Discriminative Analysis of Mild Alzheimer’s Disease and Normal Aging Using Volume of Hippocampal Subfields and Hippocampal Mean Diffusivity

Abstract

Background:

Studies discovered that the hippocampal subfields are differentially affected by pathological damage, and magnetic resonance (MR) diffusion tensor imaging parameters might be more sensitive measures of early degeneration in Alzheimer’s disease (AD) than conventional MR imaging techniques. The purpose of this study was to evaluate the significance of the volume of hippocampal subfields and the mean diffusivity (MD) value of hippocampus in discrimination between mild AD and normal aging.

Methods:

A total of 29 patients with mild AD and 30 normal aging were scanned. Binary logistic regression analysis was applied to assess the diagnostic significance of the volumes of hippocampal subfields and the MD value of hippocampus.

Results:

All hippocampal subfields except right tail atrophied significantly in the mild AD group (P < .05). The relative volumes of right CA1 and subiculum subfields entered the binary logistic regression model. The accuracy was 91.8%, which was improved to 93.9% as the MD value of right hippocampus entered the model.

Conclusion:

Atrophy was present in almost all hippocampal subfields at mild AD stage. The volumes of CA1 and subiculum were of the most diagnostic significance in discrimination of mild AD, which can be improved by the combination of volume and diffusivity analysis.

Keywords

Alzheimer’s disease hippocampus subfield volume DTI mean diffusivity

It is well known that the neuropathological changes in Alzheimer’s disease (AD), such as amyloid deposition and neurofibrillary tangles, begin in the medial temporal region encompassing the hippocampus.^1,2 As a major neural structure, the hippocampus is involved in episodic memory,³ which is the earliest and most severely impaired cognitive function in AD.⁴ Jack et al have found that the degree of hippocampal atrophy was associated with subsequent conversion from mild cognitive impairment (MCI) to AD and to progress from healthy aging to MCI and then to AD, where it correlates with the Braak and Braak stage.⁵Therefore, the hippocampus has been a focus of many structural and functional neuroimaging studies for the early detection of AD processes. Nowadays with the new developments in neuroimaging data analyses, it is possible for researchers to get a more detailed insight into the atrophy of different hippocampal subfields. Since the hippocampal subfields were found to be differentially affected by neurofibrillary tangles and neuronal loss,^6,7 it is assumed that in this study there may be better biomarkers in hippocampal subfields, which can improve the early detection of AD, than hippocampus as a unitary structure.

Diffusion tensor (DT) imaging (DTI), as a magnetic resonance (MR) imaging (MRI) technique, allows us to measure the 3-dimensional (3D) Brownian motion of water molecules in the living human brain. The mean diffusivity (MD) value derived from DTI data can quantify the variations in water diffusivity caused by microscopic structural changes. It is reported that increased MD may be associated with changes in water content, disruption and partial breakdown of tissue cytoarchitecture,⁸ sclerosis,⁹ or demyelinating processes.¹⁰ Such processes are thought to occur before neuronal degradation, and atrophy is detectable at a macroscopic level. According to many DTI studies, the random, undirected motion of water molecules is significantly elevated in the hippocampal regions of patients with mild cognitive impairment (MCI) compared to age-matched controls.^11–13 The DTI parameters might provide more sensitive and quantifiable measures of early degeneration in AD than conventional MRI techniques,¹⁰ indicating that this imaging modality has a potential for the early detection of AD. Some studies have combined multiple imaging modalities, such as morphometry, DTI, and positron emission tomography, in the diagnosis of AD.^14,15 However, no study in this field has yet investigated the combination of MRI morphometry of hippocampal subfields and DTI MD on hippocampus.

The major objective of this study was to discriminate between patients with mild AD and normal aging using the volume of hippocampal subfields and hippocampal mean diffusivity. The hypothesis is that the combination of volume of hippocampal subfields and MD value of hippocampus can improve the accuracy of discriminative analysis. Hippocampus and hippocampal subfields were automatically segmented using FreeSurfer (http://surfer.nmr.mgh.harvard.edu/).

Methods

Patients

A total of 29 patients with mild AD and 30 age-, gender-, and education-matched healthy controls were included in this MRI study (Table 1).These patients and healthy controls were recruited from the outpatient department of neurology in Lihuili Hospital, Ningbo. Inclusion criteria for the patients were age ≥ 50 years, diagnosis of AD within the past 12 months, Mini-Mental State Examination (MMSE) score ≥17, and clinical dementia rating (CDR) = 1.0. All patients met the criteria for dementia according to the Diagnostic and Statistical Manual of Mental Disorders (fourth edition)1 and the criteria for probable Alzheimer’s disease according to the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association. Normal controls were selected from a pool of volunteers. The following exclusion criteria were applied to all patients: any present serious medical, psychiatric, or neurological disorder that could affect mental function; evidence of focal brain lesions on MRI including lacunae; the presence of severe behavioral or communication problems that would make a clinical or MRI examination difficult; ambidexterity or left-handedness; and the absence of a reliable informant.

Table 1.

Demographic Data of Patients.^a

Group	N	Age in Y	Gender		Edu in Y	MMSE mean (SD)
Group	N	Age in Y	M	F	Edu in Y	MMSE mean (SD)
Mild AD	29	73.1 ± 7.4	15	14	13.8 ± 2.6	22.1 ± 4.0
Controls	30	71.6 ± 9.2	16	14	13.10 ± 2.8	29.5 ± 0.8
t value		0.216	2.68		0.73	2.77
P value		>.05	>.05		>.05	<.05

Abbreviations: MMSE, mini-mental state examination score; M, male; F, female; SD, standard deviation; AD, Alzheimer’s disease; age in Y, age in years; edu in Y, educational level in years including primary school.

^a Data are the mean ± SD. The early-stage AD and control groups differ significantly in the MMSE scores.

The study was approved by the Institutional Review Boards of Ningbo university. Informed consent was obtained from all study patients and their relatives.

The MRI Data Acquisition

Data were obtained using a 1.5-T clinical MR image scanners (Signa EXCITE, GE Healthcare, Milwaukee, Wisconsin). Conventional axial T2-weighted images were previously obtained to rule out cerebral infarction or other lesions. Three-dimensional, T1-weighted spoiled gradient-recalled images covering the whole brain were obtained for volumetric tracing and anatomical localization (repetition time [TR] = 11.3 ms, echo time [TE] = 4.2 ms, inversion time = 400 ms, field of view [FOV] = 24 × 24 mm, matrix = 256 × 192, slice thickness = 1.2 mm, number of excitations [NEX] = 1). After that a single-shot echo planar imaging sequence in alignment with the anterior–posterior commissure plane was used to acquire DTI images. The MR images with 25 noncollinear diffusion gradients and without diffusion gradients were acquired (TR = 10 000 ms, TE = 77.1 ms, B factor = 1000 s/mm², matrix = 128 × 128, slice thickness/gap = 4/0 mm, FOV = 240 mm, slice number = 38).

Image Processing

Magnetic Resonance Imaging Volume Data

The FreeSurfer¹⁶ was applied to separately estimate the left and right volumes of hippocampal fissure and 7 hippocampal subfields: CA1, CA2/3, CA4/DG, fimbria, presubiculum, subiculum, tail (Figure 1). The FreeSurfer uses a Bayesian modeling approach that first builds an explicit computational model of how an MR image around the hippocampal area is generated and subsequently uses this model to obtain fully automated segmentations of hippocampal subfields.

Figure 1.

Example of masks of hippocampal subfields segmented using FreeSurfer.

The FreeSurfer implements a technique for automatically assigning a neuroanatomical label to each location on a cortical surface model, based on the probabilistic information estimated from a manually labeled training set. This procedure incorporates both geometric information derived from the cortical model and neuroanatomical convention. After cortical parcellation, the masks for every labeled cortical structures, including bilateral hippocampi, can be obtained (Figure 2). All masks of hippocampal subfields and hippocampus were visually checked for errors in segmentation.

Figure 2.

Example of masks of bilateral hippocampi in structural (in red, upper row) and diffusion space (in light blue, lower row).

Brain tissue volume was estimated using SIENAX,¹⁷ part of the FMRIB’s Software Library (FSL).¹⁸ The SIENAX starts by extracting brain and skull images from single whole-head input data. Next, tissue-type segmentation with partial volume estimation is carried out in order to calculate the total volume of brain tissue (including separate estimates of the volumes of gray matter and white matter).¹⁹ For this study, we used the absolute volumes generated by the algorithm. The intracranial volume (ICV) was calculated by adding the volumes of cerebral spinal fluid, total gray matter, and total white matter together.

Diffusion Tensor Imaging Data

First, raw DT images were processed with the FMRIB's Diffusion Toolbox program, part of the FSL. Image distortions, induced by eddy currents and head motion, in the DTI data, were corrected by applying a full affine alignment of each image to the mean no-diffusion-weighted image. Subsequently, MD images were created. Since MRI volume data and DTI data were not in the same space, hippocampal masks obtained in MRI volume data processing procedures have to be transformed to diffusion space using transformation matrices derived by coregistering volume data to diffusion space using 6 degrees of freedom. All hippocampal masks in diffusion space were visually checked for errors in registration. Examples of hippocampal mask in diffusion space of a patient diagnosed with early-stage AD are presented in Figure 2. Finally, the MD values of hippocampal masks were calculated, with the upper threshold set as 0.002 to eliminate contamination by cerebral spinal fluid.

Statistical Analysis

The data were analyzed using the SPSS 19 package for Windows (SPSS Inc, Chicago, Illinois). The significance level was set at P < .05. The patients with AD were compared for age, educational level, and MMSE with healthy controls by 2-sample t test and compared for gender by a chi-quadrate test. Group differences in relative volume (gray matter volume/ICV × 100) of hippocampal subfields and hippocampal MD values were assessed using 2-sample t tests and 2-sample Mann-Whitney U tests, respectively. Binary logistic regression analysis was applied to build 3 models that can be used to differentiate patients with mild AD from healthy controls. The first model used the relative volumes of bilateral hippocampi as covariates. The second model used the relative volumes of bilateral hippocampal subfields as covariates. The third model used the relative volume of bilateral hippocampal subfields and MD values of bilateral hippocampi as covariates. As age is known as the strongest risk factor for developing AD,²⁰ age was also included as a covariate. A conditional likelihood ratio test was performed to determine the variables kept in the final model. The sensitivity and specificity of the final 3 models were analyzed using a relative operating characteristic (ROC) curve. The points with the maximal Yoden’s index value (sensitivity + specificity − 1) were set as cutoff points.

Results

Group demographic features are presented in Table 1. No group differences were present with regard to age, gender, and years of education. Patients with Mild AD differed significantly from healthy controls in the MMSE score.

The mean relative volume of bilateral hippocampi in patients with AD is smaller than that in healthy controls. The MD values of bilateral hippocampi in patients with AD are elevated significantly compared to those in healthy controls (Table 2). Compared to controls, the hippocampal subfields in patients with mild AD that decreased significantly in relative volume include CA1, CA2/3, CA4/DG, fimbria, presubiculum, subiculum, and left tail (Table 3).

Table 2.

Comparison of Relative Volume and MD Value of Hippocampus in Patients With Mild AD and in Healthy Controls.^a

	Mild AD	Controls	t/Z value	P value
Relative volume (10⁻²)
L HIP	1.30 ± 0.21	1.48 ± 0.16	−3.186	.003
R HIP	1.42 ± 0.26	1.55 ± 0.17	−1.903	.063
MD value (10⁻⁴ mm²/s)
L HIP	11.18 ± 1.11	10.61 ± 0.13	−3.14^a	<.00
R HIP	11.11 ± 1.04	10.48 ± 0.68	−3.82^a	<.00

Abbreviations: L, left; R, right; HIP, hippocampus; AD, Alzheimer’s disease; MD, mean diffusivity.

^a Z value.

Table 3.

Comparison of Relative Volume of Hippocampal Subfields in Patients With Mild AD and Healthy Controls.

	Relative volume (10⁻³)		t	P value
	Mild AD	Controls	t	P value
L CA1	1.50	1.67	−2.63	.01
R CA1	1.57	1.83	−3.59	.001
L CA2/3	3.78	4.71	−5.12	<.00
R CA2/3	4.20	5.18	−5.09	<.00
L CA4/DG	2.12	2.70	−5.35	<.00
R CA4/DG	2.34	2.97	−5.50	<.00
L subiculum	2.38	3.23	−6.14	<.00
R subiculum	2.50	3.44	−6.55	<.00
L presubiculum	1.74	2.33	−5.71	<.00
R presubiculum	1.81	2.45	−5.59	<.00
L fimbria	0.21	0.33	−3.60	.001
R fimbria	0.19	0.25	−2.66	.01
L tail	1.47	1.95	−4.99	<.00
R tail	2.56	2.06	−5.18	<.00
L fissure	0.23	0.22	0.28	.78
R fissure	0.25	0.26	−0.35	.73

Abbreviations: L, left; R, right; AD, Alzheimer’s disease.

In the binary logistic regression analysis, the covariates kept by a conditional likelihood ratio test in the first model are age and the relative volume of the left hippocampus. The covariates that entered the second model are the relative volume of the right CA1 and subiculum. The covariates that entered the third model are the MD value of the right hippocampus, and the relative volume of the right CA1 and subiculum. The accuracy of the 3 models was 85.7%, 91.8%, and 93.9%, respectively. The area under the ROC curve of the 3 models was 0.883, 0.933, and 0.950, respectively (Figure 3). With respect to the cutoff point, the sensitivity of the 3 models was 82.6%, 91.3%, and 99.1%, respectively. The specificity of the 3 models was 84.6%, 88.46% and 93.3%, respectively.

Figure 3.

The relative operating characteristic (ROC) curve of the binary logistic regression. The blue curve represents the first model in which age and the relative volume of bilateral hippocampi were set as covariates. The green curve represents the second model with the relative volume of hippocampal subfields as covariates. The red curve represents the third model with the relative volume of hippocampal subfields and the mean diffusivity values of bilateral hippocampi as covariates.

Discussion

Due to their intrinsic technical limitations, most of the MRI studies on AD regard the hippocampus as a unitary structure, although this is hardly the case. The hippocampus, including subfields CA1–CA4, dentate gyrus, fimbria, subiculum and presubiculum, is a highly sophisticated structure. Stimuli coming from the entorhinal cortex are processed by the dentate gyrus, subfields CA4, and CA3, before being projected outside the medial temporal lobe via CA1 or subicular efferent projections. Pathological study found that the hippocampus is affected early in AD by neurofibrillary tangle (NFT) deposition, spreading from the entorhinal cortex to the CA1 subfield and the subicular region, then to the CA2/3 subfields, the CA4 subfield, and finally to the neocortex.²¹ Apostolova et al used 3D parametric hippocampal surface mesh models, which revealed that CA1 and subicular atrophy is present in cognitively normal individuals predestined to decline to amnestic MCI (aMCI), and atrophy of CA2 and CA3 subfields in patients with aMCI suggests future conversion to AD.²² The MRI studies have found that the CA1 subfield is the highest change in patients with CDR = 0.5^23,24 and the only significant predictor of conversion to CDR = 0.5 in patients with CDR = 0.²⁵ Atrophy of the CA1 subfield appears specifically related to memory-encoding impairment, which is the predominant symptom in aMCI.²⁶ However, few studies have focused on the impaired cognitive function related to the subiculum subfield.

Adachi et al manually measured the width of the subiculum, CA1, and CA3 to CA4 on multishot diffusion- and T2-weighted MR images.²⁷ They found significant differences in the width of the subiculum and CA1 between the control and the mild or moderate AD groups but not in the width of the CA3 to CA4 between the control and the mild AD groups. In a radial mapping MR-based study, Frisoni et al discovered the atrophy of these 2 subfields.²⁸ Our study measured the volume of hippocampal subfields of patients with mild AD using an automated segmentation technique and also discovered significant atrophy of CA1 and subiculum subfields. Moreover, the volumes of the 2 subfields were the only covariates which entered the final model of binary logistic analysis with hippocampal subfields as covariates. This result is consistent with the pathological process of NFT in AD that CA1 and subiculum are the first and most severely affected subregion in hippocampus²¹ and indicates that the 2 subfields are of the most diagnositic significance in discrimination of mild AD and normal aging. However, some studies found no significant atrophy of subiculum in mild AD.^29,30 Chételat et al used voxel-based morphometry with 3D surface mapping of hippocampus and discovered that increasing age was mainly associated with subicular atrophy in the healthy population.³⁰ They pointed out that this result does not mean that the subiculum is unchanged but instead that it is not significantly more atrophied in AD than in the normal elderly individuals. In addition to CA1 and subiculum, Apostolova et al found atrophy of CA2 and CA3 subfields in patients with AD.³¹ The results in this study revealed that CA4, dentate gyrus, fimbria, presubiculum, and left tail also atrophied in mild AD stage. The hippocampal fissure is a sulcus that separates the dentate gyrus from the subiculum and the CA1 fields, which was found enlarged in probable AD.³² In this study, there was no significant contraction or enlargement in bilateral hippocampal fissure in the mild AD group.

In the logistic regression analysis with the relative volume of hippocampal subfields as covariates, there were only 2 covariates determined by a conditional likelihood ratio test in the final model: right CA1 and subiculum. However, the accuracy, sensitivity, and specificity of this model are higher than those of the final model with age and the relative volume of the bilateral hippocampi as covariates, which indicates that the hippocampal subfield volumes have more discriminative power than the total hippocampal volumes in discrimination of mild AD from normal aging. The accuracy of diagnosis was further promoted to 93.9% after the MD value of the right hippocampus was determined into the model. These results are consistent with our hypothesis that the combination of segmentation of hippocampal subfields and DTI parameter can improve the accuracy of discriminative analysis. Some studies also used multiple imaging techniques to diagnose AD. Dai et al partitioned whole brain into 90 regions and used multimodal imaging to discriminate patients with AD from healthy controls, which led to an accuracy of 89.47%.³³ Jhooa et al conducted a logistic regression analysis on a combination of hippocampal volume, parahippocampal cingulum fractional anisotropy, and hippocampal glucose metabolism to discriminate between probable patients with AD and normal controls.³⁴ That model obtained an accuracy of 94.1%. Due to the severe partial volume effect caused by 4-mm DTI slice thickness, we did not measure the MD values of the hippocampal subfield which is assumed to be able to evidently improve the discrimination of AD and normal aging and needs further investigation.

There is evidence indicating that the pathological factors may lead to asymmetry in brain.^35,36 Many MRI studies, including a meta analysis on 700 patients with AD and 365 patients with MCI in 2009,³⁷ suggested leftward laterality of hippocampal atrophy. Nevertheless, there exists disputes about laterality of the hippocampal volume in AD. In 2011, Derflinger et al,³⁸ who used voxel-based morphometry analysis on MRI data, found no laterality of hippocampal atrophy. They argued that brain atrophy in AD was asymmetric rather than lateralized, which was consistent with some histopathological studies.^39–41 Additionally, some studies discovered leftward laterality of hippocampal atrophy in MCI but not in AD.^42,43 The logistic regression analysis in this study suggested that the volume of the right CA1 and subiculum subfields, as well as the MD value of the right hippocampus, might be more responsible for group discrimination than the left. This finding implies that the right hippocampus might be preferentially involved in the development of mild AD. However, it is just conceivable that this rightward laterality may result from a selection bias, and a larger sample size is necessary in future study.

This study applied FreeSurfer to make automated and sophisticated segmentation of hippocampal subfields with no research bias compared to manual segmentation, and then measured the volume of each subfields. Voxel based analysis and shape analysis are advantageous in morphometric studies between groups of patients. However, the 2 methods are not suitable for analysis of the pattern of atrophy change in an individual patient, while the semiautomatic analysis in this study is more patient specific and can be used as a marker for disease diagnosis on an individual level.

One limitation of this study was that the relatively small sample size. A second limitation was the low vertical resolution of the MD map obtained from DTI due to the 4-mm slice thickness, which may cause partial volume effect and impede the effort to segment hippocampal subfields on MD map. A third limitation was that the patients were clinically diagnosed as probable AD without postmortem neuropathologic evaluation. Furthermore, studies have found that the number of apolipoprotein E (ApoE) ∊4 allele is significantly correlated with the hippocampal volume as well as the rate of hippocampal atrophy in probable patients with AD.^44,45 The relationship between ApoE ∊4 allele and the hippocampal subfields volume was not investigated in this study and may be a direction for future research.

In conclusion, the present study demonstrates that almost all hippocampal subfields are affected in mild AD stage. The volumes of CA1 and subiculum subfields are of higher diagnostic significance in discrimination of mild AD and normal aging than the total volume of hippocampus, which can be improved by the combination of volume and diffusivity analysis. The future study should apply more advanced imaging techniques, such as isotropic DTI, on larger a sample size.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This study was supported by the Project for Social Development of Ningbo municipality (2011C50012). The financial sponsor played no role in the design, execution, analysis and interpretation of data or writing of the study.

References

Braak

. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol. 1991;82(4):239–259.

Ball

. Neuronal loss, neurofibrillary tangles and granulovacuolar degeneration in the hippocampus with ageing and dementia. A quantitative study. Acta Neuropathol. 1997;37(2):111–118.

Deweer

Pillon

Pochon

Dubois

. Is the HM story only a “remote memory”? Some facts about hippocampus and memory in humans. Behav Brain Res. 2001;127(1-2):209–224.

Soininen

Scheltens

. Early diagnostic indices for the prevention of Alzheimer's disease. Ann Med. 1998;30(6):553–559.

Jack

Dickson

Parisi

Antemortem MRI findings correlate with hippocampal neuropathology in typical aging and dementia. Neurology. 2002;58(5):750–757.

Bobinski

de Leon

Tarnawski

Neuronal and volume loss in CA1 of the hippocampal formation uniquely predicts duration and severity of Alzheimer disease. Brain Res. 1998;805(1-2):267–269.

West

Kawas

Stewart

Rudow

Troncoso

. Hippocampal neurons in pre-clinical Alzheimer’s disease. Neurobiol Aging. 2004;25(9):1205–1212.

Beaulieu

. The basis of anisotropic water diffusion in the nervous system—A technical review. NMR Biomed. 2002;15(7-8):435–455.

Assaf

Mohamed

Abou-Khaled

Diffusion tensor imaging of the hippocampal formation in temporal lobe epilepsy. Am J Neuroradiol. 2003;24(9):1857–1862.

10.

Sun

Song

Harms

Detection of age-dependent brain injury in a mouse model of brain amyloidosis associated with Alzheimer’s disease using magnetic resonance diffusion tensor imaging. Exp Neurol. 2005;191(1):77–85.

11.

Kantarci

Jack

Jr Xu

Mild cognitive impairment and Alzheimer disease: regional diffusivity of water. Radiology. 2001;219(1):101–107.

12.

Fellgiebel

Wille

Müller

Ultrastructural hippocampal and white matter alterations in mild cognitive impairment: a diffusion tensor imaging study. Dement Geriatr Cogn Disord. 2004;18(1):101–108.

13.

Zhang

Zhao

Evaluation of apparent diffusion coefficient mappings in amnestic mild cognitive impairment using an image analysis software brain search. Acta Radiol. 2011;52(10):1147–1154.

14.

Walhovdn

Fjell

Amlien

Multimodal imaging in mild cognitive impairment: metabolism, morphometry and diffusion of the temporal-parietal memory network. Neuroimage. 2009;45(45):215–223.

15.

Wang

Goldstein

Veledar

Alterations in cortical thickness and white matter integrity in mild cognitive impairment measured by whole-brain cortical thickness mapping and diffusion tensor imaging. Am J Neuroradiol. 2009;30(5):893–899.

16.

Van Leemput

Bakkour

Benner

Automated segmentation of hippocampal subfields from ultra-high resolution in vivo MRI. Hippocampus. 2009;19(6):549–557.

17.

Smith

Zhang

Jenkinson

Accurate, robust and automated longitudinal and cross-sectional brain change analysis. Neuroimage. 2002;17(1):479–489.

18.

Smith

Jenkinson

Woolrich

Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage. 2004;23(suppl 1):208–219.

19.

Zhang

Brady

Smith

. Segmentation of brain MR images through a hidden Markov random field model and the expectation maximization algorithm. IEEE Trans Med Imaging. 2001;20(1):45–57.

20.

Petrella

Coleman

Doraiswamy

. Neuroimaging and early diagnosis of Alzheimer disease: a look to the future. Radiology. 2003;226(2):315–336.

21.

Schonheit

Zarski

Ohm

. Spatial and temporal relationships between plaques and tangles in Alzheimer-pathology. Neurobiol Aging. 2004;25(6):697–711.

22.

Apostolova

Mosconi

Thompson

Subregional hippocampal atrophy predicts Alzheimer's dementia in the cognitively normal. Neurobiol Aging. 2010;31(7):1077–1088.

23.

Csernansky

Wang

Joshi

EarlyDATis distinguished from aging by high-dimensional mapping of the hippocampus. Dementia of the Alzheimer type.

Neurology. 2000;55(11):1636–1643.

24.

Wang

Miller

Gado

Abnormalities of hippocampal surface structure in very mild dementia of the Alzheimer type. Neuroimage. 2006;30(1):52–60.

25.

Csernansky

Wang

Swank

Preclinical detection of Alzheimer’s disease: hippocampal shape and volume predict dementia onset in the elderly. Neuroimage. 2005;25(3):783–792.

26.

Fouquet

Desgranges

La Joie

Role of hippocampal CA1 atrophy in memory encoding deficits in amnestic Mild Cognitive Impairment.

Neuroimage. 2012;15;59(4):3309–3315.

27.

Adachi

Kawakatsu

Hosoya

Morphology of the inner structure of the hippocampal formation in Alzheimer disease. Am J Neuroradiol. 2003;24(8):1575–1581.

28.

Frisoni

Sabattoli

Lee

Dutton

Toga

Thompson

. In vivo neuropathology of the hippocampal formation in AD: a radial mapping MR-based study. Neuroimage. 2006;32(1):104–110.

29.

Frisoni

Ganzola

Canu

Mapping local hippocampal changes in Alzheimer's disease and normal ageing with MRI at 3 Tesla. Brain. 2008;131(pt 12):3266–3276.

30.

Chételat

Fouquet

Kalpouzos

Three-dimensional surface mapping of hippocampal atrophy progression from MCI to AD and over normal aging as assessed using voxel-based morphometry. Neuropsychologia. 2008;46(6):1721–1731.

31.

Apostolova

Dinov

, Dutton RA, et al. 3D comparison of hippocampal atrophy in amnestic mild cognitive impairment and Alzheimer’s disease. Brain. 2006;129(pt 11):2867–2873.

32.

Bastos-Leite

van Waesberghe

Oen

van der Flier

Scheltens

Barkhof

. Hippocampal sulcus width and cavities: comparison between patients with Alzheimer disease and nondemented elderly subjects. Am J Neuroradiol. 2006;27(10):2141–2145.

33.

Dai

Yan

Wang

Discriminative analysis of early Alzheimer's disease using multi-modal imaging and multi-level characterization with multi-classifier (M3). Neuroimage. 2012;59(3):2187–2195.

34.

Jhooa

Lee

Choo

Discrimination of normal aging, MCI and AD with multimodal imaging measures on the medial temporal lobe.

Psychiatry Res. 2010;183(3):237–243.

35.

Nardi

Bingman

. Asymmetrical participation of the left and right hippocampus for representing environmental geometry in homing pigeons. Behav Brain Res. 2007;178(1):160–171.

36.

Toga

Thompson

. Mapping brain asymmetry. Nat Rev Neurosci. 2003;4(1):37–48.

37.

Shi

Liu

Zhou

Hippocampal volume and asymmetry in mild cognitive impairment and Alzheimer's disease: meta-analyses of MRI studies. Hippocampus. 2009;19(11):1055–1064.

38.

Derflinger

Sorg

Gaser

Grey-matter atrophy in Alzheimer's disease is asymmetric but not lateralized. J Alzheimers Dis. 2011;25(2):347–357.

39.

Janota

Mountjoy

. Asymmetry of pathology in Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 1988;51(7):1011–1012.

40.

Wilcock

Esiri

. Asymmetry of pathology in Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 1987;50(10):1384–1386.

41.

Moossy

Zubenko

Martinez

Rao

Kopp

Hanin

. Lateralization of brain morphologic and cholinergic abnormalities in Alzheimer’s disease. Arch Neurol. 1989;46(6):639–642.

42.

Jessen

Feyen

Freymann

Volume reduction of the entorhinal cortex in subjective memory impairment. Neurobiol Aging. 2006;27(12):1751–1756.

43.

Ridha

Barnes

Bartlett

Tracking atrophy progression in familial Alzheimer’s disease: a serial MRI study. Lancet Neurol. 2006;5(10):828–834.

44.

Hashimoto

Yasuda

Tanimukai

Apolipoprotein E epsilon 4 and the pattern of regional brain atrophy in Alzheimer's disease. Neurology. 2001;57(8):1461–1466.

45.

Mori

Lee

Yasuda

Accelerated hippocampal atrophy in Alzheimer's disease with apolipoprotein E epsilon4 allele. Ann Neurol. 2002;51(2):209–214.