Delineation of hippocampal subregions using T1-weighted magnetic resonance images at 3 Tesla

Brain Struct Funct. 2015 Nov;220(6):3259-72. doi: 10.1007/s00429-014-0854-1. Epub 2014 Aug 1.

Abstract

Although several novel approaches for hippocampal subregion delineation have been developed, they need to be applied prospectively and may be limited by long scan times, the use of high field (>3T) imaging systems, and limited reliability metrics. Moreover, the majority of MR imaging data collected to date has employed a T1-weighted acquisition, creating a critical need for an approach that provides reliable hippocampal subregion segmentation using such a contrast. We present a highly reliable approach for the identification of six subregions comprising the hippocampal formation from MR images including the subiculum, dentate gyrus/cornu Ammonis 4 (DG/CA4), entorhinal cortex, fimbria, and anterior and posterior segments of cornu Ammonis 1-3 (CA1-3). MR images were obtained in the coronal plane using a standard 3D spoiled gradient sequence acquired on a GE 3T scanner through the whole head in approximately 10 min. The average ICC for inter-rater reliability across right and left volumetric regions-of-interest was 0.85 (range 0.71-0.98, median 0.86) and the average ICC for intra-rater reliability was 0.92 (range 0.66-0.99, median 0.97). The mean Dice index for inter-rater reliability across right and left hemisphere subregions was 0.75 (range 0.70-0.81, median 0.75) and the mean Dice index for intra-rater reliability was 0.85 (range 0.82-0.90, median 0.85). An investigation of hippocampal asymmetry revealed significantly greater right compared to left hemisphere volumes in the anterior segment of CA1-3 and in the subiculum.

Keywords: 3 Tesla; Delineation criteria; Hippocampus; Magnetic resonance imaging; Subregion.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Diffusion Magnetic Resonance Imaging / methods
  • Female
  • Hippocampus / anatomy & histology*
  • Hippocampus / pathology
  • Hippocampus / physiology*
  • Humans
  • Image Processing, Computer-Assisted / methods
  • Male
  • Reproducibility of Results