Functional Activation Within Hippocampal Subfields During Scene Memory Encoding In Temporal Lobe Epilepsy

Introduction We present, for the first time, functional activation patterns within hippocampal subfields in temporal lobe epilepsy (TLE) patients. The hippocampus is a complex structure consisting of subregions that express varied cell types, and are differentially affected by neurological disorders, including TLE. For example, dentate gyrus (DG) is one subfield that has been implicated in both histological [1], and more recently, in structural imaging studies [2] as an epileptogenic site. While functional activation as well as inter-hemispheric functional asymmetry within the larger whole hippocampal region of interest (ROI) has been shown to have predictive value for cognitive outcome following epilepsy surgery, as well as in pre-surgical lateralization of function, more focal functional measurements within subfield ROIs haven’t yet been studied. We use an atlas-based method to label subfields in structural MRI and report functional activation within these ROIs. We detect group differences in subfield activation between controls and patients and between patients’ epileptogenic and non-epileptogenic sides, as well as differential activation of DG and CA1 in controls, but not in patients. Methods Eighteen subjects with refractory TLE and 19 healthy controls were imaged in a 3 Tesla Siemens Trio scanner using an eight-channel head coil and body coil transmitter. T1-weighted structural MRI scans were obtained using the MP-RAGE sequence with the following parameters: TR=1620 ms, TE=3.87 ms, TI=950 ms, flip angle=15, and voxel size=0.94x0.94x1 mm. BOLD fMRI images were obtained using a gradient echoplanar (EPI) sequence with TR=3000 ms, TE=30 ms, and 3-mm isotropic voxels during a blocked design experiment consisting of a complex scene memory encoding task (see details in [3]). A binary segmentation of the whole hippocampus is obtained from the subject’s anatomical MRI using the method described in [4]. Six hippocampal subfield ROIs (Head, CA1, CA2, CA3, DG, Tail) were then labeled using shape-based normalization of this whole hippocampus mask to a postmortem hippocampus atlas [5] containing subfield labels. This mapping preserves the relative depth of structures in the subject space. Head and Tail were given separate labels as it’s harder to distinguish individual subfields in these regions. The EPI data were motion corrected, and then aligned to and resampled in the high-resolution space of the structural MRI, which helps increase the effective resolution of measured activation [6]. We do not perform any spatial smoothing, which helps preserve spatial specificity. A general linear model was used to generate task activation maps using Statistical Parametric Mapping (SPM5) software [7]. The contrast images (Fig. 1) were sampled within the subfield ROIs. Task contrast was integrated over each subfield to generate ROI-based activation measurements.