Slide Session I: Neuroimaging

PET Imaging of mGluR5 with [18F]FPEB in Epilepsy

Presenting Author: Mehdi Djekidel

Collaborating Authors: Beata Planeta Wilson, Jenna Sullivan, Evans Morris, Tore Eid, Hitten Zaveri, Colleen Malone, Nihal Chandra deLanerolle, Goncharova Irina, Jullie Pan, Richard Bronen, Pue Farooque, Richard Mattson, Kamil Detynieki, Robert Duckrow, Kenneth Vives, Lawrence Hirsch, Hal Blumenfeld, Dennis Spencer, Richard Carson

Electrochemical imbalances occurring in epilepsy are not well understood. The glutamate system including metabotropic glutamate 5-receptors (mGluR-5) has been implicated in animal and human models of epilepsy. Imaging plays a cornerstone role in epilepsy evaluation. Positron emission tomography (PET) is an optimal modality to noninvasively explore these processes. PET evaluation of glucose metabolism, benzodiazepine and serotonin receptors has been successfully performed in humans and has shown decreases in the seizure onset zone (SOZ) interictally. We seek to gain a new perspective on the role glutamate receptors play in epilepsy, through PET imaging of mGluR-5 changes using [18F]FPEB. Our study includes patients with severe medically refractory epilepsy whom are being evaluated for surgical resection of the SOZ. We identify the SOZ based on a gold standard of intracranial electrico-corticography teager energy measurements, which we then correlate with FPEB imaging results. Our preliminary data includes imaging of 5 subjects with medically refractory epilepsy with an age ranging from 22-53 yo (average age 33.6) and a 4/1 female to male ratio. We found interictally decreased mGluR5 distribution in the SOZ when compared to a normal healthy control reference -constructed from eight healthy control subjects-. Additionally a correlation with glutamate microdialysis data obtained during IEC showed no direct correlation between mGluR5 volume of distribution obtained from [18F]FPEB scans and glutamate levels in three subjects. Further exploration in a larger cohort –which is the goal of our ongoing project- is needed to further validate these results and potentially explore their value in daily clinical practice.

Functional Connectivity of Cognition in Temporal Lobe Epilepsy

Presenting Author: F. Scott Winstanley

Collaborating Authors: Hyang Woon Lee, Jagriti Arora, Dennis D. Spencer, R. Todd Constable

Rationale: Medically refractory mesial temporal lobe epilepsy (MTLE) is associated with cognitive impairment, reduced daily functioning, and overall decreased quality of life. Anterior temporal lobectomy (ATL) is an effective treatment for medically refractory temporal lobe epilepsy; however, this procedure carries the risk of additional postoperative cognitive morbidity. Identifying preoperative neural functioning as well as identifying risk factors for postoperative cognitive morbidity is an essential part of the evaluation for epilepsy surgery. Recent advances in fMRI have demonstrated that resting state intrinsic connectivity contrast (rs-ICC) analysis of fMRI BOLD signal fluctuations has also shown to be a powerful tool in identifying neural underpinnings for various cognitive functions in normal subjects. Analysis of ICC data can identify the tissue elements most affected in TLE, potentially assisting in the management of these patients. This study aims to 1) elucidate neural correlates of cognitive impairment by analyzing the fMRI intrinsic connectivity contrast (ICC) in medically refractory TLE patients; and 2) determine if these ICC correlations can predict postoperative cognitive morbidity in ATL patients.

Methods: We recruited 25 TLE patients who performed resting state fMRI, underwent neuropsychological tests as part of presurgical evaluation, and underwent standard temporal lobectomy with at least one year of post operative follow-up. There were 12 left and 13 right TLE patients, 9 males and 16 females with mean age at surgery of 39.1 ± 11.1 years old. A gradient-echo fMRI sequence was obtained from each patient, and the ICC was measured by a voxel based calculation of the degree of correlation coefficient for each voxel within the brain hemisphere and within medial or medial/lateral temporal lobes to ipsilateral (ipsi-ICC) and contralateral (contra-ICC) to the seizure focus. Analysis of rs-ICC was correlated with preoperative and postoperative scores obtained on a battery of neuropsychological tests including the verbal and non-verbal Selective Reminding Tests (SRT), the California Verbal Learning Test (CVLT), and the Boston Naming test (BNT).

Results: Verbal SRT scores were positively correlated with ipsi-ICC in left hippocampus and contra-ICC in both hippocampi using medial temporal mask. Verbal SRT showed similar positive correlation with ipsi- and contra-ICC values with whole brain mask in left middle and part of inferior temporal gyri. Impaired non-verbal SRT was correlated with decreased ipsi- and contra-ICC in right hippocampus, right superior temporal gyrus with medial/lateral temporal mask. Decreased CVLT scores were correlated with decreased ipsi- and contra-ICC in left middle temporal gyrus, and BNT showed similar positive correlation with left hippocampus, left superior/middle temporal and parahippocampal gyri, using whole brain mask.

Conclusion: These findings suggest that impairments in cognitive function in TLE patients are reflected in the intrinsic functional organization of the brain as measured by ICC. Impairments of verbal learning and naming in TLE patients showed alterations in dominant temporal lobe structures while non-verbal learning abnormalities were closely related to ICC changes in non-dominant temporal structures, in good agreement with the previous knowledge of memory and language networks. This study also demonstrated a relationship between pre and postoperative test scores and rs-ICC, indicating rs-ICC may be a useful predictor of cognitive outcome in patients undergoing ATL.

The Face of Category-Selectivity in Human Visual Cortex

Presenting Author: Andrew Engell

Collaborating Authors: Gregory McCarthy

The ultimate goal of cortical resection is to disrupt pathological processes while preserving normal function. Stimulation mapping or fMRI are thus used identify those regions that are critical to typical cognitive, perceptual, or sensorimotor behavior. Importantly, this approach is predicated on understanding the extent of localized function in a given region, and what role the region might play in a broader brain network. Here we investigate the properties of a core set of brain regions that area activated by face perception. Although the role of these regions as 'face-selective' patches of cortex has been reinforced by twenty years of fMRI research, we argue that they play a broader role in the perception of animate agents. For instance, it has been shown that viewing biological motion evokes a greater response in these areas than does non-biological motion. The current study uses a large sample of fMRI data (N > 100) to pursue three aims: 1) investigate the nature of the overlap between face and biological motion activation maps, 2) define and contrast the spatial reliability of activation across key nodes (lateral temporal, ventral temporal, and occipitotemporal cortices) of the face and biological motion networks, and 3) create probabilistic atlases of visual-category perception. We find that face and biological motion activations substantially overlap, but nonetheless have spatially separable peaks. However, in the ventral temporal and inferior occipitotemporal cortices these peaks do not maintain a consistent spatial relationship. We also characterize the spatial reliability of the response to each task and find that the Euclidean distance of the category peaks measured at different times (within-category variance) is not substantially different from the difference between face and biological motion peaks (between-category variance). Finally, we make freely available probabilistic atlases of face-, biological motion-, scene-, and house-selective BOLD activations.

3D Color Brain Movies to Improve Seizure Localization with >200 Channel Intracranial EEG

Presenting Author: Pue Farooque

Collaborating Authors: William Chen, Hyang Woon Lee, Mark W. Youngblood, Xiao Han, Stephen Jhun, Irina Goncharova, Kenneth Vives, Dennis D. Spencer, Hitten Zaveri, Lawrence J. Hirsch, Hal Blumenfeld

In cases of medically intractable epilepsy, intracranial EEG (icEEG) is the gold standard for identifying the seizure generating region. As rapid advancements in acquisition technology have led to improvements in the spatial resolution of icEEG data, the complexity of analyzing a large number of 2D, un-contextualized data channels has become a major impediment to understanding icEEG information. In an attempt to simplify data analysis and increase reviewer consistency, this study aimed to develop a new technique for clinical seizure localization using 3D color movies. The clinical usefulness of this technique was tested through a patient-by-patient comparison with a traditional icEEG analysis and again with surgical resection area. We analyzed 140 seizures from 48 intractable epilepsy patients (28 temporal and 20 extratemporal) with icEEG recordings, epilepsy surgery, and at least one year of postsurgical follow-up. The icEEG data was processed by performing a fast-fourier transform on the calculated signal power and the Beta frequency band was plotted over time onto a MRI-based surface rendering of each patient's cortex. Two independent blinded reviewers interpreted the icEEG data using a conventional visual analysis followed by 3D movies. Agreement between two reviewers was good, especially for 3D movies. Overall agreement rates between 3D movies and icEEG visual analysis or surgery were around 90% for side, 80% for lobe, and slightly lower than 80% for sublobar using a fixed Maximum Power Coloration Value (MPCV) and the Beta frequency range. These agreement rates were improved when variable MPCV and frequency range was allowed during review, especially for sublobar. In general seizure localization using 3D movies was comparable with the traditional icEEG visual analysis, but was measurably faster and considered qualitatively easier by both reviewers. These findings suggest that 3D movies could be a useful diagnostic method for presurgical seizure localization in patients with intractable epilepsy.

EEG and fMRI Correlates of Behavior during Childhood Absence Seizures

Presenting Author: Jennifer Guo

Collaborating Authors: Stephen Jhun, Robert Kim, Jorge Rodriquez-Fernandez, Adam Kundishora, Wendy Xiao, Xiaoiao Bai, Michiro Negishi, Hetal Mistry, Christopher Bailey, Michael Crowley, R. Todd Constable, Linda Mayes, Hal Blumenfeld

Seizures in childhood absence epilepsy (CAE) are characterized by 3-4 Hz spike-and-wave discharges on electroencephalography (EEG) and temporary loss of consciousness. While CAE is considered a generalized type of epilepsy, prior studies using functional magnetic resonance imaging (fMRI) show involvement of focal attentional and default mode networks, including the orbital/medial frontal cortex, medial/lateral parietal cortex, and thalamus. We recorded 256-lead high density EEG in 20 pediatric patients as well as simultaneous 32-lead EEG and fMRI in 34 patients as they performed behavioral tasks requiring attentional vigilance.

Overall, seizures caused greater impaired performance during the more difficult continuous performance task (CPT) compared to the less attentionally demanding repetitive tapping task (RTT). However, performance on both tasks varied from patient to patient and seizure to seizure. Behavioral performance was then related to EEG measures. Time-frequency analysis of ictal data on EEG show greater power changes in the 3-4Hz range for seizures associated with poor behavioral performance (>75% omissions) compared to seizures with good behavioral performance (<25% omissions). Source localization for seizure activity using both distributed and single dipole models localize spikes to medial frontal cortex. Timecourse of fMRI signal changes of seizure activity showed early increases in medial frontal cortex, bilateral insula, thalamus, then occipital cortex followed by decreases in a widespread frontal-parietal network. Finally, concordance between sources on high-density EEG and fMRI regions during seizures was determined. Characterizing electrophysiology and functional imaging signals during absence seizures may shed light on mechanisms of loss of consciousness during absence seizures.

Contribution of 7T MR Spectroscopic Imaging to Seizure Localization in Surgically Resected Epilepsy

Presenting Author: Jullie Pan

Collaborating Authors: RB Duckrow, J Gerrard, C Ong, L Hirsch, SR Resor Jr, Y Zhange, HP Hetherington, DD Spencer

Outcome studies in epilepsy surgery show it to be an excellent option for medically intractable epilepsy. However the challenge in these cases commonly depends on seizure localization with sufficient accuracy for surgery. In this initial report we evaluate the contribution of ultra-high field (7T) MR metabolic imaging in the process of localization for surgical epilepsy. Patient data are included in this analysis if their region of resective surgery overlapped with available MRSI data, resulting in a group of n=24 patients (a mix of medial temporal lobe MTL and neocortical epilepsy patients). The classification of interest was whether MRSI abnormalities completely agree with surgical resection (class 1), partially agreed (class 2), or did not agree (class 3). Two major outcome groups of ILAE I-III and IV-VI are used. With three classes and two outcomes, a 3x2 contingency table is devised. Of the n=24 patients, 16 underwent intracranial EEG monitoring for seizure localization. All MTL patients with MRI=defined asymmetric hippocampal atrophy (n=8) demonstrated concordance between surgery and MRSI (class 1) and excellent outcome (I-III). Out of the n=16 remainder, n=5 good outcome (I-III) was found with class 1 concordance; 10 poor outcomes (IV-VI) were found if there was either partial concordance or discordance (class 2 or class 3). n=1 patient was inaccurately classified, with good outcome and class 3 discordance. We discuss the results of this initial analysis.




1 Concordant: region of surgical resection agrees with MRSI (NA/Cr) abnormality 13 0
2 Partial Concordance: region of surgical resection displays MRSI abnormality but there are other regions of abnormality at least or worse 0 6
3 Discordance: region of surgical resection does not display MRSI abnormality 1 4