Abstracts

Rationale: Functional MRI (fMRI) studies of memory encoding in adults with temporal lobe epilepsy (TLE) show promise as a potential noninvasive replacement for the Wada test for presurgical memory lateralization. Complex visual scene-encoding activates both the left and right mesial temporal lobe (mTL) in healthy adult controls, and shows asymmetries in patients with TLE. The utility of presurgical fMRI lateralization has not been assessed in children. We created a child friendly version of this task to be applied to healthy children and children with TLE to determine if clinically significant mTL asymmetries exist. Methods: Subjects underwent at least two 5-minute BOLD fMRI sessions, using high-resolution 2 mm isotropic echoplanar imaging on a Siemens 3 Tesla Trio scanner. A block design was used alternating complex visual scenes with a scrambled control image. Subjects were asked to remember the scenes for subsequent testing. Scanning was also performed during recognition testing with 36 targets mixed with 36 foils. Subject responses were recorded and a discrimination score (d’) was calculated and served as our measure of task performance. All image sets were processed using SPM5. Two regions of interest (ROI) were used; a hippocampal ROI as well as a broader hippocampal, parahippocampal, and fusiform (HPF) ROI. All positively activated voxels were extracted from each ROI and plotted as a function of its own t statistic. The area under this curve was used for calculation of a laterality index (LI) with the formula L(ROI)-R(ROI)/L(ROI)+R(ROI). Results: Twelve healthy control subjects and 4 TLE patients (3 left, 1 right) were studied. The mean age for controls was 11.3 years and for patients was 16.9 years. Patients and controls correctly identified an average of 30/36 scenes with only 2/36 false positives. During encoding, the mean laterality index for controls was –0.123 (std dev=0.228) and –0.112 (std dev= 0.349) in the HPF and hippocampal ROI’s, respectively. During recognition scanning a contrast was generated using “remembered” vs. “novel” scenes. The mean laterality index during recognition scanning was –0.193 (std dev 0.253) and –0.152 (std dev 0.474) in the HPF and hippocampal ROI’s, respectively. An example of an activation map from a healthy control is shown in Figure 1. All 4 TLE subjects had a LI within one standard deviation of the control group with a range of –0.23 to 0.049 in the HPF ROI and –0.052 to 0.074 in the hippocampal ROI. Only 2 of the TLE subjects (both left) had Wada tests performed, both of which suggested left greater than right support for memory.Conclusions: An fMRI complex visual scene-encoding task in healthy children results in symmetric activation of the mTL. In a small sample of adolescents with TLE, this fMRI task also revealed symmetric activation similar to healthy controls. This study supports further investigation using fMRI studies of memory in children and should be validated against the Wada procedure in a larger sample of children with temporal lobe epilepsy.