Prior to implantation with depth electrodes, 3-T high-resolution T2-weighted structural MR images of the hippocampus were obtained in all patients (TR 5.2 seconds, TE 105 msec, 19 slices, voxel size 0.4 × 0.4 × 3 mm). These parameters optimize signal strength in the hippocampal area, particularly in the anterior portion.13,14
Slices were obtained perpendicular to the long axis of the hippocampus. Patients also underwent 3-T whole-brain T1-weighted magnetization-prepared rapid acquisition gradient echo MR imaging (TR 1800 msec, TE 2.93 seconds, voxel size 0.9 × 0.9 × 0.8 mm) as part of depth electrode placement planning. Patients then underwent implantation of depth electrodes for surgical monitoring. At the tip of each electrode, a set of eight 40-μm platinum–iridium microwires provided possible cellular and electroencephalographic signals.6
After implantation with depth electrodes, patients underwent a spiral CT scan (1 second rotation, high-quality [also called HQ] mode, helical pitch 1.5, 1-mm slice collimation, and a 0.5-mm reconstruction interval delineated using axial slices) to localize electrodes. All patients who participated provided informed consent. All studies conformed to the guidelines of the Medical Institutional Review Board at our institution.
We performed a 3-way registration of the patients’ CT, whole-brain MR imaging, and high-resolution MR imaging studies to maximize our localization accuracy. This was done because we occasionally obtained unsatisfactory registrations between the high-resolution MR images and CT scans (> 1-mm translations). Three-way registrations were performed using BrainLab stereotactic and localization software (www.brainlab.com
The CT scan was first registered to the whole-brain MR image, followed by registration of the whole-brain image to the high-resolution MR imaging study. In most cases, microwire tracks were visible as artifacts in the coregistered CT–MR images (, Row 2
). By localizing microwires on the CT scan, we could then visualize these locations on the high-resolution and whole-brain MR image simultaneously. In all cases, anatomical locations targeted by microwire tips on the high-resolution MR images corresponded with those on the whole-brain imaging studies.
Fig. 1 Neuroimaging studies and maps used in localization of microelectrodes to hippocampal subregions. Row 1: High-resolution coronal–oblique MR images (perpendicular to the long axis of the hippocampus) obtained in 3 different patients. “Left” (more ...)
High-resolution structural scans were computationally unfolded and flattened from 3 dimensions to 2 dimensions, producing 2D maps oriented from bottom to top (posterior to anterior) along the long axis of the hippocampus. Computational unfolding involved several steps. We first segmented hippocampal gray matter by outlining white matter and cerebrospinal fluid along the hippocampus proper and extending through the fusiform cortex. To improve the quality of the segmentation and our overall voxel resolution within the hippocampus, slices were computationally and then manually interpolated by a factor of 7 along the long axis of the hippocampus, producing a final voxel size of 0.391 × 0.391 × 0.429 mm. Gray matter voxels were computationally connected using the region-expansion algorithm “MrGray.”10
This 3D gray matter strip was then computationally flattened into 2 dimensions by using “mrUnfold” software.3
Using landmarks on the 3D hippocampal scan, we defined anatomical boundaries of the following structures: CA2/CA3/dentate gyrus (labeled CA23DG); CA1–subiculum; subiculum–ERC; ERC–PRC; subiculum–PHC; and PHC–fusiform cortex (for example, the collateral sulcus).1,2
We also demarcated the beginning of the hippocampal head (anterior CA/dentate gyrus) and the ERC-PRC-PHC boundary. These boundaries were then projected into 2D hippocampal space, thus outlining the anatomical locations defined by these boundaries in 3D space. A more complete explanation of the methods involved in computational unfolding, including validation and precision of the method, can be found in Zeineh et al.14