Voxel- and VOI-based analysis of SPECT CBF in relation to clinical and psychological heterogeneity of mild cognitive impairment

doi:10.1016/S1053-8119(03)00168-X

NeuroImage

Volume 19, Issue 3, July 2003, Pages 1137-1144

https://doi.org/10.1016/S1053-8119(03)00168-X Get rights and content

Abstract

This study aimed to explore the heterogeneity of mild cognitive impairment (MCI) and detect differences in regional cerebral blood flow (rCBF) and cognitive function between progressive mild cognitive impairment (PMCI) and stable mild cognitive impairment (SMCI) in order to identify specific changes useful for early diagnosis of dementia. SPECT was performed in 82 MCI subjects and 20 controls using Tc-99m hexamethylpropyleneamine oxime. Cognitive functions were tested in five domains which included episodic memory, semantic memory, visuospatial function, attention, and general cognitive function. After the initial examination, MCI subjects were clinically followed for an average of 2 years. Twenty-eight subjects progressed to dementia and were defined as PMCI at baseline and 54 subjects remained stable and were defined as SMCI at baseline. The baseline rCBF and cognitive function of PMCI, SMCI, and controls were compared. PMCI had decreased relative rCBF in the parietal lobes and increased relative rCBF in prefrontal cortex compared to SMCI and controls at baseline. The cognitive function of PMCI was more severely impaired compared to SMCI with respect to episodic memory and visuospatial and general cognitive function. Both SPECT and neuropsychological tests had moderate discriminant function between PMCI and SMCI at baseline with the area under the receiver operating characteristic (ROC) curve at 75–77%. The combination of these two methods improved the diagnostic accuracy with the area under the ROC curve at 82–84%. Semantic memory and attention were negatively correlated with left prefrontal relative rCBF among the study population. The results show that the clinical heterogeneity of MCI is reflected in different patterns of psychological and CBF changes. Combined SPECT investigation and neuropsychological testing might predict the future development of dementia in patients with MCI.

Introduction

Mild cognitive impairment (MCI) is an operational diagnostic term which includes the preclinical phase of dementia. However, it is a large heterogeneous group with a variety of clinical outcomes. Some MCI subjects go on developing dementia, particularly Alzheimer’s disease (AD). These MCI subjects are defined as having progressive mild cognitive impairment (PMCI). Some may remain stable and never progress to any significant extent. Subjects with functional memory impairment induced by psychological disorders such as depression are also included in the MCI group. Such patients might improve after several months’ follow-up. Identification of PMCI versus nonprogressive or stable mild cognitive impairment (SMCI) subjects is important for the early diagnosis of dementia and effective therapeutical interventions, which in turn may delay or even prevent the onset of the disease process.

Memory deficits, especially delayed recall, have been frequently reported to be the first cognitive function affected in the early stage of AD Elias et al 2000, Small et al 2000, Tierney et al 2000. However, the memory deficits remain relatively stable for a long preclinical phase until dementia can be diagnosed (Small et al., 2000). Simultaneous impairments have also been found in multiple cognitive domains, such as verbal, visuospatial processes, attention and executive function, abstract reasoning, and retention. Procedural memories, as well as sensory and motor abilities, are relatively well preserved Howieson et al 1997, Small et al 1997, Tierney et al 2000. Moreover, Tierney et al. (2000) showed that the Mini-Mental State Examination (MMSE) is a useful test in predicting future onset of AD, with a specificity of 96 to 98%, but a sensitivity of 31 to 41%.

However, results from functional imaging studies have so far shown some inconsistency. Small et al. (1995) showed that the metabolic reduction in parietal lobe and hemisphere asymmetry were more pronounced among the genetic at-risk subjects than controls. The single photon emission computed tomography (SPECT) study of Kogure et al. (2000) found that posterior cingulate and precuneus were the earliest signs of the disease progress, and even preceded the hypoperfusion of hippocampus and parahippocampus. Moreover, the functional changes of the parietotemporal region, thalamus, and prefrontal cortex in preclinical dementia are also under debate Johnson et al 1998, Kennedy et al 1995, Kogure et al 2000, Minoshima et al 1997, Reiman et al 1996.

In a previous study, we found a reduction in posterior cingulate perfusion was related to later development of dementia in patients with MCI (Huang et al., 2002). In that study we used an automated volumes-of-interest (VOI)-based analysis, designed to map significant regional changes at an individual level. However, especially when comparing group data, the VOI method is restricted by the rather large predefined anatomical VOIs. A voxel-based analysis of Statistical Parametric Mapping (SPM) (Wellcome Department of Cognitive Neurology, University College, London) might give the possibility of finding more anatomically discrete changes. However, the feasibility of SPM for analyzing the subjects at an individual level is limited. For these reasons, we have now examined a larger sample of MCI patients with voxel-based analysis of SPM and explore the possibility of calculating the diagnostic accuracy between groups using VOI-based BRASS analysis, with VOI selection guided by the SPM results.

Our study aimed to detect the differences of brain perfusion and compare the patterns of cognitive function among PMCI, SMCI, and age-matched controls in order to identify the subjects at risk for the future development of dementia. We also correlated the regional cerebral blood flow (rCBF) and the neuropsychological tests to explore the relation between brain perfusion and cognitive function.

Section snippets

Subject selection

Eighty-two MCI subjects were evaluated. The subjects were selected from those individuals consecutively investigated for suspected dementia at the Geriatric Clinic, Huddinge University Hospital, Stockholm, Sweden. Twenty control subjects were recruited through advertisements in the press, the Swedish Pensioner Society, and a Driving and Aging project.

All subjects underwent general medical, neurological, psychiatric, and neuropsychological evaluation, as well as neuroimage diagnostic procedures

Neuropsychological tests

Group comparisons showed that both PMCI and SMCI had impaired function at baseline in all the tested cognitive domains compared to controls (Table 1). PMCI had more severe cognitive deficits compared to SMCI at baseline in episodic memory represented by the word recognition test, in visuospatial function represented by Block Design, and in general cognitive function represented by MMSE.

Group comparison

The relative rCBF was obtained using global CBF as a reference region. Similar patterns were found before and

Discussion

The results demonstrated that the discriminative factors between PMCI and SMCI at baseline involve both rCBF and cognitive function. It is important to emphasize that in the clinical diagnostic procedure at baseline, these groups both fulfilled MCI criteria, and could not be separated without follow-up data. Using voxel-based analysis we could detect selective rCBF changes in PMCI with reduced blood flow in parietal lobe and enhanced brain perfusion in prefrontal lobe. The neuropsychological

Conclusion

This study combining voxel- and VOI-based analysis shows that the clinical heterogeneity of MCI is reflected in different patterns of psychological and rCBF changes. PMCI had an impaired cognitive function and significant rCBF changes compared to SMCI. Combined SPECT investigation and neuropsychological testing might predict the future development of dementia in patients with MCI.20

Acknowledgements

This work was supported by the Swedish Medical Research Council, the Swedish Council for Social Research, the Gamla Tjänarinnor Foundation, the Greta Lindenau-Hansell Foundation, Alzheimerfonden, the Karolinska Institute, and the Swedish Society of Medicine. Dr. Jesper Andersson at the MR center, Karolinska Hospital, is gratefully acknowledged for his kind and efficient and in the Statistical Parametric Mapping analysis.

References (23)

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)
C. Huang et al.
Discrimination of Alzheimer’s disease and mild cognitive impairment by equivalent EEG sourcesa cross-sectional and longitudinal study
Clin. Neurophysiol.
(2000)
C. Huang et al.
Cingulate cortex hypoperfusion predicts Alzheimer’s disease in mild cognitive impairment
BMC Neurol.
(2002)
S.-C. Johnson et al.
The relationship between fMRI activation and cerebral atrophycomparison of normal aging and Alzheimer disease
NeuroImage
(2000)
A.-M. Kennedy et al.
Deficits in cerebral glucose metabolism demonstrated by positron emission tomography in individuals at risk of familial Alzheimer’s disease
Neurosci. Lett.
(1995)
American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, fourth ed. American...
L. Backman et al.
Episodic memory functioning in a community-based sample of old adults with major depressionutilization of cognitive support
J. Abnorm. Psychol.
(1994)
S.-Y. Bookheimer et al.
Patterns of brain activation in people at risk for Alzheimer’s disease
N. Engl. J. Med.
(2000)
H. Braak et al.
Neuropathological stageing of Alzheimer-related changes
Acta Neuropathol.
(1991)
D. Cardebat et al.
Brain correlates of memory processes in patients with dementia of Alzheimer’s typea SPECT activation study
J. Cereb. Blood Flow Metab.
(1998)

M.-F. Elias et al.

The preclinical phase of Alzheimer diseasea 22-year prospective study of the Framinghan Cohort

Arch. Neurol.

(2000)

Cited by (82)

Comprehensive review of task-based neuroimaging studies of cognitive deficits in Alzheimer's disease using electrophysiological methods
2023, Ageing Research Reviews
With an aging population, cognitive decline and neurodegenerative disorders are an emerging public health crises with enormous, yet still under-recognized burdens. Alzheimer’s disease (AD) is the most common type of dementia, and the number of cases is expected to dramatically rise in the upcoming decades. Substantial efforts have been placed into understanding the disease. One of the primary avenues of research is neuroimaging, and while positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) are most common, crucial recent advancements in electrophysiological methods such as magnetoencephalography (MEG) and electroencephalography (EEG) have provided novel insight into the aberrant neural dynamics at play in AD pathology. In this review, we outline task-based M/EEG studies published since 2010 using paradigms probing the cognitive domains most affected by AD, including memory, attention, and executive functioning. Furthermore, we provide important recommendations for adapting cognitive tasks for optimal use in this population and adjusting recruitment efforts to improve and expand future neuroimaging work.
HYDRA: Revealing heterogeneity of imaging and genetic patterns through a multiple max-margin discriminative analysis framework
2017, NeuroImage
Citation Excerpt :
However, this approach ignores the heterogeneous nature of diseases, which is supported by ample evidence. Typical examples of brain disorders that are characterized by a heterogeneous clinical presentation include both neurodevelopmental and neurodegenerative disorders: autism spectrum disorder (ASD) comprises neurodevelopmental disorders characterized by deficits in social communication and repetitive behaviors (Geschwind and Levitt, 2007; Jeste and Geschwind, 2014); schizophrenia and Parkinson's disease can be subdivided into distinct groups by separating its symptomatology to discrete symptom domains (Buchanan and Carpenter, 1994; Graham and Sagar, 1999; Koutsouleris et al., 2008; Nenadic et al., 2010; Zhang et al., 2015; Lewis et al., 2005); Alzheimer's disease (AD) can be separated into three subtypes on the basis of the distribution of neurofibrillary tangles (Murray et al., 2011); and mild cognitive impairment (MCI) may be further classified based on the type of specific cognitive impairment (Huang et al., 2003; Whitwell et al., 2007). Disentangling disease heterogeneity may significantly contribute to our understanding and lead to a more accurate diagnosis, prognosis and targeted treatment.
Multivariate pattern analysis techniques have been increasingly used over the past decade to derive highly sensitive and specific biomarkers of diseases on an individual basis. The driving assumption behind the vast majority of the existing methodologies is that a single imaging pattern can distinguish between healthy and diseased populations, or between two subgroups of patients (e.g., progressors vs. non-progressors). This assumption effectively ignores the ample evidence for the heterogeneous nature of brain diseases. Neurodegenerative, neuropsychiatric and neurodevelopmental disorders are largely characterized by high clinical heterogeneity, which likely stems in part from underlying neuroanatomical heterogeneity of various pathologies. Detecting and characterizing heterogeneity may deepen our understanding of disease mechanisms and lead to patient-specific treatments. However, few approaches tackle disease subtype discovery in a principled machine learning framework. To address this challenge, we present a novel non-linear learning algorithm for simultaneous binary classification and subtype identification, termed HYDRA (Heterogeneity through Discriminative Analysis). Neuroanatomical subtypes are effectively captured by multiple linear hyperplanes, which form a convex polytope that separates two groups (e.g., healthy controls from pathologic samples); each face of this polytope effectively defines a disease subtype. We validated HYDRA on simulated and clinical data. In the latter case, we applied the proposed method independently to the imaging and genetic datasets of the Alzheimer's Disease Neuroimaging Initiative (ADNI 1) study. The imaging dataset consisted of T1-weighted volumetric magnetic resonance images of 123 AD patients and 177 controls. The genetic dataset consisted of single nucleotide polymorphism information of 103 AD patients and 139 controls. We identified 3 reproducible subtypes of atrophy in AD relative to controls: (1) diffuse and extensive atrophy, (2) precuneus and extensive temporal lobe atrophy, as well some prefrontal atrophy, (3) atrophy pattern very much confined to the hippocampus and the medial temporal lobe. The genetics dataset yielded two subtypes of AD characterized mainly by the presence/absence of the apolipoprotein E (APOE) ε4 genotype, but also involving differential presence of risk alleles of CD2AP, SPON1 and LOC39095 SNPs that were associated with differences in the respective patterns of brain atrophy, especially in the precuneus. The results demonstrate the potential of the proposed approach to map disease heterogeneity in neuroimaging and genetic studies.
Machine learning as a means toward precision diagnostics and prognostics
2016, Machine Learning and Medical Imaging
Machine learning plays an essential role in medical imaging. Pattern analysis techniques can identify, and quantify, subtle and spatially complex patterns of disease-induced changes in the brain despite confounding statistical noise and inter-individual variability. This allows the construction of sensitive biomarkers that can identify disease, or risk of developing it, and characterize future clinical progression on an individual patient basis. Thus pattern analysis techniques have become an indispensable part of the growing need for personalized, predictive medicine. However, despite important advances, several challenges remain before they can gain widespread acceptance as tools for precision diagnostics and prognostics in clinical practice. These include: (i) feature extraction and dimensionality reduction; (ii) readily interpreting complex multivariate models; and (iii) elucidating disease heterogeneity. In this chapter, we describe these challenges, putting emphasis on possible solutions, and present evidence of the usefulness of machine learning techniques at the clinical and research levels.
Utility of a semi-quantification and automatic statistical analysis for the interpretation of brain perfusion SPECT in Alzheimer's disease
2014, Medecine Nucleaire
L’interprétation visuelle de la tomographie par émission monophotonique (TEMP) de perfusion cérébrale pour le diagnostic de maladie d’Alzheimer pouvant être difficile, on souhaite bénéficier de techniques permettant d’augmenter les performances diagnostiques, notamment pour des observateurs peu expérimentés.
Trois observateurs ont rétrospectivement interprété en aveugle 77 TEMP à l’ECD-^99mTc demandées dans un contexte de neurodégénérescence suspectée. La probabilité de maladie d’Alzheimer a été cotée de 1 à 5 et la sensibilité et la spécificité optimales ont été déterminées à partir de courbes ROC. Les observateurs ont aussi attribué un score d’hypoperfusion à 13 régions cérébrales ; un diagnostic a été porté à partir de cette semi-quantification en utilisant une règle logique simple. Les TEMP ont aussi été analysées avec le logiciel Planet Neuro et un diagnostic porté à partir d’un seuil pour 2 des 90 régions analysées. Le diagnostic de référence était établi par un bilan neurologique spécialisé avec au moins 5 ans de recul.
La sensibilité et la spécificité des trois observateurs étaient 50 %, 56 %, 28 % et 76 %, 83 % et 90 %, respectivement, pour l’interprétation visuelle. Avec la semi-quantification, elles étaient de 50 %, 67 %, 61 % et 54 %, 85 % et 69 % et les résultats de deux observateurs étaient plus homogènes. Avec la quantification automatique, on obtenait 78 % et 81 %.
La semi-quantification paraît améliorer les résultats de certains observateurs et la cohérence entre certains observateurs. La quantification automatique semble toutefois le meilleur moyen d’aider tous les observateurs à améliorer leurs performances diagnostiques.
As the visual interpretation of the brain perfusion SPECT for the diagnosis of Alzheimer's disease might be difficult, it is wished to benefit from techniques allowing to improve the diagnostic accuracy especially for the inexperienced readers.
Three readers retrospectively interpreted 77 ^99mTc-ECD SPECT studies within the context of suspected neurodegenerative disease. The probability of Alzheimer's disease was quoted from 1 to 5, and the optimum sensitivity and specificity values were determined from ROC curves. The readers also attributed a hypoperfusion score to 13 brain regions; a diagnosis was obtained from this semi-quantification using a simple logic rule. The SPECT scans were also analyzed using the Planet Neuro software and a diagnosis was reached from a threshold for 2 of the 90 analyzed regions. The gold standard was established by a comprehensive neurological evaluation with at least 5 years of follow-up.
The sensitivities and specificities of the three readers were 50 %, 56 %, 28 % and 76 %, 83 % and 90 %, respectively. Using the semi-quantification they were 50 %, 67 %, 61 % and 54 %, 85 %, 69 %. The results of two readers were more in agreement. Using the automatic quantification, sensitivity was 78 % and specificity 81 %.
The semi-quantification seems to ameliorate the results of and the agreement between certain readers. However, the automatic quantification seems to be the only way of assisting all of the readers to improve their diagnostic performances.
Biomarkers of Alzheimer's disease
2014, Biomarkers in Toxicology
Alzheimer’s disease (AD) is characterized by progressive synaptic and neuronal loss accompanied by the presence of two forms of amyloid: extracellular plaques and intracellular neurofibrillary tangles. Amyloid plaques are composed of the peptide Aβ, which self-associates to form insoluble fibrils, while neurofibrillary tangles are composed of hyperphosphorylated forms of the microtubule-associated protein tau. Along the way to forming fibrils, Aβ also forms soluble oligomeric toxins, which are hypothesized to be disease initiating factors. Decades before AD is clinically detectable, Aβ accumulation in the brain has already begun, giving Aβ biomarkers excellent prognostic value. Tau biomarkers are also prognostic and precede AD onset, although they reflect neural damage and disease severity much better than Aβ. Other diagnostic biomarkers, including brain atrophy, hypometabolism, hypoperfusion, and oxidative damage, are better diagnostic tools for estimating disease progression than for predicting AD onset, although no biomarkers have been validated for clinical use. The etiology, disease mechanisms, and biomarkers of AD are discussed in this chapter.
Generation of realistic HMPAO SPECT images using a subresolution sandwich phantom
2013, NeuroImage
Citation Excerpt :
Such interpretation is subjective and dependent on the operator's understanding of normality. Consequently the technique of anatomic standardisation and comparison of patients with controls and/or other patient groups is increasingly used in clinical practice and research (Huang et al., 2003; Imabayashi et al., 2004; Matsuda et al., 2007; Van Laere et al., 2002b). Several software packages are available that allow automated whole-brain analysis of brain SPECT studies based on anatomic standardisation.
Traditional interpretation of rCBF SPECT data is of a qualitative nature and is dependent on the observer's understanding of the normal distribution of the tracer. The use of a normal database in quantitative regional analysis facilitates the detection of functional abnormality in individual and group studies by accounting for inter-subject variability. The ability to simulate realistic images would allow various important areas related to the use of normal databases to be studied. These include the optimisation of the detection of abnormal blood flow and the portability of normal databases between gamma camera systems. To investigate this further we have constructed a hardware phantom and scanned various configurations of radioactive brain patterns and simulated skull configurations.
A subresolution sandwich phantom was constructed with a simulated skull which was assembled using a high-resolution segmented MR scan printed with a ^99mTcO₄− mixture and scanned using a double-headed gamma camera with parallel-hole collimators. Various different grey-to-white matter (GM:WM) ratios and aluminium simulated skull configurations were used. A single difference measure between the phantom data and a control database mean image was used for optimisation. The realism of phantom data was assessed using statistical parametric mapping (SPM) and ROI analysis.
Optimisation was achieved with a range of WM:GM ratios from 1.9 to 2.4:1 with various simulated skull configurations.
The ability to simulate realistic HMPAO SPECT scans has been demonstrated using a subresolution sandwich phantom. Further work, involving scanning the optimised phantom on different gamma camera systems and comparison with camera-specific normal databases should further refine the phantom configuration.

View all citing articles on Scopus

View full text

Regular articleVoxel- and VOI-based analysis of SPECT CBF in relation to clinical and psychological heterogeneity of mild cognitive impairment

Abstract

Introduction

Section snippets

Subject selection

Neuropsychological tests

Group comparison

Discussion

Conclusion

Acknowledgements

J. Psychiatr. Res.

Clin. Neurophysiol.

BMC Neurol.

NeuroImage

Neurosci. Lett.

Episodic memory functioning in a community-based sample of old adults with major depressionutilization of cognitive support

J. Abnorm. Psychol.

Patterns of brain activation in people at risk for Alzheimer’s disease

N. Engl. J. Med.

Neuropathological stageing of Alzheimer-related changes

Acta Neuropathol.

Brain correlates of memory processes in patients with dementia of Alzheimer’s typea SPECT activation study

J. Cereb. Blood Flow Metab.

The preclinical phase of Alzheimer diseasea 22-year prospective study of the Framinghan Cohort

Arch. Neurol.

Regular article
Voxel- and VOI-based analysis of SPECT CBF in relation to clinical and psychological heterogeneity of mild cognitive impairment