness) indicating the presence of similar mechanisms for language lateralization in left- and right-handed children.

Assessing language and visuospatial functions with one task: a "dual use" approach to performing fMRI in children

In order to increase the rate of successful functional MR studies in children it is helpful to shorten the time spent in the scanner. To this effect, assessing two cognitive functions with one task seems to be a promising approach. The hypothesis of this study was that the control condition of an established language task (vowel identification task, VIT) requires visuospatial processing and that the control condition (VIT(CC)) therefore may also be applicable to localize visuospatial functions. As a reference task, a visual search task (VST, previously established for use in children) was employed. To test this hypothesis, 43 children (19 f, 24 m; 12.0±2.6, range 7.9 to 17.8 years) were recruited and scanned using both tasks. Second-level random effects group analyses showed activation of left inferior-frontal cortex in the active condition of the VIT, as in previous studies. Additionally, analysis of the VIT(CC) demonstrated activation in right-dominant superior parietal and high-frontal brain regions, classically associated with visuospatial functions; activation seen in the VST was similar with a substantial overlap. However, lateralization in the parietal lobe was significantly more bilateral in the VST than in the VIT(CC). This suggests that the VIT can not only be applied to assess language functions (using the active>control contrast), but also that the control>active condition is useful for assessing visuospatial functions. Future task design may benefit from such a "dual use" approach to performing fMRI not only, but also particularly in children.