Elsevier

NeuroImage

Volume 52, Issue 1, 1 August 2010, Pages 186-197
NeuroImage

Joint analysis of structural and perfusion MRI for cognitive assessment and classification of Alzheimer's disease and normal aging

https://doi.org/10.1016/j.neuroimage.2010.04.033Get rights and content

Abstract

Structural magnetic resonance imaging (MRI) of brain tissue loss and physiological imaging of regional cerebral blood flow (rCBF) can provide complimentary information for the characterization of brain disorders, such as Alzheimer's disease (AD) but studies into gains in classification power for AD using these image modalities jointly have been limited. Our aim in this study was to determine the joint contribution of structural and perfusion-weighted imaging for the classification of AD in a cross-sectional study using an integrated multimodality MRI processing framework and a cortical surface-based analysis approach. We used logistic regression analysis to determine sequentially the value of cortical thickness, rCBF, and cortical thickness and rCBF jointly for classification for diagnosis of AD compared to controls. We further tested the extent to which cortical thinning and reduced rCBF explain individually or together variability in dementia severity. Separate analysis of structural MRI and perfusion-weighted MRI data yielded the well-established pattern of cortical thinning and rCBF reduction in AD, affecting predominantly temporo-parietal brain regions. Using structural MRI and perfusion-weighted MRI jointly indicated that cortical thinning dominated the classification of AD and controls without significant contributions from rCBF. However there was also a positive interaction between reduced rCBF and cortical thinning in the right superior temporal sulcus, implying that structural and physiological brain alterations in AD can be complementary. Compared to reduced rCBF, regional cortical thinning better explained the variability in dementia severity. In conclusion, structural brain alterations compared to physiological variations are the dominant features of MRI in AD.

Introduction

Most neurodegenerative disorders, such as Alzheimer's disease (AD) and other types of dementia, are associated with characteristic patterns of regional brain alterations that can be visualized using neuroimaging. Moreover, the patterns of structural, functional, and physiological alterations can be regionally discordant, suggesting that each pattern may provide complementary information (Hayasaka et al., 2006). In AD, for example, structural MRI studies consistently revealed a pattern of brain tissue loss that predominantly involves structures in the medial temporal cortex (i.e., hippocampus and the entorhinal cortex (deToledo-Morrell et al., 2004, Morra et al., 2008, Morra et al., 2009a, Morra et al., 2009b, Schroeter et al., 2009, Stoub et al., 2005, Thompson et al., 2004)), consistent with the known distribution of early AD pathology from histopathological studies (Braak and Braak, 1991). As the severity of AD progresses, structural MRI also shows a gradual expansion of tissue loss into temporo-parietal cortical areas (Chetelat and Baron, 2003, Desikan et al., 2008, Hua et al., 2008a, Whitwell et al., 2007, Whitwell et al., 2008). On the other hand, functional studies in AD using positron emission tomography (PET) for measurements of cerebral glucose consumption or single photon emission computed tomography (SPECT) and more recently arterial spin labeling magnetic resonance imaging (ASL-MRI) for measurements of regional cerebral blood flow (rCBF) generally found the most prominent alterations in the association cortices (Alsop et al., 2008, Callen et al., 2002, Nebu et al., 2001, Rodriguez et al., 2000), spatially separated from the main structural changes. The affected areas include the posterior temporal and parietal association cortices (Bradley et al., 2002, Keilp et al., 1996, Schroeter et al., 2009), as well as in the posterior cingulate, precuneus, and medial temporal cortices (Asllani et al., 2008, Du et al., 2006, Ishii et al., 1996, Johnson et al., 2005, Kobayashi et al., 2008, Warkentin et al., 2004). However, functional alterations can also be seen in mesial temporal lobe structures and the hippocampus (Alsop et al., 2008, Mosconi et al., 2005), in overlap with early structural changes in these regions. The diversity of these patterns is of clinical interest as it may help separating AD from normal aging as well as staging of the disease, since the patterns generally correlate with the progression of clinical symptoms, especially with decline in memory function (Arbizu et al., 1997, Basso et al., 2006, Benoit et al., 1999, Bruen et al., 2008, Gilboa et al., 2005, Jagust et al., 1989, Keilp et al., 1996, Lampl et al., 2003, Leube et al., 2008, Maestu et al., 2003, Mungas et al., 2005, Nobili et al., 2005, Nobili et al., 2007, O'Brien et al., 1992, Reed et al., 1989, Rémy et al., 2005, Rodriguez et al., 1999, Rodriguez et al., 2000, Sabbagh et al., 1997, Schwartz et al., 1991, Wolfe et al., 1995). However, most imaging studies have exploited structural or physiological alterations separately for the classification of AD patients and healthy subjects. Moreover, among studies that used structural and physiological changes together for classification (Jagust, 2006, Kawachi et al., 2006, Matsunari et al., 2007), many ignored potential interactions between structural and physiological changes and often limited the analysis to predetermined regions of interest, potentially under-utilizing information available with imaging.

Our overall goal in this study was to assess in full the value added by using jointly MRI measures of regional cortical thinning and rCBF, including their interaction, for the classification of AD patients and elderly controls. To avoid regional bias, we further aimed to determine the joint classification power of structural and perfusion MRI on a point-by-point basis. In addition to mere group classification, we also aimed to determine the joint value of cortical atrophy and rCBF measures in explaining the variance in the severity of cognitive impairment in AD.

Toward these study goals, we present an integrated multimodality image processing and analysis framework for an effective joint analysis of regional cortical thinning and rCBF variations on a point-by-point basis across the whole brain. Since we are mainly interested in cortical alterations, pertaining cortical thinning and cortical rCBF, we pursued a cortical surface-based analysis approach, which provides better spatial normalization of cortical data across subjects compared to voxel-based approaches (Tosun and Prince, 2008). In addition, the dense analysis of cortical atrophy and rCBF on cortical surface representations benefits a data-driven approach that overcomes the restrictions of region-of-interest-based methods. Similar cortical surface-based approaches were reported recently for analysis of fMRI data (Anticevic et al., 2008, Hagler et al., 2006, Hashikawa et al., 1995). However, this is – to our knowledge – the first investigation aimed to evaluate structural and perfusion alterations in AD together by using an integrated multimodality MR image-processing framework coupled with 3D cortical surface-based data analysis. In the following sections, the technical challenges for an integrated multimodality MR image-processing framework are described, especially in the context of dementia, where extensive brain atrophy requires accurate spatial alignment of the intra-subject inter-modality MR images, including corrections for nonlinear geometric distortions and partial volume effects in the low-resolution perfusion images. We then present a logistic regression analysis to determine sequentially the value of cortical thickness, rCBF, and cortical thickness and rCBF jointly for the classification of AD patients and cognitively normal controls. We further test the extent to which cortical thinning and reduced rCBF explain individually or together severity of cognitive impairment in AD.

Section snippets

Subjects

The study included 38 healthy elderly subjects, aged 51–81 years with Mini-Mental State Examination (MMSE) scores between 26 and 30, and 24 patients diagnosed with Alzheimer's disease, aged 51–85 years with MMSE scores between 8 and 29. All subjects were recruited from the Memory and Aging Center of the University of California, San Francisco and had extensive physical, neurological, and neurocognitive examinations at the center. The MR images were used to rule out other major neuropathologies

Structural MR image processing

The following key processing steps were performed on each brain image volume for estimations of cortical thickness. First, an expectation maximization segmentation (EMS) algorithm including correction for intensity inhomogeneity (Van Leemput et al., 1999a, Van Leemput et al., 1999b) was applied to the T1w image with supplementary T2w image input, to separate skull, scalp, extra-cranial tissue, cerebellum, and brain stem (at the level of the diencephalon) from the rest of brain volume. The

Cortical spatial normalization

An image analysis technique known as cortical spatial normalization was used to match anatomically homologous cortical features across subjects before performing cross-subject comparisons. Specifically, the central cortical surface model of each subject was spatially normalized with respect to the geometry of a representative reference brain using an automated surface-based cortical warping method (Tosun and Prince, 2008). Structural brain MRI scan from a healthy female of age 65 years old was

Cortical thinning

Maps of logistic regression coefficients using cortical thickness alone for the correct classification of CN and AD are shown in Fig. 4a. Widespread cortical thinning in the temporo-parietal, middle frontal, superior frontal, posterior cingulate, anterior cingulate, precuneus, cuneus, and entorhinal cortices bilaterally had the best classification power, as indicated in the corresponding significance map corrected at p = 0.05 in Fig. 4b.

Hypoperfusion

Maps of logistic regression coefficients using rCBFPVE alone

Discussion

The major findings are: (1) separate analyses of structural MRI and cASL-MRI data yielded the well-established pattern of cortical thinning and rCBF reduction in AD, consistent with previous neuroimaging studies, including PET and SPECT. (2) Using measurements from structural MRI and cASL-MRI jointly indicated that cortical thinning dominated the classification of AD and controls without significant diagnostic contributions from rCBF measurements. (3) Considering furthermore the relationship

Acknowledgments

This work was supported by the National Institutes of Health grant P41 RR23953 and the Department of Defense grant W81XWH-05-2-0094. This work has also been made possible by use of research facilities at the Veterans Affairs Medical Center in San Francisco.

References (100)

  • M.F. Folstein et al.

    “Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician

    J. Psychiatr. Res.

    (1975)
  • D.J. Hagler et al.

    Smoothing and cluster thresholding for cortical surface-based group analysis of fMRI data

    NeuroImage

    (2006)
  • X. Han et al.

    CRUISE: cortical reconstruction using implicit surface evolution

    NeuroImage

    (2004)
  • S. Hayasaka et al.

    A non-parametric approach for co-analysis of multi-modal brain imaging data: application to Alzheimer's disease

    NeuroImage

    (2006)
  • X. Hua et al.

    3D characterization of brain atrophy in Alzheimer's disease and mild cognitive impairment using tensor-based morphometry

    NeuroImage

    (2008)
  • X. Hua et al.

    Tensor-based morphometry as a neuroimaging biomarker for Alzheimer's disease: an MRI study of 676 AD, MCI, and normal subjects

    NeuroImage

    (2008)
  • W. Jagust

    Positron emission tomography and magnetic resonance imaging in the diagnosis and prediction of dementia

    Alzheimer's Dement.

    (2006)
  • J.G. Keilp et al.

    Inferior parietal perfusion, lateralization, and neuropsychological dysfunction in Alzheimer's disease

    Brain Cogn.

    (1996)
  • S. Kobayashi et al.

    Quantitative analysis of brain perfusion SPECT in Alzheimer's disease using a fully automated regional cerebral blood flow quantification software, 3DSRT

    J. Neurol. Sci.

    (2008)
  • J.H. Morra et al.

    Validation of a fully automated 3D hippocampal segmentation method using subjects with Alzheimer's disease mild cognitive impairment, and elderly controls

    NeuroImage

    (2008)
  • J.H. Morra et al.

    Automated mapping of hippocampal atrophy in 1-year repeat MRI data from 490 subjects with Alzheimer's disease, mild cognitive impairment, and elderly controls

    NeuroImage

    (2009)
  • F. Nobili et al.

    Resting SPECT-neuropsychology correlation in very mild Alzheimer's disease

    Clin. Neurophysiol.

    (2005)
  • F. Rémy et al.

    Verbal episodic memory impairment in Alzheimer's disease: a combined structural and functional MRI study

    NeuroImage

    (2005)
  • G. Rodriguez et al.

    Hippocampal perfusion in mild Alzheimer's disease

    Psychiatry Res.

    (2000)
  • M.L. Schroeter et al.

    Neural correlates of Alzheimer's disease and mild cognitive impairment: a systematic and quantitative meta-analysis involving 1351 patients

    NeuroImage

    (2009)
  • J.A. Teresi et al.

    Item bias in cognitive screening measures: comparisons of elderly white, Afro-American, Hispanic and high and low education subgroups

    J. Clin. Epidemiol.

    (1995)
  • P.M. Thompson et al.

    Mapping hippocampal and ventricular change in Alzheimer disease

    NeuroImage

    (2004)
  • D. Tosun et al.

    Cortical reconstruction using implicit surface evolution: accuracy and precision analysis

    NeuroImage

    (2006)
  • D. Tosun et al.

    Mapping techniques for aligning sulci across multiple brains

    Med. Image Anal.

    (2004)
  • I. Yakushev et al.

    SPM-based count normalization provides excellent discrimination of mild Alzheimer's disease and amnestic mild cognitive impairment from healthy aging

    NeuroImage

    (2009)
  • J. Arbizu et al.

    Cerebral SPECT and neuropsychological function in Alzheimer's disease

    Rev. Med. Univ. Navarra

    (1997)
  • I. Asllani et al.

    Multivariate and univariate analysis of continuous arterial spin labeling perfusion MRI in Alzheimer's disease

    J. Cereb. Blood Flow Metab.

    (2008)
  • K.A. Bayles

    Alzheimer's disease symptoms: prevalence and order of appearance

    J. Appl. Gerontol.

    (1991)
  • M. Benoit et al.

    Behavioral and psychological symptoms in Alzheimer's disease. Relation between apathy and regional cerebral perfusion

    Dement. Geriatr. Cogn. Disord.

    (1999)
  • L. Bidzan

    Vascular factors in dementia

    Psychiatr. Pol.

    (2005)
  • H. Braak et al.

    Neuropathological stageing of Alzheimer-related changes

    Acta Neuropathol.

    (1991)
  • K.M. Bradley et al.

    Cerebral perfusion SPET correlated with Braak pathological stage in Alzheimer's disease

    Brain

    (2002)
  • P.D. Bruen et al.

    Neuroanatomical correlates of neuropsychiatric symptoms in Alzheimer's disease

    Brain

    (2008)
  • R. Cabeza et al.

    Imaging cognition II: an empirical review of 275 PET and fMRI studies

    J. Cogn. Neurosci.

    (2000)
  • R. Cacabelos et al.

    Cerebrovascular risk factors in Alzheimer's disease: brain hemodynamics and pharmacogenomic implications

    Neurol. Res.

    (2003)
  • V.D. Calhoun et al.

    Method for multimodal analysis of independent source differences in schizophrenia: combining gray matter structural and auditory oddball functional data

    Hum. Brain Mapp.

    (2006)
  • D.J. Callen et al.

    Limbic system perfusion in Alzheimer's disease measured by MRI-coregistered HMPAO SPET

    Eur. J. Nucl. Med. Mol. Imaging

    (2002)
  • P. Calvini et al.

    Automatic analysis of medial temporal lobe atrophy from structural MRIs for the early assessment of Alzheimer disease

    Med. Phys.

    (2009)
  • J.A. Claassen et al.

    Cholinergically mediated augmentation of cerebral perfusion in Alzheimer's disease and related cognitive disorders: the cholinergic-vascular hypothesis

    J. Gerontol. A Biol. Sci. Med. Sci.

    (2006)
  • J.J. Claus et al.

    The diagnostic value of SPECT with Tc 99m HMPAO in Alzheimer's disease: a population-based study

    Neurology

    (1994)
  • W. Dai et al.

    Mild cognitive impairment and Alzheimer disease: patterns of altered cerebral blood flow at MR imaging

    Radiology

    (2009)
  • R.S. Desikan et al.

    MRI measures of temporoparietal regions show differential rates of atrophy during prodromal AD

    Neurology

    (2008)
  • J.A. Detre et al.

    Perfusion imaging

    Magn. Reson. Med.

    (1992)
  • M. Devous

    Functional brain imaging in the dementias: role in early detection, differential diagnosis, and longitudinal studies

    Eur. J. Nucl. Med. Mol. Imaging

    (2002)
  • A.T. Du et al.

    Hypoperfusion in frontotemporal dementia and Alzheimer disease by arterial spin labeling MRI

    Neurology

    (2006)
  • Cited by (35)

    • Image processing: an early detection tool for Alzheimer’s disease

      2023, Smart Diagnostics for Neurodegenerative Disorders: Neuro-sensors
    • Hyperbaric oxygen therapy ameliorates pathophysiology of 3xTg-AD mouse model by attenuating neuroinflammation

      2018, Neurobiology of Aging
      Citation Excerpt :

      These changes were accompanied with reduction in proinflammatory cytokines, elevation in antiinflammatory cytokines and phagocytic markers, and improvement in behavioral deficits in old 3xTg mice. Growing evidence suggests that AD patients present reduced cerebral perfusion, which can be detected in early stages of the disease and deteriorates with disease progression (Alsop et al., 2010; Binnewijzend et al., 2013; Chao et al., 2010; Chen et al., 2011; Roher et al., 2012; Thomas et al., 2015; Tosun et al., 2010). Moreover, hypoxia has been shown to contribute to several AD pathologies, such as accumulation of Aβ, hyperphosphorylation of tau, neuroinflammation, and degeneration of neurons (Zhang and Le, 2010).

    • Brain perfusion in polysubstance users: Relationship to substance and tobacco use, cognition, and self-regulation

      2015, Drug and Alcohol Dependence
      Citation Excerpt :

      All participants were instructed to remain awake with eyes closed during the 7 min whole-brain ASL perfusion sequence. Structural MRI data were aligned with perfusion data using a fluid-flow warping based distortion correction algorithm, and corrections for partial volume effects controlled for brain atrophy; method details are described elsewhere (Tosun et al., 2010). Regional cerebral blood flow images were corrected for partial volume effects and co-aligned with FreeSurfer v5.1 parcelated labels (Fischl et al., 2002, 2004) to yield subject-specific perfusion averages in pre-defined regions.

    View all citing articles on Scopus
    View full text