Multi-voxel pattern analysis in human hippocampal subfields

Abstract

A complete understanding of the hippocampus depends on elucidating the representations and computations that exist in its anatomically distinct subfields. High-resolution structural and functional MRI scanning is starting to permit insights into hippocampal subfields in humans. In parallel, such scanning has facilitated the use of multi-voxel pattern analysis (MVPA) to examine information present in the distributed pattern of activity across voxels. The aim of this study was to combine these two relatively new innovations and deploy MVPA in the hippocampal subfields. Delineating subregions of the human hippocampus, a prerequisite for our study, remains a significant challenge, with extant methods often only examining part of the hippocampus, or being unable to differentiate CA3 and dentate gyrus (DG). We therefore devised a new high-resolution anatomical scanning and subfield segmentation protocol that allowed us to overcome these issues, and separately identify CA1, CA3, DG, and subiculum (SUB) across the whole hippocampus using a standard 3T MRI scanner. We then used MVPA to examine fMRI data associated with a decision-making paradigm involving highly similar scenes that had relevance for the computations that occur in hippocampal subfields. Intra- and inter-rater scores for subfield identification using our procedure confirmed its reliability. Moreover, we found that decoding of information within hippocampal subfields was possible using MVPA, with findings that included differential effects for CA3 and DG. We suggest that MVPA in human hippocampal subfields may open up new opportunities to examine how different types of information are represented and processed at this fundamental level.

Keywords: CA3; DG; MVPA; fMRI; hippocampus; memory; scenes; subfields.

Figure 1

The stimuli and experimental task. (A) The two original scenes—note they were not labeled A and B in the actual experiment. (B) The morph continuum proceeding from 100% scene A to 100% scene B. (C) A timeline of an example single trial with stimulus duration of 2.5 s during which the participant registered their decision. Participants then indicated their confidence in that decision during the next 3 s from a choice of “not sure,” “fairly sure,” and “very sure”. There was a 2 s rest period before the next trial.

Figure 2

Subfield segmentation in the coronal plane. Coronal sections through an averaged T2-weighted image of both the left and right hippocampus of a participant.

Figure 3

Subfield segmentation in the sagittal plane. (A) Sagittal section through an averaged T2-weighted image of the hippocampus of a participant. Blue indicates CA1, red indicates CA3, the DG in green becoming visible. (B) Proceeding medially through the hippocampus. (C) The body starts to thin and subiculum (yellow) replaces part of CA1. (D) Only head and tail remain at the medial end of the hippocampus, with now most of the subiculum in view.

Figure 4

Summary of the subfield segmentation procedure. (A) An example coronal slice before segmentation. (B) The same coronal slice with the subfield boundaries demarcated. (C) The final subfield segmentation of this coronal slice. (D) An example sagittal slice before segmentation. (E) The same sagittal slice with the subfield boundaries demarcated. (F) The final subfield segmentation of this sagittal slice.

Figure 5

An example of subfield segmentation in 3D. (A) The left and right hippocampus of a participant shown in 3D with the subfields conjoined. (B) The same left hippocampus is shown with its subfields separated.

Figure 6

Behavioural data. Means ± 1 SE are shown. (A) The psychometric function for accuracy for the 16 participants showed a sigmoid profile. (B) Participants were less accurate and slower with increasing noise in the sensory input. (C) Participants' pattern of confidence ratings also followed the expected distribution. Morphs approaching the two original scenes were afforded higher confidence ratings, and more ambiguous morphs lower ratings.

Figure 7

Classifier performance in hippocampal subfields. Chance was 50%. Means ± 1 SE are shown. All subfield classifier performances were significantly above chance. The y axis shows the percentage accuracy of the classifiers (A) under perceptual certainty, when discriminating between the 100% A and 100% B scenes, and (B) under perceptual ambiguity, when making decisions about the 50% morphed scenes. Under perceptual ambiguity classifiers operating on voxels in both CA1 and CA3 showed significantly better performance (^*p < 0.05) than those operating on voxels in DG and subiculum.