Simulating functional interactions in the brain: a model for examining correlations between regional cerebral metabolic rates

Predicting functional connectivity from structural connectivity via computational models using MRI: an extensive comparison study

The relationship between structural connectivity (SC) and functional connectivity (FC) in the human brain can be studied using magnetic resonance imaging (MRI). However many of the underlying physiological mechanisms and parameters cannot be directly observed with MRI. This limitation has motivated the recent use of various computational models meant to bridge the gap. However their absolute and relative explanatory power and the properties that actually drive that power remain insufficiently characterized. We performed an extensive comparison of seven mainstream computational models predicting FC from SC. We investigated the extent to which simulated FC could predict empirical FC. We also applied graph theory to the entire set of simulated and empirical FCs in order to further characterize the relationships between the models and the MRI data. The comparison was performed at three different spatial scales. We found that (i) there were significant effects of scale and model on predictive power; (ii) among all models, the simplest model, the simultaneous autoregressive (SAR) model, was found to consistently perform better than the other models; (iii) the SAR also appeared more 'central' from a graph theory perspective; and (iv) empirical FC only appeared weakly correlated with simulated FCs, and was featured as 'peripheral' in the graph analysis. We conclude that the substantial differences existing between these computational models have little impact on their predictive power for FC and that their capacity to predict FC from SC appears to be both moderate and essentially underlined by a simple core linear process embodied by the SAR model.

Relating structural and functional connectivity in MRI: a simple model for a complex brain

Advances in magnetic resonance imaging (MRI) allow to gain critical insight into the structure of neural networks and their functional dynamics. To relate structural connectivity [as quantified by diffusion-weighted imaging (DWI) tractography] and functional connectivity [as obtained from functional MRI (fMRI)], increasing emphasis has been put on computational models of brain activity. In the present study, we use structural equation modeling (SEM) with structural connectivity to predict functional connectivity. The resulting model takes the simple form of a spatial simultaneous autoregressive model (sSAR), whose parameters can be estimated in a Bayesian framework. On synthetic data, results showed very good accuracy and reliability of the inference process. On real data, we found that the sSAR performed significantly better than two other reference models as well as than structural connectivity alone, but that the Bayesian procedure did not bring significant improvement in fit compared to two simpler approaches. Nonetheless, we also found that the values of the region-specific parameters inferred using Bayesian inference differed significantly across resting-state networks. These results demonstrate 1) that a simple abstract model is able to perform better that more complex models based on more realistic assumptions, 2) that the parameters of the sSAR can be estimated and can potentially be used as biomarkers, but also 3) that the sSAR, while being the best-performing model, is at best still a very crude model of the relationship between structure and function in MRI.

Pathway-based genome-wide association analysis of coronary heart disease identifies biologically important gene sets

Genome-wide association (GWA) studies of complex diseases including coronary heart disease (CHD) challenge investigators attempting to identify relevant genetic variants among hundreds of thousands of markers being tested. A selection strategy based purely on statistical significance will result in many false negative findings after adjustment for multiple testing. Thus, an integrated analysis using information from the learned genetic pathways, molecular functions, and biological processes is desirable. In this study, we applied a customized method, variable set enrichment analysis (VSEA), to the Framingham Heart Study data (404,467 variants, n=6421) to evaluate enrichment of genetic association in 1395 gene sets for their contribution to CHD. We identified 25 gene sets with nominal P<0.01; at least four sets are previously known for their roles in CHD: vascular genesis (GO:0001570), fatty-acid biosynthetic process (GO:0006633), fatty-acid metabolic process (GO:0006631), and glycerolipid metabolic process (GO:0046486). Although the four gene sets include 170 genes, only three of the genes contain a variant ranked among the top 100 in single-variant association tests of the 404,467 variants tested. Significant enrichment for novel gene sets less known for their importance to CHD were also identified: Rac 1 cell-motility signaling pathway (h_rac1 Pathway, P<0.001) and sulfur amino-acid metabolic process (GO:0000096, P<0.001). In summary, we showed that the pathway-based VSEA can help prioritize association signals in GWA studies by identifying biologically plausible targets for downstream searches of genetic variants associated with CHD.

Neural substrates for the motivational regulation of motor recovery after spinal-cord injury

It is believed that depression impedes and motivation enhances functional recovery after neuronal damage such as spinal-cord injury and stroke. However, the neuronal substrate underlying such psychological effects on functional recovery remains unclear. A longitudinal study of brain activation in the non-human primate model of partial spinal-cord injury using positron emission tomography (PET) revealed a contribution of the primary motor cortex (M1) to the recovery of finger dexterity through the rehabilitative training. Here, we show that activity of the ventral striatum, including the nucleus accumbens (NAc), which plays a critical role in processing of motivation, increased and its functional connectivity with M1 emerged and was progressively strengthened during the recovery. In addition, functional connectivities among M1, the ventral striatum and other structures belonging to neural circuits for processing motivation, such as the orbitofrontal cortex, anterior cingulate cortex and pedunculopontine tegmental nucleus were also strengthened during the recovery. These results give clues to the neuronal substrate for motivational regulation of motor learning required for functional recovery after spinal-cord injury.

Effective and structural connectivity in the human auditory cortex

Language processing involves multiple neuronal structures in the human auditory cortex. Although a variety of neuroimaging and mapping techniques have been implemented to better understand language processing at the level of the auditory cortex, much is unknown regarding how and by what pathways these structures interact during essential tasks such as sentence comprehension. In this study, the effective and structural connectivity at the level of the auditory cortex were investigated. First, blood oxygenation level-dependent (BOLD) responses were measured with time-resolved functional magnetic resonance imaging (fMRI) during audition of short sentences. Once BOLD activation maps were obtained, the effective connectivity between primary auditory cortex and the surrounding auditory regions on the supratemporal plane and superior temporal gyrus (STG) were investigated using Granger causality mapping (GCM). Effective connectivity was observed between the primary auditory cortex and (1) the lateral planum polare and anterior STG, and (2) the lateral planum temporale and posterior STG. By using diffusion tensor probabilistic mapping (DTPM), rostral and caudal fiber pathways were detected between regions depicting effective connectivity. The effective and structural connectivity results of the present study provide further insight as to how auditory stimuli (i.e., human language) is processed at the level of the auditory cortex. Furthermore, combining BOLD fMRI-based GCM and DTPM analysis could provide a novel means to study effective and structural connectivity not only in the auditory cortex, but also in other cortical regions.

Metabolic thalamocortical correlations during a verbal learning task and their comparison with correlations among regional volumes

Methods based on the analysis of metabolic and volumetric interregional correlations have been used in neuroimaging research, yet metabolic and volumetric interregional correlations for identical regions of interest have never been compared in the same group of subjects. Magnetic resonance and [18F]-fluorodeoxyglucose positron emission tomography brain images were acquired in 59 healthy subject. Correlation matrices for relative glucose metabolic rates during a verbal learning task and for relative gray matter volumes were compiled between the manually traced mediodorsal, centromedian, and pulvinar nuclei of the thalamus and 39 cortical Brodmann's areas. Metabolic correlations between the cortex and these thalamic nuclei followed the known patterns of anatomical connectivity in non-human primates. Intercorrelations of the mediodorsal nucleus were widespread with the prefrontal cortex (9 out of 10 Brodmann's areas in the left hemisphere) and temporal lobe (10 out of 11 Brodmann's areas in the left hemisphere) while the pulvinar correlated only with the parietal and occipital cortical areas. Different correlation patterns were observed for the regional gray matter volumes whereby only the pulvinar yielded extensive cortical intercorrelations, primarily with the occipital, parietal, anterior cingulate, and orbitofrontal areas in the right hemisphere. Metabolic thalamocortical correlations were much more extensive for the mediodorsal and centromedian nuclei whereas structural correlations were more extensive for the pulvinar. Therefore, metabolic and volumetric correlational methods are sensitive to different aspects of interregional relations in the brain and their comparison in the same group of subjects may render complementary and only partially overlapping connectivity information.

Sex-related differences in amygdala functional connectivity during resting conditions

Recent neuroimaging studies have established a sex-related hemispheric lateralization of amygdala involvement in memory for emotionally arousing material. Here, we examine the possibility that sex-related differences in amygdala involvement in memory for emotional material develop from differential patterns of amygdala functional connectivity evident in the resting brain. Seed voxel partial least square analyses of regional cerebral blood flow data revealed significant sex-related differences in amygdala functional connectivity during resting conditions. The right amygdala was associated with greater functional connectivity in men than in women. In contrast, the left amygdala was associated with greater functional connectivity in women than in men. Furthermore, the regions displaying stronger functional connectivity with the right amygdala in males (sensorimotor cortex, striatum, pulvinar) differed from those displaying stronger functional connectivity with the left amygdala in females (subgenual cortex, hypothalamus). These differences in functional connectivity at rest may link to sex-related differences in medical and psychiatric disorders.

Task difficulty in a simultaneous face matching task modulates activity in face fusiform area

The level of difficulty of a task can alter the neural network that activates for performance of the task. Previous studies have shown increased activation with task difficulty in the frontal lobes while the effects in the extrastriate visual areas have been unclear. We hypothesized that the face fusiform area (FFA), an area specialized for face processing, would increase activation as task difficulty increased in a face matching task. The difficulty level was increased by degrading the quality of the images. The degradation levels were 10%, 20%, 40% and 60%. Based on the correct response rate, the data were divided into a baseline level (composed of non-degraded and 10% degraded images) and a difficult level (composed of the 20%, 40% and 60% degraded images). Brain activation was measured using functional magnetic resonance imaging. The baseline face matching task activated a wide network of regions that included bilaterally the occipital, temporal and parietal lobes and the right frontal lobe. A novel behavioral finding was that task difficulty did not linearly increase with image degradation. The novel brain imaging finding was that the FFA is modulated by task difficulty and performance in the task was linearly correlated to activation in FFA. In addition, we found that activation in the dorsolateral prefrontal cortex (DLPFC) had increased activation as task difficulty increased. When adding the response time as a covariate, the differences in the DLPFC did not remain statistically significant. Increased task difficulty also led to a decrease in activation of visual areas in the extrastriate cortex. Task difficulty increased activation in the FFA to enhance the face processing and suppressed activation in visual extrastriate areas that processed low level properties of the stimuli. Task difficulty led to enhanced response in the FFA and suppressed response in other visual areas.

Mapping directed influence over the brain using Granger causality and fMRI

We propose Granger causality mapping (GCM) as an approach to explore directed influences between neuronal populations (effective connectivity) in fMRI data. The method does not rely on a priori specification of a model that contains pre-selected regions and connections between them. This distinguishes it from other fMRI effective connectivity approaches that aim at testing or contrasting specific hypotheses about neuronal interactions. Instead, GCM relies on the concept of Granger causality to define the existence and direction of influence from information in the data. Temporal precedence information is exploited to compute Granger causality maps that identify voxels that are sources or targets of directed influence for any selected region-of-interest. We investigated the method by simulations and by application to fMRI data of a complex visuomotor task. The presented exploratory approach of mapping influences between a region of interest and the rest of the brain can form a useful complement to existing models of effective connectivity.

Amygdalar interhemispheric functional connectivity differs between the non-depressed and depressed human brain

The amygdalae are important, if not critical, brain regions for many affective, attentional and memorial processes, and dysfunction of the amygdalae has been a consistent finding in the study of clinical depression. Theoretical models of the functional neuroanatomy of both normal and psychopathological affective processes which posit cortical hemispheric specialization of functions have been supported by both lesion and functional neuroimaging studies in humans. Results from human neuroimaging studies in support of amygdalar hemispheric specialization are inconsistent. However, recent results from human lesion studies are consistent with hemispheric specialization. An important, yet largely ignored, feature of the amygdalae in the primate brain--derived from both neuroanatomical and electrophysiological data--is that there are virtually no direct interhemispheric connections via the anterior commissure (AC). This feature stands in stark contrast to that of the rodent brain wherein virtually all amygdalar nuclei have direct interhemispheric connections. We propose this feature of the primate brain, in particular the human brain, is a result of influences from frontocortical hemispheric specialization which have developed over the course of primate brain evolution. Results consistent with this notion were obtained by examining the nature of human amygdalar interhemispheric connectivity using both functional magnetic resonance imaging (FMRI) and positron emission tomography (PET). We found modest evidence of amygdalar interhemispheric functional connectivity in the non-depressed brain, whereas there was strong evidence of functional connectivity in the depressed brain. We interpret and discuss the nature of this connectivity in the depressed brain in the context of dysfunctional frontocortical-amygdalar interactions which accompany clinical depression.

Independent component model for cognitive functions of multiple subjects using [15O]H2O PET images

An independent component model of multiple subjects' positron emission tomography (PET) images is proposed to explore the overall functional components involved in a task and to explain subject specific variations of metabolic activities under altered experimental conditions utilizing the Independent component analysis (ICA) concept. As PET images represent time-compressed activities of several cognitive components, we derived a mathematical model to decompose functional components from cross-sectional images based on two fundamental hypotheses: (1) all subjects share basic functional components that are common to subjects and spatially independent of each other in relation to the given experimental task, and (2) all subjects share common functional components throughout tasks which are also spatially independent. The variations of hemodynamic activities according to subjects or tasks can be explained by the variations in the usage weight of the functional components. We investigated the plausibility of the model using serial cognitive experiments of simple object perception, object recognition, two-back working memory, and divided attention of a syntactic process. We found that the independent component model satisfactorily explained the functional components involved in the task and discuss here the application of ICA in multiple subjects' PET images to explore the functional association of brain activations.

Effect of Nucleus Basalis Magnocellularis Ablation on Local Brain Glucose Utilization in the Rat: Functional Brain Reorganization

After unilateral destruction of the nucleus basalis magnocellularis (NBM) in 3-month-old rats, which reduces cholinergic inputs to the ipsilateral frontoparietal neocortex, regional cerebral metabolic rates for glucose (rCMRglc) of denervated cortex are initially reduced, but nearly normalize by 2 weeks. To examine functional reorganization of the brain after unilateral destruction of the NBM, a correlation analysis of rCMRglc was performed on two groups of 16 young rats 2 weeks after stereotaxic ablation of the right NBM with ibotenate or sham surgery. rCMRglc was measured in 117 brain regions of awake rats with the [14C]deoxyglucose method. For each region pair, a partial correlation coefficient was calculated for rCMRglc across animals. Most correlations between cholinergic nuclei of both left and right forebrain (medial septum and diagonal band) and right (66/72, mean increase 0.44) but not left (39/72) frontoparietal cortical regions were larger (P < 0.001) in lesioned rats, as were those between most frontoparietal region pairs (516/630, P < 0.001). These results suggest that, after unilateral NBM ablation, (1) functional interactions are established between the remaining cholinergic forebrain and the deafferented cortex, (2) the neocortex becomes more integrated, and (3) functional reorganization involves both cortical hemispheres. These changes do not correspond to those reported to occur in Alzheimer's disease.

Comparison of regional cerebral blood flow and glucose metabolism in the normal brain: effect of aging

The regional cerebral blood flow (rCBF) and metabolic rate for glucose (rCMRGlc) are associated with functional activity of the neural cells. The present work reports a comparison study between rCBF and rCMRGlc in a normal population as a function of age. 10 young (25.9+/-5.6 years) and 10 old (65.4+/-6.1 years) volunteers were similarly studied at rest. In each subject, rCBF and rCMRGlc were measured in sequence, during the same session. Both rCBF and rCMRGlc values were found to decrease from young (mean rCBF=43.7 ml/100 g per min; mean rCMRGlc=40.6 micromol/100 g per min) to old age (mean rCBF=37.3 ml/100 g per min; mean rCMRGlc=35.2 micromol/100 g per min), resulting in a drop over 40 years of 14.8% (0.37%/year) and 13.3% (0.34%/year), respectively. On a regional basis, the frontal and the visual cortices were observed to have, respectively, the highest and the lowest reduction in rCBF, while, for rCMRGlc, these extremes were observed in striatum and cerebellum. Despite these differences, the ratio of rCBF to rCMRGlc was found to have a similar behavior in all brain regions for young and old subjects as shown by a correlation coefficient of 88%. This comparative study indicates a decline in rCBF and rCMRGlc values and a coupling between CBF and CMRGlc as a function of age.

Neural modeling and functional brain imaging: an overview

This article gives an overview of the different functional brain imaging methods, the kinds of questions these methods try to address and some of the questions associated with functional neuroimaging data for which neural modeling must be employed to provide reasonable answers.

Towards the networks of the brain: from brain imaging to consciousness

The manner in which the brain computes in various tasks is being probed at a deep level by modern brain imaging techniques, with an increasing appreciation of the different networks being used to solve these tasks. There is simultaneously developing a neural modelling technology, which attempts to explain the underlying computations being performed by this set of networks. This paper describes results from brain imaging and how they may be related to the underlying neural networks by means of structural modelling. It thereby attempts to give an initial glimpse of the emerging picture of the functionality of brain networks. It concludes with a discussion of the role of consciousness in global processing, and how particular styles of neural processing can attain this.

Visualizing fronto-striatal circuitry and neuroleptic effects in schizophrenia

Disturbances in fronto-striatal circuitry have been postulated to be important in schizophrenia. Positron emission tomography typically shows decreased metabolic rates in these areas relative to other brain areas in schizophrenia. After treatment with typical neuroleptics, striatal metabolic rates are increased, but other brain areas tend not to show significant changes. Atypical neuroleptics less markedly affect striatal metabolic rates, but show wider cortical effects. In order to examine fronto-striatal circuitry, a technique for visualizing the correlations between metabolic rates in all brain areas was applied in 33 controls and 27 unmedicated schizophrenic patients. Correlation images revealed strong fronto-striatal connections in controls, but weak fronto-striatal links in schizophrenic patients. Changes in striatal circuits, also reflected in recent anatomical studies, may be important for understanding antipsychotic effects.

Neural modeling, functional brain imaging, and cognition

The richness and complexity of data sets acquired from PET or fMRI studies of human cognition have not been exploited until recently by computational neural-modeling methods. In this article, two neural-modeling approaches for use with functional brain imaging data are described. One, which uses structural equation modeling, estimates the functional strengths of the anatomical connections between various brain regions during specific cognitive tasks. The second employs large-scale neural modeling to relate functional neuroimaging signals in multiple, interconnected brain regions to the underlying neurobiological time-varying activities in each region. Delayed match-to-sample (visual working memory for form) tasks are used to illustrate these models.

Large-scale functional connectivity in associative learning: interrelations of the rat auditory, visual, and limbic systems

Large-scale functional connectivity in associative learning: interrelations of the rat auditory, visual, and limbic systems. J. Neurophysiol. 80: 3148-3162, 1998. Functional relations between specialized parts of the brain may be important determinants of learned behaviors. To study this, we examined the interrelations of the auditory system with several extraauditory structures in two groups of rats having different behavioral histories. Both groups were trained to associate a tone conditional stimulus (CS) with an aversive unconditional stimulus (US). For one group, a light presented with the tone predicted the absence of the US (group TL-). In the other group, the light was a neutral stimulus (group TL0). Fluorodeoxyglucose (FDG) incorporation was measured in the presence of the tone-light compound. Because the tone-light compound was physically identical for both groups, neural differences between groups reflected differences in the learned associative properties of the stimuli. Covariances of FDG uptake in the auditory system and extraauditory structures were examined using partial least squares. Three strong covariance or functional connectivity patterns were identified. The first pattern mainly reflected similarities between groups, with strong interrelations between the subcortical auditory system and the thalamocortical visual system, cerebellum, deep cerebellar nuclei, and midline thalamus. This pattern of interactions may represent part of a common circuit for relaying the associative value of the tone CS to the cerebellum and the midline thalamus. The external nucleus of the inferior colliculus and medial division of the medial geniculate nucleus were associated more strongly with this pattern for group TL-, which was interpreted as representing the change of the associative value of the tone by the light, mediated through extraauditory influences on these two regions. A second pattern involved midbrain auditory regions, superior colliculus, zona incerta, and subiculum and was stronger for group TL0. The relations between midbrain structures may represent the excitatory conditioned response (CR) evoked by the tone in this group. The final pattern was strongest in group TL- and involved interrelations of the thalamocortical auditory system with hippocampus, basolateral amygdala, and hypothalamus. This pattern may represent the learned inhibition of the CR to the tone in the presence of the light. These findings are consistent with behavioral studies suggesting that at least two types of associations are formed during associative learning. One is the sensory relation of the stimuli and another is the relation between the CS and the affective components of the US. These behavioral associations are mapped to the patterns of functional connectivity between auditory and extraauditory regions.

Elucidating dynamic brain interactions with across-subjects correlational analyses of positron emission tomographic data: the functional connectivity of the amygdala and orbitofrontal cortex during olfactory tasks

Covariance analyses of positron emission tomography (PET) data are used increasingly to elucidate the functional connectivity between brain regions during different cognitive tasks. Functional connectivity may be estimated by examining the covariance between regions over time or across subjects. In functional brain-mapping studies, across-subjects covariance matrices derived from within-task (nonsubtracted) and between-task (subtracted) data characterize different, complementary aspects of functional interactions. The authors study amygdala-orbitofrontal interactions during three task conditions (aversive olfaction, odor detection, and resting with eyes closed) to illustrate the strengths and limitations of across-subjects covariance analyses based on subtracted and nonsubtracted data. This example underscores the dynamic nature of connectivity between the amygdalae and orbitofrontal cortices and highlights the importance of including data from resting conditions in covariance analyses.

Images of the working brain: understanding human brain function with positron emission tomography

In the past 15 years positron emission tomography (PET) has become a settled method of imaging the functioning human brain, both in normal volunteers and in patients with various disorders. Much of the work on sensory systems has been on the visual system, a conveniently studied and very important part of the brain. The motor system in health and disease has attracted as much interest. In this short review I will explain some of the technical aspects of PET, and illustrate some of the research that has been done on visual and motor brain function in the human.

Commentary and opinion: I. Principal component analysis, variance partitioning, and "functional connectivity"

We briefly review the need for careful study of "variance partitioning" and "optimal model selection" in functional positron emission tomography (PET) data analysis, emphasizing the use of principal component analysis (PCA) and the importance of data analytic techniques that allow for heterogeneous spatial covariance structures. Using an [15O]water dataset, we demonstrate that--even after data processing--the intrasubject signal component of primary interest in baseline activation studies constitutes a very small fraction of the intersubject variance. This small intrasubject variance component is subtly but significantly changed by using analysis of covariance instead of scaled subprofile model processing before applying PCA. Finally, we argue that the concept of "functional connectivity" should be interpreted very generally until the relative roles of inter- and intrasubject variability in both disease and normal PET datasets are better understood.

Reduction of functional neuronal connectivity in long-term treated hypertension

BACKGROUND AND PURPOSE

Anatomic imaging of patients with chronic well-treated hypertension has demonstrated dilatation of the lateral cerebral ventricles and left brain atrophy, whereas positron emission tomography has shown only subtle reductions in regional cerebral metabolic rates for glucose in some subcortical nuclei. To further explore the implications of the imaging changes, an analytic technique designed to determine functional neuronal connectivity between regions of interest (ROIs) was applied to the data on regional cerebral metabolic rates for glucose to determine if and where in the brain reduction of functional neuronal connectivity occurred.

METHODS

Glucose metabolism was measured by positron emission tomography in 17 older men (age, 68 +/- 8 years) with well-controlled, noncomplicated hypertension of at least 10 years' duration and in 25 age- and sex-matched healthy control subjects. A significant correlation difference analysis was performed to determine which ROI pairs had reduced correlation coefficients (reduced functional neuronal connectivity). The vascular pattern of the reduction was determined after allocating the ROIs to their appropriate vascular territories.

RESULTS

Compared with the control subjects, hypertensive patients had reduced correlation coefficients in cortical territories of the internal carotid arteries but not of the vertebrobasilar arteries. The border zone supplied by the middle and anterior cerebral arteries was most affected.

CONCLUSIONS

The border zone between the anterior and middle cerebral arteries is vulnerable to ischemia from carotid pathology, systemic hypotension, or both. We hypothesize that although these hypertensive patients were "well controlled" and had normal neuropsychological tests, they may have experienced ischemia severe enough to cause border zone reduction of functional neuronal connectivity as a result of carotid pathology, antihypertensive medications, hypotensive episodes with a right-shifted autoregulation curve, or other factors in isolation or combination.

Individual differences in cerebral metabolic patterns during pharmacotherapy in obsessive-compulsive disorder: a multiple regression/discriminant analysis of positron emission tomographic data

A multiple regression/discriminant analysis of positron emission tomographic cerebral metabolic (rCMRglc) data in 10 obsessive-compulsive disorder (OCD) patients before and during pharmacotherapy was carried out to see if rCMRglc interdependencies distinguished OCD patients from controls. Before therapy, a discriminant function reflecting parietal, sensorimotor, and midbrain rCMRglc interdependencies correctly classified eight (80%) of the 10 patients as OCD; after therapy, six (70%) were classified as controls, most of whom were responders. Before therapy, rCMRglc interdependencies involving basal ganglia, thalamus, limbic, and sensory and association cortical regions distinguished 67% of patients who clinically responded to drug (RESP, n = 6) and 75% of patients who did not (NRESP, n = 4) from controls. After therapy, all RESP were classified as controls; classification of NRESP remained unchanged. The results suggest the conjunctive utility of this method to assess individual differences in rCMRglc during pharmacotherapy, and to explore the neurobiology of OCD.

Early detection of Alzheimer's disease: a statistical approach using positron emission tomographic data

Correlational analysis of regional cerebral glucose metabolism (rCMRglc) obtained by high-resolution positron emission tomography (PET) has demonstrated reduced neocortical rCMRglc interactions in mildly/moderately demented patients with probable Alzheimer's disease (AD). Thus, identification of individual differences in patterns of rCMRglc interactions may be important for the early detection of AD, particularly among individuals at greater risk for developing AD (e.g., those with a family history of AD). Recently, a statistical procedure, using multiple regression and discriminant analysis, was developed to assess individual differences in patterns of rCMRglc interdependencies. We applied this new statistical procedure to resting rCMRglc PET data from mildly/moderately demented patients with probable AD and age/sex-matched controls. The aims of the study were to identify a discriminant function that would (a) distinguish patients from controls and (b) identify an AD pattern in an individual at risk for AD with isolated memory impairment whose initial PET scan showed minor abnormalities, but whose second scan showed parietal hypometabolism, coincident with further cognitive decline. Two discriminant functions, reflecting interactions involving regions most involved in reduced correlations in probable AD, correctly classified 87% of the patients and controls, and successfully identified the first scan of the at-risk individual as AD (probability > 0.70). The results suggest that this statistical approach may be useful for the early detection of AD.

Parametric in vivo brain imaging during activation to examine pathological mechanisms of functional failure in Alzheimer disease

Alzheimer disease (AD) patients demonstrate reduced "resting state" regional cerebral metabolic rates for glucose (rCMRglc) and reduced regional cerebral blood flow (rCBF) in cortical association areas, early and throughout the course of disease. In this paper, we hypothesize that parametric cognitive or passive stimulation, during in vivo brain imaging, can be used to elucidate the pathological basis of these flow and metabolic abnormalities in individual AD patients. Experimental data suggest that sigmoidal relations (nonliner monotonically increasing relations reaching a horizontal asymptote) exist in the normal brain between rCBF or rCMRglc, and a function F(D,P) of task difficulty D (intensity, duration, pattern complexity) and subject performance P (reaction time, accuracy, effort, attention). Pathological mechanism I, under some conditions reversible and involving neural (including synaptic) element dropout or modification with retained capacity for full activation at high values of F(D,P), is expected early in AD and should shift the rising phase of the normal sigmoidal curve to the right. Observed rCBF responses in a face-matching task in mildly-moderately demented AD patients are consistent with mechanism I. Pathological mechanisms II and III, both irreversible and involving neural element dropout with loss of capacity for maximum activation, should alter the sigmoidal brain response at all values of F(D,P), and are expected late in disease. Our hypothesis predicts that activation paradigms with a wide range of F(D,P) values could help to distinguish among the reversible and irreversible pathological mechanisms in AD, and to evaluate drug action on these mechanisms.

Functional connectivity: the principal-component analysis of large (PET) data sets

The distributed brain systems associated with performance of a verbal fluency task were identified in a nondirected correlational analysis of neurophysiological data obtained with positron tomography. This analysis used a recursive principal-component analysis developed specifically for large data sets. This analysis is interpreted in terms of functional connectivity, defined as the temporal correlation of a neurophysiological index measured in different brain areas. The results suggest that the variance in neurophysiological measurements, introduced experimentally, was accounted for by two independent principal components. The first, and considerably larger, highlighted an intentional brain system seen in previous studies of verbal fluency. The second identified a distributed brain system including the anterior cingulate and Wernicke's area that reflected monotonic time effects. We propose that this system has an attentional bias.

Functional Associations among Human Posterior Extrastriate Brain Regions during Object and Spatial Vision

Primate extrastriate visual cortex is organized into an occipitotemporal pathway for object vision and an occipitoparietal pathway for spatial vision. Correlations between normalized regional cerebral blood flow values (regional divided by global flows), obtained using H2 15O and positron emission tomography, were used to examine functional associations among posterior brain regions for these two pathways in 17 young men during performance of a face matching task and a dot-location matching task. During face matching, there was a significant correlation in the right hemisphere between an extrastriate occipital region that was equally activated during both the face matching and dot-location matching tasks and a region in inferior occipitotemporal cortex that was activated more during the face matching task. The corresponding correlation in the left hemisphere was not significantly different from zero. Significant intrahemispheric correlations among posterior regions were observed more often for the right than for the left hemisphere. During dot-location matching, many significant correlations were found among posterior regions in both hemispheres, but significant correlations between specific regions in occipital and parietal cortex shown to be reliably activated during this spatial vision test were found only in the right cerebral hemisphere. These results suggest that (1) correlational analysis of normalized rCBF can detect functional interactions between components of proposed brain circuits, and (2) face and dot-location matching depend primarily on functional interactions between posterior cortical areas in the right cerebral hemisphere. At the same time, left hemisphere cerebral processing may contribute more to dot-location matching than to face matching.

Interregional correlations of resting cerebral glucose metabolism in old and young women

A correlational analysis of normalized (regional to whole-brain) regional cerebral metabolic rates for glucose obtained in the 'resting' state (eyes covered, ears plugged) using [18F]fluorodeoxyglucose, demonstrated differences between old and young women in patterns of functional associations. Fifteen healthy young (age less than 40 years) and 17 healthy old women (age greater than 64 years) were scanned with a Scanditronix PC1024-7B tomograph. The brain was divided into 65 regions of interest. The old women had fewer and less positive correlations between pairs of metabolic ratios in the frontal and parietal cortices. The results suggest an age-related reduction in frontal and parietal functional interactions in the 'resting' state that is consistent with a prior correlation analysis using a low resolution ECAT II scanner on young and old men. Reduced functional interactions may reflect age-related cognitive changes.

Gender differences in correlations of cerebral glucose metabolic rates in young normal adults

Correlational analysis of normalized regional cerebral metabolic data obtained by positron emission tomography, in healthy subjects in the 'resting' state (eyes covered, ears plugged) using [18F]2-fluoro-2-deoxy-D-glucose, demonstrated gender differences in patterns of functional associations. Fifteen women and 18 men (less than 40 yr) were scanned with a Scanditronix PC1024-7B tomograph. The brain was divided into 65 regions of interest (ROIs). There were no differences between men and women in global or regional metabolic rates, or in metabolic right-left asymmetries. Although the total number of significant correlations did not differ between men and women, patterns differed: female correlations rF were most positive than male correlations rM more often than rM greater than rF; and most rF greater than rM cases involved left frontal and sensorimotor ROIs, whereas most rM greater than rF cases involved right sensorimotor and occipital ROIs. The findings demonstrate gender differences in the pattern of functional neocortical interactions at the 'resting' state.

Cerebral metabolic pattern in obsessive-compulsive disorder: altered intercorrelations between regional rates of glucose utilization

Correlations between normalized regional cerebral metabolic rates for glucose, determined by positron emission tomography with 18F-2-fluoro-2-deoxy-D-glucose, were used to investigate functional associations between pairs of brain regions in 18 adult patients with primary obsessive-compulsive disorder (OCD) of childhood-onset, as compared with 18 age- and sex-matched control subjects. The number of correlations that differed significantly between the two groups exceeded chance, although as many of these correlations were larger in the OCD group relative to controls as were smaller. The two regions that had the largest number of correlations that differed significantly between groups were a left hemisphere superior parietal region and the left hemisphere anterior medial temporal area (which includes principally the amygdala). Correlations involving the caudate nuclei did not differ between the two groups for the most part. Anterior limbic/paralimbic regions had correlations in the OCD group that were significantly larger with frontal areas than in controls, and correlations that were significantly smaller with posterior brain regions. This pattern was especially pronounced for the left hemisphere anterior medial temporal region. These results suggest that the correlation pattern in OCD is not characterized by an overall loss of functional integration but, rather, by functional reorganization.

Functional interactions in the brain: use of correlations between regional metabolic rates

Correlation coefficients between pairs of regional metabolic rates have been used to study patterns of functional associations among brain regions in humans and animals. An overview is provided concerning the additional information about brain functioning this type of analysis yields. A computer simulation model is presented for the purpose of giving a partial validation for correlational analysis. The model generates a set of simulated metabolic data upon which correlational analysis is performed. Because the underlying pattern of functional couplings in the model is known, these simulations demonstrate that the correlation coefficient between normalized metabolic rates is proportional to the strength of the functional coupling constant and that correlational analysis yields information on regional involvement in neural systems not evident in the pattern of absolute metabolic values.