Hippocampal subfield diffusivity changes in healthy ageing
Daniel J Cox1,2, Hamied A Haroon2, Daniela Montaldi1, and Laura M Parkes2

1School of Psychological Sciences, University of Manchester, Manchester, United Kingdom, 2Centre for Imaging Sciences, University of Manchester, Manchester, United Kingdom


Alterations to hippocampal microstructure may precede gross volumetric changes in ageing, and these changes may occur preferentially in different hippocampal subfields. We investigated both established (FA and mADC) and novel (DOC) measurements of diffusion in these regions, in addition to volume, in order to determine where age-related changes occurred. The results showed changes across the majority of subfields for mADC and FA, but only in left CA 2/3 for DOC measures 1, 3 and >3. We suggest this could be related to differential degradation of particular cellular structures in these regions.


As normal healthy aging occurs, research has shown that the hippocampus undergoes both macro and microstructural changes1,2. Whilst many studies investigate gross volumetric or microstructural changes across the hippocampus as a whole, it is likely that there is a degree of separation between subfields in terms of microstructure3. We investigated age-related changes in established (fractional anisotropy (FA) and mean apparent diffusion coefficient (mADC)) and novel (diffusion orientation complexity (DOC)) measures of diffusion, in addition to volume, across the cornu ammonis (CA) subfields. We hypothesised that measures of hippocampal microstructure will change with age, and that there will be a differentiation between subfields in this change which may not be apparent from volumetric analysis.


34 healthy volunteers participated in the study, which was approved by a local NHS research ethics committee. A 3T Philips Achieva system with an 8-element head coil was used to collect data. Both structural T1 weighted (0.94x0.94x1mm) and High Angle Resolution Diffusion Imaging (HARDI) data (TE=59ms, matrix 128x128, slice thickness 2.1mm, 60 contiguous slices, in-plane resolution 1.875x1.875mm, 43 non-collinear diffusion sensitization directions, b=1200s/mm2, 1 at b=0, cardiac gating) were collected. Diffusion data was corrected for susceptibility and eddy current induced distortion4. In-house software calculated voxel-wise probability maps (fig 1)5 for n number of fibre orientations (1, 2, 3 or greater than 3, which are referred to as DOC 1, 2, 3 and >3 respectively).

All T1-weighted images were segmented using the cortical parcellation and hippocampal subfield (v5.33)6 segmentation pipelines available in Freesurfer. Hippocampal subfields were transformed into native space using the inverse transform provided by Freesurfer, and thresholded so only voxels having at least 51% probability of belonging to a subfield and 75% probability for consisting of grey matter were included (fig 2). These regions were then registered into diffusion space for each subject. Subfield volumes were also normalised to intracranial volume for each subject. 4 hippocampal segmentations failed or did not register correctly, so the final dataset used for this study totalled 30 subjects (age 18-30 (n15) and 60-82 (n15), mean age 48.1 ± 26.8).

Image analysis was performed in SPM12 using MarsBaR (v0.437) to extract data values for left and right hippocampal CA regions (CA1, CA2-3 and CA 4-dentate gyrus (DG)). Independent t-test analyses were then performed in SPSS 22 on the data to determine in which subfields age-related volumetric or diffusion changes occurred.


All results given are corrected for multiple comparisons8 and are p<0.05. Independent sample t-tests were used to investigate volume change between age groups across all subfields, though no significant results were found after correction. Next, additional t-test analyses were performed to examine age-related differences for FA, mADC and DOC in each subfield. FA and mADC showed significant age-related changes between age groups in all subfields except for left CA1 and right CA1/CA2-3 for FA, with older adults showing decreased FA and increased mADC overall. T-tests also showed significant decreases to DOC1, and increases to DOC3 and DOC>3 in left CA 2-3 only (figs 3 and 4). These results suggest that measures of diffusivity may be more sensitive to age-related physiological changes in hippocampal subfields than gross volume.


The results of this study support our hypotheses and suggest that novel measures of tissue microstructure (such as DOC) may provide additional information to that given by current measures. They may also act as sensitive markers of age-related neurophysiological decline in the hippocampus compared to tissue atrophy, as shown by highlighting microstructural changes in specific hippocampal subfields. This may be because particular cellular structures (for example, pyramidal neurons) degrade to a greater extent in certain subregions during aging. These findings highlight the need for further investigation into this effect, particularly with regards to cognitive measures of recognition memory which have been shown to be correlated with particular subfields.


No acknowledgement found.


Wisse, L. et al (2014) 'Hippocampal subfield volumes at 7T in early Alzheimer's disease and normal aging.', Neurobiology of Aging, vol. 35 no. 9 pp. 2039-45

Cox, D. J. et al (2015) 'Quantification of age-related hippocampal microstructure changes and recollection memory decline', Proceedings of the Organisation for Human brain Mapping, Honolulu, Hawaii poster number 1748

Pereira, J. B. et al (2014) 'Regional vulnerability of hippocampal subfields to aging measured by structural and diffusion MRI' Hippocampus, vol. 24 no. 4 pp. 403 - 414

Embleton, K. E. et al. (2010) ‘Distortion correction for diffusion-weighted MRI Tractography and fMRI in the temporal lobes’, Human Brain Mapping, vol. 31 no.10 pp. 1570-1587

Haroon, H et al. (2010) 'Probabilistic Quantification of Microstructural Complexity in Cortical and Subcortical Regions' Proceedings of the Organisation for Human Brain Mapping, Barcelona, Spain poster number 829 Wh-AM

Van Leemput, K et al (2009) 'Automated Segmentation of Hippocampal Subfields from Ultra-High Resolution In Vivo MRI', Hippocampus, vol. 19, no. 6 pp. 549-557

Brett, M et al. (2002) 'Region of interest analysis using an SPM toolbox', 8th International Conference on Functional Mapping of the Human Brain, Sendai, Japan

Benjamini, Y., and Y. Hochberg. (1995) ‘Controlling the false discovery rate: a practical and powerful approach to multiple testing’, Journal of the Royal Statistical Society B, 57 pp. 289-300


Fig 1 - Sagittal slices of diffusion images in one subject. a: FA, b-e: Probability maps of n fibre orientations (DOC 1, 2, 3 and >3). Colour bar indicates 0-100% probability (DOC), greyscale bar indicates values between 0 and 1 (FA). Upper right image shows location of image slices.

Fig 2 - Sagittal, coronal and axial views of hippocampal CA regions post-thresholding and masking in a single subject. Left/right CA1 - red/purple, left/right CA 2/3 - dark blue/yellow, left/right CA 4/DG - green/light blue

Fig 3 - Bar charts showing mean diffusion measures (a-FA; b-mADC; c-DOC1; d-DOC2; e-DOC3; f-DOC>3) between young and old groups across CA subfields. Significant group differences (p<0.05) are indicated by a star. Error bars - 1 SEM.

Fig 4 - Group means, standard deviations, t and p values for diffusion measures showing significant group differences after correction for multiple comparisons.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)