Sterling C Johnson

University of Wisconsin–Madison, Madison, Wisconsin, United States

Are you Sterling C Johnson?

Claim your profile

Publications (149)925.87 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Subtle cognitive and behavioral changes are common in early Parkinson's disease. The cause of these symptoms is probably multifactorial but may in part be related to extra-striatal dopamine levels. 6-[(18) F]-Fluoro-L-dopa (FDOPA) positron emission tomography has been widely used to quantify dopamine metabolism in the brain; the most frequently measured kinetic parameter is the tissue uptake rate constant, Ki. However, estimates of dopamine turnover, which also account for the small rate of FDOPA loss from areas of specific trapping, may be more sensitive than Ki for early disease-related changes in dopamine biosynthesis. The purpose of the present study was to compare effective distribution volume ratio (eDVR), a metric for dopamine turnover, to cognitive and behavioral measures in Parkinson's patients. We chose to focus the investigation on anterior cingulate cortex, which shows highest FDOPA uptake within frontal regions and has known roles in executive function. Fifteen non-demented early-stage PD patients were pretreated with carbidopa and tolcapone, a central catechol-O-methyl transferase (COMT) inhibitor, and then underwent extended imaging with FDOPA PET. Anterior cingulate eDVR was compared with composite scores for language, memory, and executive function measured by neuropsychological testing, and behavior change measured using two informant-based questionnaires, the Cambridge Behavioral Inventory and the Behavior Rating Inventory of Executive Function-Adult Version. Lower mean eDVR (thus higher dopamine turnover) in anterior cingulate cortex was related to lower (more impaired) behavior scores. We conclude that subtle changes in anterior cingulate dopamine metabolism may contribute to dysexecutive behaviors in Parkinson's disease.
    Brain Imaging and Behavior 12/2014; · 3.39 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The serotonin-1A (5-HT1A; 5-HT is 5-hydroxytryptamine) receptor is implicated in an array of neurologic and psychiatric disorders. Current PET radioligands targeting 5-HT1A receptors have limitations hindering widespread PET studies of this receptor system. The 5-HT1A–specific antagonist radioligand N-{2-[4-(2-methoxyphenyl)piperazinyl]ethyl}-N-(2-pyridyl)-N-(trans-4-18F-fluoromethylcyclohexane)carboxamide (18F-mefway) exhibited promising in vivo properties in rhesus monkeys. The goal of this work was to examine the in vivo cerebral binding profile and metabolism of 18F-mefway in humans. Methods: Dynamic 18F-mefway PET data were acquired for 6 healthy volunteers (4 women, 2 men; age, 22–38 y). Scans were initiated with the injection of 192–204 MBq of radiotracer, and data were acquired for 2 h. Venous blood samples were collected and assayed to examine the in vivo metabolism profile of 18F-mefway. To examine the test–retest variability of 18F-mefway, a second PET scan was acquired at least 2 wk later for 4 subjects. Regional binding potentials (BPNDs) were calculated with the multilinear reference tissue model, and voxelwise BPND maps were calculated with Logan graphical analysis. Regions surrounding the brain were carefully inspected for uptake of radiolabeled species in bone. Results: 18F-mefway uptake in the brain occurred quickly, with a peak standardized uptake value (SUV) of 1.7. Rapid washout in the cerebellum resulted in SUVs of 0.2 at 120 min, whereas regions with specific 5-HT1A binding exhibitedretention of radioligand, yielding SUVs of 0.4–0.9 at 120 min. Rapid metabolism of 18F-mefway was observed, with no detected 18F-fluoride ions in plasma. BPND values of 2.4 were measured in the mesial temporal lobe, with values of 1.6 in the insular cortex and 0.7–1.0 in other cortical regions. Stable BPND estimates were obtained using 90 min of dynamic data. Average test–retest variability was 8%. No evidence of radioactivity uptake in bone was observed. Conclusion: 18F-mefway exhibits favorable in vivo properties for serotonin 5-HT1A receptor measurements in humans. The simple radiosynthesis, high specific binding profile, and absence of PET signal in bone make 18F-mefway an attractive radiotracer for PET experiments examining the 5-HT1A receptor in neuropsychiatric disorders and drug intervention.
    Journal of Nuclear Medicine 11/2014; 55(12):1973-1979. · 5.56 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: This study tested the hypothesis that frequent participation in cognitively-stimulating activities, specifically those related to playing games and puzzles, is beneficial to brain health and cognition among middle-aged adults at increased risk for Alzheimer's disease (AD). Three hundred twenty-nine cognitively normal, middle-aged adults (age range, 43.2-73.8 years) enrolled in the Wisconsin Registry for Alzheimer's Prevention (WRAP) participated in this study. They reported their current engagement in cognitive activities using a modified version of the Cognitive Activity Scale (CAS), underwent a structural MRI scan, and completed a comprehensive cognitive battery. FreeSurfer was used to derive gray matter (GM) volumes from AD-related regions of interest (ROIs), and composite measures of episodic memory and executive function were obtained from the cognitive tests. Covariate-adjusted least squares analyses were used to examine the association between the Games item on the CAS (CAS-Games) and both GM volumes and cognitive composites. Higher scores on CAS-Games were associated with greater GM volumes in several ROIs including the hippocampus, posterior cingulate, anterior cingulate, and middle frontal gyrus. Similarly, CAS-Games scores were positively associated with scores on the Immediate Memory, Verbal Learning & Memory, and Speed & Flexibility domains. These findings were not modified by known risk factors for AD. In addition, the Total score on the CAS was not as sensitive as CAS-Games to the examined brain and cognitive measures. For some individuals, participation in cognitive activities pertinent to game playing may help prevent AD by preserving brain structures and cognitive functions vulnerable to AD pathophysiology.
    Brain Imaging and Behavior 10/2014; · 3.39 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Positive cerebrospinal fluid (CSF) biomarkers of tau and amyloid beta42 suggest possible active underlying Alzheimer's disease (AD) including neurometabolic dysfunction and neurodegeneration leading to eventual cognitive decline. But the temporal relationship between CSF, imaging markers of neural function, and cognition has not been described. Using a statistical mediation model, we examined relationships between cerebrospinal fluid (CSF) analytes (hyperphosphorylated tau (p-Tau181p), β-amyloid peptides 1-42 (Aβ1-42), total tau (t-Tau), and their ratios); change in cognitive function; and change in [18F]fluorodeoxyglucose (FDG) uptake using positron emission tomography (PET). We hypothesized that a) abnormal CSF protein values at baseline, result in cognitive declines by decreasing neuronal glucose metabolism across time, and b) the role of altered glucose metabolism in the assumed causal chain varies by brain region and the nature of CSF protein alteration. Data from 412 individuals participating in Alzheimer's Disease Neuroimaging (ADNI) cohort studies were included in analyses. At baseline, individuals were cognitively normal (N=82), or impaired: 241 with mild cognitive impairment, and 89 with Alzheimer's disease. A parallel-process latent growth curve model was used to test mediational effects of changes in regional FDG-PET uptake over time in relation to baseline CSF biomarkers and changes in cognition, measured with the 13-item Alzheimer Disease's Assessment Scale-cognitive subscale (ADAS-Cog). Findings suggested a causal sequence of events; specifically, FDG hypometabolism acted as a mediator between antecedent CSF biomarker alterations and subsequent cognitive impairment. Higher baseline concentrations of t-Tau, and p-Tau181p were more predictive of decline in cerebral glucose metabolism than lower baseline concentrations of Aβ1-42. FDG-PET changes appeared to mediate t-Tau or t-Tau/Aβ1-42-associated cognitive change across all brain regions examined. Significant direct effects of alterations in Aβ1-42 levels on hypometabolism were observed in a single brain region: middle/inferior temporal gyrus. Results support a temporal framework model in which reduced CSF amyloid-related biomarkers occur earlier in the pathogenic pathway, ultimately leading to detrimental cognitive effects. Also consistent with this temporal framework model, baseline markers of neurofibrillary degeneration predicted changes in brain glucose metabolism in turn causing longitudinal cognitive changes, suggesting that tau-related burden precedes neurometabolic dysfunction. While intriguing, the hypothesized mediational relationships require further validation. Copyright © 2014. Published by Elsevier Inc.
    NeuroImage 10/2014; · 6.13 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Episodic memory decline is one of the earliest preclinical symptoms of AD, and has been associated with an upregulation in the BOLD response in the prodromal stage (e.g. MCI) of AD. In a previous study, we observed upregulation in cognitively normal (CN) subjects with subclinical episodic memory decline compared to non-decliners. In light of this finding, we sought to determine if a separate cohort of Decliners will show increased brain activation compared to Stable subjects during episodic memory processing, and determine whether the BOLD effect was influenced by cerebral blood flow (CBF) or gray matter volume (GMV). Individuals were classified as a "Decliner" if scores on the Rey Auditory Verbal Learning Test (RAVLT) consistently fell ≥ 1.5 SD below expected intra- or inter-individual levels. FMRI was used to compare activation during a facial recognition memory task in 90 Stable (age = 59.1) and 34 Decliner (age = 62.1, SD = 5.9) CN middle-aged adults and 10 MCI patients (age = 72.1, SD = 9.4). Arterial spin labeling and anatomical T1 MRI were used to measure resting CBF and GMV, respectively. Stables and Decliners performed similarly on the episodic recognition memory task and significantly better than MCI patients. Compared to Stables, Decliners showed increased BOLD signal in the left precuneus on the episodic memory task that was not explained by CBF or GMV, familial AD risk factors, or neuropsychological measures. These findings suggest that subtle changes in the BOLD signal reflecting altered neural function may be a relatively early phenomenon associated with memory decline.
    Brain Imaging and Behavior 10/2014; · 3.39 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Cardiorespiratory fitness (CRF) is an objective measure of habitual physical activity (PA), and has been linked to increased brain structure and cognition. The gold standard method for measuring CRF is graded exercise testing (GXT), but GXT is not feasible in many settings. The objective of this study was to examine whether a non-exercise estimate of CRF is related to gray matter (GM) volumes, white matter hyperintensities (WMH), cognition, objective and subjective memory function, and mood in a middle-aged cohort at risk for Alzheimer's disease (AD). Three hundred and fifteen cognitively healthy adults (mean age =58.58 years) enrolled in the Wisconsin Registry for Alzheimer's Prevention underwent structural MRI scanning, cognitive testing, anthropometric assessment, venipuncture for laboratory tests, and completed a self-reported PA questionnaire. A subset (n = 85) underwent maximal GXT. CRF was estimated using a previously validated equation incorporating sex, age, body-mass index, resting heart rate, and self-reported PA. Results indicated that the CRF estimate was significantly associated with GXT-derived peak oxygen consumption, validating its use as a non-exercise CRF measure in our sample. Support for this finding was seen in significant associations between the CRF estimate and several cardiovascular risk factors. Higher CRF was associated with greater GM volumes in several AD-relevant brain regions including the hippocampus, amygdala, precuneus, supramarginal gyrus, and rostral middle frontal gyrus. Increased CRF was also associated with lower WMH and better cognitive performance in Verbal Learning & Memory, Speed & Flexibility, and Visuospatial Ability. Lastly, CRF was negatively correlated with self- and informant-reported memory complaints, and depressive symptoms. Together, these findings suggest that habitual participation in physical activity may provide protection for brain structure and cognitive function, thereby decreasing future risk for AD.
    Brain Imaging and Behavior 10/2014; · 3.39 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To examine whether engagement in physical activity might favorably alter the age-dependent evolution of Alzheimer disease (AD)-related brain and cognitive changes in a cohort of at-risk, late-middle-aged adults.
    Neurology 10/2014; · 8.30 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Nearly all adults with Down syndrome show neuropathology of Alzheimer's disease, including amyloid-β deposition, by their fifth decade of life. In the current study, we examined the association between brain amyloid-β deposition, assessed via in vivo assessments of neocortical Pittsburgh compound B, and scores on an extensive neuropsychological battery of measures of cognitive functioning in 63 adults (31 male, 32 female) with Down syndrome aged 30-53 years who did not exhibit symptoms of dementia. Twenty-two of the adults with Down syndrome were identified as having elevated neocortical Pittsburgh compound B retention levels. There was a significant positive correlation (r = 0.62, P < 0.0001) between age and neocortical Pittsburgh compound B retention. This robust association makes it difficult to discriminate normative age-related decline in cognitive functioning from any potential effects of amyloid-β deposition. When controlling for chronological age in addition to mental age, there were no significant differences between the adults with Down syndrome who had elevated neocortical Pittsburgh compound B retention levels and those who did not on any of the neuropsychological measures. Similarly, when examining Pittsburgh compound B as a continuous variable, after controlling for mental age and chronological age, only the Rivermead Picture Recognition score was significantly negatively associated with neocortical Pittsburgh compound B retention. Our findings indicate that many adults with Down syndrome can tolerate amyloid-β deposition without deleterious effects on cognitive functioning. However, we may have obscured true effects of amyloid-β deposition by controlling for chronological age in our analyses. Moreover, our sample included adults with Down syndrome who were most 'resistant' to the effects of amyloid-β deposition, as adults already exhibiting clinical symptoms of dementia symptoms were excluded from the study.
    Brain 07/2014; · 10.23 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background Insulin resistance (IR) increases Alzheimer's disease (AD) risk. IR is related to greater amyloid burden post-mortem and increased deposition within areas affected by early AD. No studies have examined if IR is associated with an in vivo index of amyloid in the human brain in late middle-aged participants at risk for AD. Methods Asymptomatic, late middle-aged adults (N = 186) from the Wisconsin Registry for Alzheimer's Prevention underwent [C-11]Pittsburgh compound B (PiB) positron emission tomography. The cross-sectional design tested the interaction between insulin resistance and glycemic status on PiB distribution volume ratio in three regions of interest (frontal, parietal, and temporal). Results In participants with normoglycemia but not hyperglycemia, higher insulin resistance corresponded to higher PiB uptake in frontal and temporal areas, reflecting increased amyloid deposition. Conclusions This is the first human study to demonstrate that insulin resistance may contribute to amyloid deposition in brain regions affected by AD.
    Alzheimer's and Dementia 07/2014; · 17.47 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Linear regression is a parametric model which is ubiquitous in scientific analysis. The classical setup where the observations and responses, i.e., (xi , yi ) pairs, are Euclidean is well studied. The setting where yi is manifold valued is a topic of much interest, motivated by applications in shape analysis, topic modeling, and medical imaging. Recent work gives strategies for max-margin classifiers, principal components analysis, and dictionary learning on certain types of manifolds. For parametric regression specifically, results within the last year provide mechanisms to regress one real-valued parameter, xi ∈ R, against a manifold-valued variable, yi ∈ . We seek to substantially extend the operating range of such methods by deriving schemes for multivariate multiple linear regression -a manifold-valued dependent variable against multiple independent variables, i.e., f : R (n) → . Our variational algorithm efficiently solves for multiple geodesic bases on the manifold concurrently via gradient updates. This allows us to answer questions such as: what is the relationship of the measurement at voxel y to disease when conditioned on age and gender. We show applications to statistical analysis of diffusion weighted images, which give rise to regression tasks on the manifold GL(n)/O(n) for diffusion tensor images (DTI) and the Hilbert unit sphere for orientation distribution functions (ODF) from high angular resolution acquisition. The companion open-source code is available on nitrc.org/projects/riem_mglm.
    06/2014; 2014:2705-2712.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Statistical analysis on arbitrary surface meshes such as the cortical surface is an important approach to understanding brain diseases such as Alzheimer's disease (AD). Surface analysis may be able to identify specific cortical patterns that relate to certain disease characteristics or exhibit differences between groups. Our goal in this paper is to make group analysis of signals on surfaces more sensitive. To do this, we derive multi-scale shape descriptors that characterize the signal around each mesh vertex, i.e., its local context, at varying levels of resolution. In order to define such a shape descriptor, we make use of recent results from harmonic analysis that extend traditional continuous wavelet theory from the Euclidean to a non-Euclidean setting (i.e., a graph, mesh or network). Using this descriptor, we conduct experiments on two different datasets, the Alzheimer's Disease NeuroImaging Initiative (ADNI) data and images acquired at the Wisconsin Alzheimer's Disease Research Center (W-ADRC), focusing on individuals labeled as having Alzheimer's disease (AD), mild cognitive impairment (MCI) and healthy controls. In particular, we contrast traditional univariate methods with our multi-resolution approach which show increased sensitivity and improved statistical power to detect a group-level effects. We also provide an open source implementation.
    NeuroImage 06/2014; 93:107–123. · 6.13 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Cellular studies suggest sphingolipids may cause or accelerate amyloid-beta (Aβ) and tau pathology but in vivo human studies are lacking. We determined cerebrospinal fluid levels of sphingolipids (ceramides and sphingomyelins), amyloid-beta (Aβ1-42, AβX-38, AβX-40, and AβX-42) and tau (T-tau and p-tau181) in 91 cognitively normal individuals, aged 36-69 years, with a parental history of Alzheimer's disease. The 18-carbon acyl chain length ceramide species was associated with AβX-38 (r = 0.312, p = 0.003), AβX-40 (r = 0.327, p = 0.002), and T-tau (r = 0.313, p = 0.003) but not with AβX-42 (r = 0.171, p = 0.106) or p-tau (r = 0.086, p = 0.418). All sphingomyelin species correlated (most p < 0.001) with all Aβ species and T-tau; many also correlated with p-tau. Results remained in regression models after controlling for age and APOE genotype. These results suggest in vivo relationships between cerebrospinal fluid ceramides and sphingomyelins and Aβ and tau levels in cognitively normal individuals at increased risk for Alzheimer's disease, indicating these sphingolipids may be associated with early pathogenesis.
    Neurobiology of Aging 05/2014; · 4.85 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The rhesus macaque exhibits age-related brain changes similar to humans, making an excellent model of normal aging. Calorie restriction is a dietary intervention that reduces age-related comorbidities in short-lived animals, and its effects are still under study in rhesus macaques. Here, using deterministic fiber tracking method, we examined the effects of age and calorie restriction on a diffusion tensor imaging measure of white matter integrity, fractional anisotropy (FA), within white matter tracks traversing the anterior (genu) and posterior (splenium) corpus callosum in rhesus monkeys. Our results show: (1) a significant inverse relationship between age and mean FA of tracks traversing the genu and splenium; (2) higher mean FA of the splenium tracks as compared to that of genu tracks across groups; and (3) no significant diet effect on mean track FA through either location. These results are congruent with the age-related decline in white matter integrity reported in humans and monkeys, and the anterior-to-posterior gradient in white matter vulnerability to normal aging in humans. Further studies are warranted to critically evaluate the effect of calorie restriction on brain aging in this unique cohort of elderly primates.
    Neuroscience Letters 03/2014; · 2.06 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The relative influence of amyloid burden, neuronal structure and function, and prior cognitive performance on prospective memory decline among asymptomatic late middle-aged individuals at risk for Alzheimer's disease (AD) is currently unknown. We investigated this using longitudinal cognitive data from 122 middle-aged adults (21 "Decliners" and 101 "Stables") enrolled in the Wisconsin Registry for Alzheimer's Prevention who underwent multimodality neuroimaging [11C-Pittsburgh Compound B (PiB), 18F-fluorodeoxyglucose (FDG), and structural/functional magnetic resonance imaging (fMRI)] 5.7 ± 1.4 years (range = 2.9-8.9) after their baseline cognitive assessment. Covariate-adjusted regression analyses revealed that the only imaging measure that significantly distinguished Decliners from Stables (p = .027) was a Neuronal Function composite derived from FDG and fMRI. In contrast, several cognitive measures, especially those that tap episodic memory, significantly distinguished the groups (p's<.05). Complementary receiver operating characteristic curve analyses identified the Brief Visuospatial Memory Test-Revised (BVMT-R) Total (.82 ± .05, p < .001), the BVMT-R Delayed Recall (.73 ± .06, p = .001), and the Reading subtest from the Wide-Range Achievement Test-III (.72 ± .06, p = .002) as the top three measures that best discriminated the groups. These findings suggest that early memory test performance might serve a more clinically pivotal role in forecasting future cognitive course than is currently presumed. (JINS, 2014, 20, 1-12).
    Journal of the International Neuropsychological Society 03/2014; · 2.70 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Caloric restriction (CR) without malnutrition increases longevity and delays the onset of age-associated disorders in short-lived species, from unicellular organisms to laboratory mice and rats. The value of CR as a tool to understand human ageing relies on translatability of CR's effects in primates. Here we show that CR significantly improves age-related and all-cause survival in monkeys on a long-term ~30% restricted diet since young adulthood. These data contrast with observations in the 2012 NIA intramural study report, where a difference in survival was not detected between control-fed and CR monkeys. A comparison of body weight of control animals from both studies with each other, and against data collected in a multi-centred relational database of primate ageing, suggests that the NIA control monkeys were effectively undergoing CR. Our data indicate that the benefits of CR on ageing are conserved in primates.
    Nature Communications 02/2014; 5:3557. · 10.74 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Aim: It is difficult to reliably detect the earliest signs of Alzheimer's disease (AD)-associated cognitive impairment. Our aim was to compare 3 psychometric methods of identifying amnestic mild cognitive impairment (aMCI) in a middle-aged longitudinal cohort enriched for AD risk. Methods: Wisconsin Registry for Alzheimer's Prevention (WRAP) participants with 3 waves of cognitive assessment over approximately 6 years were coded as meeting each of 3 psychometric aMCI definitions: (a) 'aMCI standard-baseline' used published norms to establish cutoffs for baseline performance; (b) 'aMCI robust-baseline' applied WRAP-specific robust norms to baseline, and (c) 'aMCI robust-multiwave' applied these robust norms across 3 waves of assessment. Each group was compared to a cognitively healthy subset. Results: Half the aMCI standard-baseline and one third of the aMCI robust-baseline group reverted to normal ranges at follow-up. Only the aMCI robust-multiwave method had an aMCI × age interaction showing significantly worse age-related memory declines in the aMCI group compared to the cognitively healthy group over 6 years of follow-up. Conclusion: Both cross-sectional methods showed instability over time, with many reverting to normal performance after baseline. The multiwave approach identified a group who showed progressive memory declines over 3 visits. Being able to detect progressive decline in late middle age is a critical step in improving prevention efforts. © 2014 S. Karger AG, Basel.
    Dementia and Geriatric Cognitive Disorders 02/2014; 38(1-2):16-30. · 2.79 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Precise detection and quantification of white matter hyperintensities (WMH) observed in T2-weighted Fluid Attenuated Inversion Recovery (FLAIR) Magnetic Resonance Images (MRI) is of substantial interest in aging, and age-related neurological disorders such as Alzheimer's disease (AD). This is mainly because WMH may reflect co-morbid neural injury or cerebral vascular disease burden. WMH in the older population may be small, diffuse, and irregular in shape, and sufficiently heterogeneous within and across subjects. Here, we pose hyperintensity detection as a supervised inference problem and adapt two learning models, specifically, Support Vector Machines and Random Forests, for this task. Using texture features engineered by texton filter banks, we provide a suite of effective segmentation methods for this problem. Through extensive evaluations on healthy middle-aged and older adults who vary in AD risk, we show that our methods are reliable and robust in segmenting hyperintense regions. A measure of hyperintensity accumulation, referred to as normalized effective WMH volume, is shown to be associated with dementia in older adults and parental family history in cognitively normal subjects. We provide an open source library for hyperintensity detection and accumulation (interfaced with existing neuroimaging tools), that can be adapted for segmentation problems in other neuroimaging studies. Hum Brain Mapp, 2014. © 2014 Wiley Periodicals, Inc.
    Human Brain Mapping 02/2014; · 6.92 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Canonical correlation analysis (CCA) is a widely used statistical technique to capture correlations between two sets of multi-variate random variables and has found a multitude of applications in computer vision, medical imaging and machine learning. The classical formulation assumes that the data live in a pair of vector spaces which makes its use in certain important scientific domains problematic. For instance, the set of symmetric positive definite matrices (SPD), rotations and probability distributions, all belong to certain curved Riemannian manifolds where vector-space operations are in general not applicable. Analyzing the space of such data via the classical versions of inference models is rather sub-optimal. But perhaps more importantly, since the algorithms do not respect the underlying geometry of the data space, it is hard to provide statistical guarantees (if any) on the results. Using the space of SPD matrices as a concrete example, this paper gives a principled generalization of the well known CCA to the Riemannian setting. Our CCA algorithm operates on the product Riemannian manifold representing SPD matrix-valued fields to identify meaningful statistical relationships on the product Riemannian manifold. As a proof of principle, we present results on an Alzheimer's disease (AD) study where the analysis task involves identifying correlations across diffusion tensor images (DTI) and Cauchy deformation tensor fields derived from T1-weighted magnetic resonance (MR) images.
    01/2014; 8690:251-267.
  • [Show abstract] [Hide abstract]
    ABSTRACT: There is growing body of research devoted to designing imaging-based biomarkers that identify Alzheimer's disease (AD) in its prodromal stage using statistical machine learning methods. Recently several authors investigated how clinical trials for AD can be made more efficient (i.e., smaller sample size) using predictive measures from such classification methods. In this paper, we explain why predictive measures given by such SVM type objectives may be less than ideal for use in the setting described above. We give a solution based on a novel deep learning model, randomized denoising autoencoders (rDA), which regresses on training labels y while also accounting for the variance, a property which is very useful for clinical trial design. Our results give strong improvements in sample size estimates over strategies based on multi-kernel learning. Also, rDA predictions appear to more accurately correlate to stages of disease. Separately, our formulation empirically shows how deep architectures can be applied in the large d, small n regime--the default situation in medical imaging. This result is of independent interest.
    01/2014; 17(Pt 2):470-8.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Some cognitively healthy individuals develop brain amyloid accumulation, suggestive of incipient Alzheimer's disease (AD), but the effect of amyloid on other potentially informative imaging modalities, such as Diffusion Tensor Imaging (DTI), in characterizing brain changes in preclinical AD requires further exploration. In this study, a sample (N = 139, mean age 60.6, range 46 to 71) from the Wisconsin Registry for Alzheimer's Prevention (WRAP), a cohort enriched for AD risk factors, was recruited for a multimodal imaging investigation that included DTI and [C-11]Pittsburgh Compound B (PiB) positron emission tomography (PET). Participants were grouped as amyloid positive (Aβ+), amyloid indeterminate (Aβi), or amyloid negative (Aβ-) based on the amount and pattern of amyloid deposition. Regional voxel-wise analyses of four DTI metrics, fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (Da), and radial diffusivity (Dr), were performed based on amyloid grouping. Three regions of interest (ROIs), the cingulum adjacent to the corpus callosum, hippocampal cingulum, and lateral fornix, were selected based on their involvement in the early stages of AD. Voxel-wise analysis revealed higher FA among Aβ+ compared to Aβ- in all three ROIs and in Aβi compared to Aβ- in the cingulum adjacent to the corpus callosum. Follow-up exploratory whole-brain analyses were consistent with the ROI findings, revealing multiple regions where higher FA was associated with greater amyloid. Lower fronto-lateral gray matter MD was associated with higher amyloid burden. Further investigation showed a negative correlation between MD and PiB signal, suggesting that Aβ accumulation impairs diffusion. Interestingly, these findings in a largely presymptomatic sample are in contradistinction to relationships reported in the literature in symptomatic disease stages of Mild Cognitive Impairment and AD, which usually show higher MD and lower FA. Together with analyses showing that cognitive function in these participants is not associated with any of the four DTI metrics, the present results suggest an early relationship between PiB and DTI, which may be a meaningful indicator of the initiating or compensatory mechanisms of AD prior to cognitive decline.
    NeuroImage: Clinical. 01/2014;

Publication Stats

4k Citations
925.87 Total Impact Points

Institutions

  • 2006–2014
    • University of Wisconsin–Madison
      • • Wisconsin Alzheimer's Disease Research Center
      • • Department of Medicine
      Madison, Wisconsin, United States
  • 2009
    • Wisconsin National Primate Research Center
      Madison, Wisconsin, United States
    • St. Mary's Hospital (WI, USA)
      Madison, Wisconsin, United States
  • 2007
    • University of Toronto
      Toronto, Ontario, Canada
    • Case Western Reserve University
      • Institute of Pathology
      Cleveland, OH, United States
  • 2002–2004
    • Barrow Neurological Institute
      • Department of Neurology
      Phoenix, Arizona, United States
  • 2003
    • St. Joseph Medical Center
      Houston, Texas, United States
  • 2000
    • Dartmouth Medical School
      • Department of Psychiatry
      Hanover, New Hampshire, United States