Principal component analysis in mild and moderate Alzheimer's disease — A novel approach to clinical diagnosis
Introduction
Comparisons between regional cerebral blood flow (rCBF) patterns in Alzheimer's disease (AD) patients and normal controls (CTR) by single photon emission computed tomography (SPECT) or positron emission tomography (PET) have in the past mainly been carried out by either visual evaluation or outlining the regions of interest (ROIs) in a manual or semiautomatic mode (Syed et al., 1992, Salmon et al., 1994, Bonte et al., 1997, Jagust et al., 1997, Jagust et al., 2001). In recent years, several digitized spatial standardization software programs have been introduced, and some of them have been extensively used both in research and clinical investigations (Friston et al., 1995, Worsley et al., 1996, Hirsch et al., 1997, Houston et al., 1998, Bartenstein et al., 1997, Imran et al., 1999). Among them, the Computerized Brain Atlas (CBA) (Greitz et al., 1991, Thurfjell et al., 1995, Andersson and Thurfjell, 1997), when registered to imaging data, defines volumes of interest (VOIs), corresponding to anatomical and functional areas of the brain (Brodmann areas and deep grey structures). It thus allows investigating rCBF relationships between anatomically distributed but physiologically correlated brain regions using principal component analysis (PCA) (Pagani et al., 2002). Moreover, the univariate approach treats all variables as if they represent independent measures across brain regions, thus ignoring the extensive and structured connections between them. Possible networking is better investigated by multivariate analysis describing pattern of covariance at both the voxel and VOI levels.
Multivariate spatial covariance methods have been proposed in functional neuroimaging by several authors (Moeller et al., 1987, Moeller et al., 1999, Eidelberg et al., 1990, Eidelberg et al., 1994, Friston et al., 1993, Friston et al., 1999, Strother et al., 1995, McIntosh et al., 1996, Jones et al., 1998, Zuendorf et al., 2003, Greicius et al., 2004, Lozza et al., 2004, Habeck et al., 2005, Hu et al., 2005, Kerrouche et al., 2006). All such studies were meant to overcome the concept of functional segregation as derived by univariate analysis of independent variables and to interpret the results in the frame of knowledge of brain circuitry and thus from a more holistic point of view.
The application of PCA aims to reduce the dimensionality of the data matrix through the grouping of VOIs into principal components that will undergo further statistical analysis. In AD, the rCBF distribution is likely to be the result of a combination of pathological lesions and the effects they can have on distant parts of the brain (Stoub et al., 2006). Such an approach has been applied to 18F-FDG PET data to characterize patterns of covariance in AD (Salmon et al., 2007) and to distinguishing between AD and vascular dementia (VaD) (Kerrouche et al., 2006). In studies of resting state rCBF with H215O PET (Scarmeas et al., 2004, Devanand et al., 2006) or SPECT (Huang et al., 2007), AD patients and subjects with mild cognitive impairment showed a distinct covariance pattern versus controls. PCA has been successfully performed implementing both a voxel-based analysis (Kerrouche et al., 2006, Scarmeas et al., 2004, Devanand et al., 2006) and using discrete regions of interest (Huang et al., 2007). These studies have consistently shown that PC composition may allow inferences on the peculiar functional re-organization of brain networking in both AD dementia and pre-dementia states, as well as in VaD (Kerrouche et al., 2006) and frontotemporal dementia (Salmon et al., 2006). Therefore, we hypothesised that investigating AD by means of multivariate analysis (i.e. PCA) might add knowledge about the rCBF pattern and neural connectivity underlying this complex neurodegenerative disorder.
The aim of the present study is to use multivariate statistical methodology to assess rCBF distribution differences between two groups of AD patients with different levels of disease severity and a group of normal controls.
Section snippets
Normal subjects
About 100 subjects older than 50 years were contacted by the Unit of Clinical Neurophysiology of the University of Genoa during University courses reserved for elderly people. Forty-four subjects agreed to participate and underwent brain SPECT examination by the same procedure used for patients. All subjects were carefully screened by general medical history and clinical examination. Complete blood counts, serum glucose, creatinine and total cholesterol, blood urea nitrogen, and urinalysis were
Results
In the VOI analysis, there was a significant interaction between VOIs and Groups (F(52,2262) = 4.825; P < 0.001). No hemispheric effect was found. The ANCOVA performed on the average value of bilateral VOIs found in both comparisons (i.e., CTR vs. mildAD and CTR vs. modAD) six bilateral Brodmann areas (BAs, 21, 37, 38, 39, 31 and nc. caudatus) belonging to temporal, parietal and limbic cortex and to the deep grey structures in which patients showed a significant rCBF decrease as compared with
Interpretation of the findings
The main finding of the study was the identification by PCA of large regions in which perfusion significantly decreased in mildAD as compared with controls, i.e., the real clinical question in early diagnosis. In fact, in the CTR vs. mildAD comparison, the significant areas highlighted by CBA-PCA-ANCOVA disclosed in mildAD a significantly lower rCBF distribution in large regions in both hemispheres (i.e. temporal lobe, inferior parietal lobe, precuneus and posterior cingulate cortex; see Fig. 1
Conclusions
In conclusion, the present investigation confirmed the rCBF decrease in several cortical areas in two groups of AD patients, as already widely known in literature. The implementation of principal component analysis revealed a strong covariance between regions belonging to the temporo-parietal cortex and to the limbic system. These finding add depth to the statistical analysis and yield more complete information on the networking underlying pathological changes in AD. Moreover, highlighting
Acknowledgments
The authors thank Dr. Antonietta Di Salvatore for her wise suggestions on statistical methodology and Ms Charlotte Williams for her help in English language revision. The experiments comply with the current laws of the country in which they were performed.
References (62)
- et al.
Combined 99mTc-ECD SPECT and neuropsychological studies in MCI for the assessment of conversion to AD
Neurobiology of Aging
(2006) - et al.
“Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician
Journal of Psychiatric Research
(1975) - et al.
Unified SPM-ICA for fMRI analysis
Neuroimage
(2005) - et al.
Imaging markers of mild cognitive impairment: multivariate analysis of CBF SPECT
Neurobiology of Aging
(2007) - et al.
Spatial pattern analysis of functional brain images using partial least squares
Neuroimage
(1996) - et al.
Effects of long-term Donepezil therapy on rCBF of Alzheimer's patients
Clinical Neurophysiology
(2002) - et al.
Decomposition of metabolic brain clusters in the frontal variant of frontotemporal dementia
Neuroimage
(2006) - et al.
On the multivariate nature of brain metabolic impairment in Alzheimer's disease
Neurobiology of Aging
(2009) - et al.
Covariance PET patterns in early Alzheimer's disease and subjects with cognitive impairment but no dementia: utility in group discrimination and correlations with functional performance
Neuroimage
(2004) - et al.
Multivariate network analysis of fiber tract integrity in Alzheimer's disease
Neuroimage
(2007)
CBA — an atlas based software tool used to facilitate the interpretation of neuroimaging data
Computer Methods & Programs in Biomedicine
Positron emission tomography and single-photon emission computed tomography brain imaging in the evaluation of Dementia
Seminars in Nuclear Medicine
Implementation and validation of a fully automatic system for intra- and inter-individual registration of PET brain scans
Journal of Computer Assisted Tomography
Quantitative assessment of cerebral blood flow in patients with Alzheimer's disease by SPECT
Journal of Nuclear Medicine
Brain blood flow in the dementias, SPECT with histopathologic correlation in 54 patients
Radiology
A method for attenuation correction in radionuclide computed tomography
IEEE Transactions on Nuclear Science
Mild cognitive impairment. Can FDG-PET predict who is to rapidly convert to Alzheimer's disease?
Neurology
PET network abnormalities and cognitive decline in patients with mild cognitive impairment
Neuropsychopharmacology
Functional brain imaging in the dementias: role in early detection, differential diagnosis, and longitudinal studies
European Journal of Nuclear Medicine and Molecular Imaging
Systematic review of the diagnostic accuracy of 99mTc-HMPAO-SPECT in Dementia
American Journal of Geriatric Psychiatry
The metabolic anatomy of Parkinson's disease: complementary (18F)fluorodeoxyglucose and (18F)fluorodopa positron emission tomographic studies
Movement Disorders
The metabolic topography of parkinsonism
Journal of Cerebral Blood Flow and Metabolism
Functional connectivity: the principal-component analysis of large (PET) data set
Journal of Cerebral Blood Flow and Metabolism
Statistical parametric maps in functional imaging: a general linear approach
Human Brain Mapping
Revealing interactions among brain systems with nonlinear PCA
Human Brain Mapping
The development of ASPECT, an annular single crystal brain camera for high efficiency SPECT
IEEE Transactions on Nuclear Science
Brain SPET perfusion in early Alzheimer's disease: where to look?
European Journal of Nuclear Medicine and Molecular Imaging
Default-mode network activity distinguishes Alzheimer's disease from healthy aging: evidence from functional MRI
Proceedings of the National Academy of Sciences of the United States of America
A computerized brain atlas: construction, anatomical content, and some applications
Journal of Computer Assisted Tomography
A new approach to spatial covariance modeling of functional brain imaging data: ordinal trend analysis
Neural Computation
Reduction of regional cerebral blood flow and cognitive impairment in patients with Alzheimer's disease: evaluation of an observer-independent analytic approach
Dementia and Geriatric Cognitive Disorders
Cited by (33)
Parkinson's Disease Diagnosis beyond Clinical Features: A Bio-marker using Topological Machine Learning of Resting-state Functional Magnetic Resonance Imaging
2023, NeuroscienceCitation Excerpt :There are mainly six approaches for rs-fMRI analysis in the open literature: Seed-based approach, Network-based approaches, Graph Analysis, Regional Homogeneity, Amplitude of Low-frequency Fluctuations, and Machine Learning-based approaches (Tahmasian et al., 2015). Among them, linear dimensionality reduction techniques (e.g., Principal Components Analysis (PCA) (Pagani et al., 2009), and Independent Components Analysis (ICA) (Calhoun et al., 2009)), Machine Learning-based Classifiers (e.g., Support Vector Machine (SVM) (Fan et al., 2006)), and Manifold Learning (e.g., Local Linear Embedding (LLE) (Shen et al., 2010), and t-distributed Stochastic Neighbor Embedding (t-SNE) (Chang et al., 2018)), have been used in neurological disorder analysis. These methods embed high-dimensional temporal BOLD signals into low-dimension space for visualization or pattern recognition.
Social brain network predicts real-world social network in individuals with social anhedonia
2021, Psychiatry Research - NeuroimagingCitation Excerpt :Second, this study combined the two groups of high and low SA for PCA to circumvent the problem of small sample size. Although several previous studies applied a similar method to analyze brain characteristics across different participant groups (Pagani et al., 2009; Rudie et al., 2012), future study should recruit a larger sample and conduct tests independently to each group to verify our preliminary findings. Third, we used self-report questionnaires to measure social network.
Multi-modal AD classification via self-paced latent correlation analysis
2019, NeurocomputingCitation Excerpt :Feature selection is an effective technique for dimensionality reduction by removing the irrelevant features. At present, a number of feature selection approaches have been applied into AD classification methods include multi-task feature selection (MTFS) [7], group lasso [12] and principal component analysis (PCA) [13]. In the multi-modal methods, these selected features of each modality are often directly combined to predict the class label [14].
Volume of interest-based [<sup>18</sup>F]fluorodeoxyglucose PET discriminates MCI converting to Alzheimer's disease from healthy controls. A European Alzheimer's Disease Consortium (EADC) study
2015, NeuroImage: ClinicalCitation Excerpt :However, the power of the present analysis including large cohorts of subjects and a sophisticated regional and asymmetry analysis might have allowed the identification of subtle but significant changes already at the early stages of the disease. Frontal and orbitofrontal cortex regions were also grouped by exploratory data analysis into separated components (Nobili et al., 2008; Pagani et al., 2009) and, besides showing significantly lower FDG uptake in MCI converters, in conjunction with a verbal memory test, were among the better discriminators between MCI-converters and healthy controls, with a sensitivity of 82% and a specificity of 100% (Nobili et al., 2008). The presence of the Postcentral Gyrus/Precentral Gyrus/Supplementary Motor Area complex among the most discriminating meta-VOIs underscores an interesting methodological issue.
Towards mapping the brain connectome in depression: Functional connectivity by perfusion SPECT
2014, Psychiatry Research - NeuroimagingCitation Excerpt :In Alzheimer´s disease (AD), brain connectome studies based on results from electroencephalography (EEG), positron emission tomography (PET) and magnetic resonance imaging (MRI) demonstrated disrupted network connectivity pattern in support of the theory that AD is a disconnection syndrome which might be the functional basis of the cognitive deficits (reviewed in Delbeuck et al., 2003; Xie and He, 2012). In this respect, factorial analysis was implemented to discriminate patients from controls resulting in a high accuracy of the method (Nobili et al., 2008; Pagani et al., 2009). However these studies investigated the correlations between regions defined a priori, an approach that could have resulted in missing brain networking at the cluster level.