Comparison of CPU and GPU bayesian estimates of fibre orientations from diffusion MRI.

BACKGROUND: The correct estimation of fibre orientations is a crucial step for reconstructing human brain tracts. Bayesian Estimation of Diffusion Parameters Obtained using Sampling Techniques (bedpostx) is able to estimate several fibre orientations and their diffusion parameters per voxel using Markov Chain Monte Carlo (MCMC) in a whole brain diffusion MRI data, and it is capable of running on GPUs, achieving speed-up of over 100 times compared to CPUs. However, few studies have looked at whether the results from the CPU and GPU algorithms differ. In this study, we compared CPU and GPU bedpostx outputs by running multiple trials of both algorithms on the same whole brain diffusion data and compared each distribution of output using Kolmogorov-Smirnov tests. RESULTS: We show that distributions of fibre fraction parameters and principal diffusion direction angles from bedpostx and bedpostx_gpu display few statistically significant differences in shape and are localized sparsely throughout the whole brain. Average output differences are small in magnitude compared to underlying uncertainty. CONCLUSIONS: Despite small amount of differences in output between CPU and GPU bedpostx algorithms, results are comparable given the difference in operation order and library usage between CPU and GPU bedpostx.

Prenatal antidepressant exposure and sex differences in neonatal corpus callosum microstructure.

Prenatal exposure to selective serotonin reuptake inhibitor (SSRI) antidepressants may influence white matter (WM) development, as previous studies report widespread microstructural alterations and reduced interhemispheric connectivity in SSRI-exposed infants. In rodents, perinatal SSRIs had sex-specific disruptions in corpus callosum (CC) axon architecture and connectivity; yet it is unknown whether SSRI-related brain outcomes in humans are sex specific. In this study, the neonate CC was selected as a region-of-interest to investigate whether prenatal SSRI exposure has sex-specific effects on early WM microstructure. On postnatal day 7, diffusion tensor imaging was used to assess WM microstructure in SSRI-exposed (n = 24; 12 male) and nonexposed (n = 48; 28 male) term-born neonates. Fractional anisotropy was extracted from CC voxels and a multivariate discriminant analysis was used to identify latent patterns differing between neonates grouped by SSRI-exposure and sex. Analysis revealed localized variations in CC fractional anisotropy that significantly discriminated neonate groups and correctly predicted group membership with an 82% accuracy. Such effects were identified across three dimensions, representing sex differences in SSRI-exposed neonates (genu, splenium), SSRI-related effects independent of sex (genu-to-rostral body), and sex differences in nonexposed neonates (isthmus-splenium, posterior midbody). Our findings suggest that CC microstructure may have a sex-specific, localized, developmental sensitivity to prenatal SSRI exposure.

Warped rhythms: Epileptic activity during critical periods disrupts the development of neural networks for human communication.

It is well established that temporal lobe epilepsy-the most common and well-studied form of epilepsy-can impair communication by disrupting social-emotional and language functions. In pediatric epilepsy, where seizures co-occur with the development of critical brain networks, age of onset matters: The earlier in life seizures begin, the worse the disruption in network establishment, resulting in academic hardship and social isolation. Yet, little is known about the processes by which epileptic activity disrupts developing human brain networks. Here we take a synthetic perspective-reviewing a range of research spanning studies on molecular and oscillatory processes to those on the development of large-scale functional networks-in support of a novel model of how such networks can be disrupted by epilepsy. We seek to bridge the gap between research on molecular processes, on the development of human brain circuitry, and on clinical outcomes to propose a model of how epileptic activity disrupts brain development.

Advanced neuroimaging: A window into the neural correlates of fetal programming related to prenatal exposure to maternal depression and SSRIs.

Fetal programming is a conceptual framework whereby the in utero environment shapes the offspring's neurodevelopment. Maternal depression and treatment with selective serotonin reuptake inhibitor (SSRI) antidepressants during pregnancy are common prenatal exposures that affect critical early life developmental programming processes. Prenatal depression and SSRIs both have been reported to increase the risks for preterm birth, low birth weight, and are associated with behavioral disturbances across the early life span. However, not all exposures lead to adverse developmental outcomes and distinguishing how each exposure contributes to variations in development remains challenging. Advances in neuroimaging, using MR and EEG, offer novel insights into central processes that might reveal the neural correlates of fetal programming. This review focuses on emerging findings from neuroimaging studies reflecting early brain functional and structural development associated with prenatal exposure to maternal depression and SSRI antidepressants. Suggestions for future research directions that use neuroimaging as a tool to advancing our understanding of the early origins of developmental plasticity are offered.

Hub distribution of the brain functional networks of newborns prenatally exposed to maternal depression and SSRI antidepressants.

BACKGROUND: Prenatal maternal depression (PMD) and selective serotonin reuptake inhibitor (SSRI) antidepressants are associated with increased developmental risk in infants. Reports suggest that PMD is associated with hyperconnectivity of the insula and the amygdala, while SSRI exposure is associated with hyperconnectivity of the auditory network in the infant brain. However, associations between functional brain organization and PMD and/or SSRI exposure are not well understood. METHODS: We examined the relation between PMD or SSRI exposure and neonatal brain functional organization. Infants of control (n = 17), depressed SSRI-treated (n = 20) and depressed-only (HAM-D ≥ 8) (n = 16) women, underwent resting-state functional magnetic resonance imaging at postnatal Day 6. At 6 months, temperament was assessed using Infant Behavioral Questionnaire (IBQ). We applied GTA and partial least square regression (PLSR) to the resting-state time series to assess group differences in modularity, and connector and provincial hubs. RESULTS: Modularity was similar across all groups. The depressed-only group showed higher connector hub values in the left anterior cingulate, insula, and caudate as well as higher provincial hub values in the amygdala compared to the control group. The SSRI group showed higher provincial hub values in Heschl's gyrus relative to the depressed-only group. PLSR showed that newborns' hub values predicted 10% of the variability in infant temperament at 6 months, suggesting different developmental patterns between groups. CONCLUSIONS: Prenatal exposures to maternal depression and SSRIs have differential impacts on neonatal functional brain organization. Hub values at 6 days predict variance in temperament between infant groups at 6 months of age.

Alterations in Resting-State Networks Following In Utero Selective Serotonin Reuptake Inhibitor Exposure in the Neonatal Brain.

BACKGROUND: Selective serotonin reuptake inhibitors (SSRIs) are commonly used to treat depression during pregnancy. SSRIs cross the placenta, inhibit serotonin reuptake, and thereby are thought to alter central fetal serotonin signaling. Both prenatal maternal mood disturbances and in utero SSRI exposure have been associated with altered fetal and infant behavior. Resting-state functional magnetic resonance imaging has identified resting-state networks (RSNs) in newborns, reflecting functional capacity of auditory and visual networks and providing opportunities to examine early experiences effects on neurodevelopment. We sought to examine the effect of in utero SSRI exposure on neonatal RSN functional organization. We hypothesized that prenatal SSRI exposure would be associated with alterations in neonatal RSNs compared with healthy control infants and infants exposed to mothers with depression. METHODS: Clinician-rated Hamilton Depression Rating Scale and self-reported Pregnancy Experiences Scale were completed during the third trimester. Control (n = 17), maternal depression-exposed (Hamilton Depression Rating Scale ≥8 without SSRI exposure, n = 16), and SSRI-exposed (n = 20) 6-day-old neonates underwent resting-state functional magnetic resonance imaging. Independent component analysis was used as a data-driven approach to extract 22 RSNs. RESULTS: SSRI-exposed neonates had higher connectivity in a putative auditory RSN compared with depressed-only (p = .01) and control (p = .02) infants (corrected for multiple comparisons), controlling for sex, age at the magnetic resonance imaging, and Pregnancy Experiences Scale score. CONCLUSIONS: Hyperconnectivity in auditory RSN in neonates with in utero SSRI exposure relative to neonates of depressed but not pharmacologically treated mothers and control infants may offer an insight into the functional organization origins of shifts in language perception and altered language development, previously reported in infants and children with prenatal SSRI exposure.

Learning, neural plasticity and sensitive periods: implications for language acquisition, music training and transfer across the lifespan.

Sensitive periods in human development have often been proposed to explain age-related differences in the attainment of a number of skills, such as a second language (L2) and musical expertise. It is difficult to reconcile the negative consequence this traditional view entails for learning after a sensitive period with our current understanding of the brain's ability for experience-dependent plasticity across the lifespan. What is needed is a better understanding of the mechanisms underlying auditory learning and plasticity at different points in development. Drawing on research in language development and music training, this review examines not only what we learn and when we learn it, but also how learning occurs at different ages. First, we discuss differences in the mechanism of learning and plasticity during and after a sensitive period by examining how language exposure versus training forms language-specific phonetic representations in infants and adult L2 learners, respectively. Second, we examine the impact of musical training that begins at different ages on behavioral and neural indices of auditory and motor processing as well as sensorimotor integration. Third, we examine the extent to which childhood training in one auditory domain can enhance processing in another domain via the transfer of learning between shared neuro-cognitive systems. Specifically, we review evidence for a potential bi-directional transfer of skills between music and language by examining how speaking a tonal language may enhance music processing and, conversely, how early music training can enhance language processing. We conclude with a discussion of the role of attention in auditory learning for learning during and after sensitive periods and outline avenues of future research.

The survey of autobiographical memory (SAM): a novel measure of trait mnemonics in everyday life.

Compared to the abundance of laboratory-based memory tasks, few measures exist to assess self-reported memory function. This need is particularly important for naturalistic mnemonic capacities, such as autobiographical memory (recall of events and facts from one's past), because it is difficult to reliably assess in the laboratory. Furthermore, naturalistic mnemonic capacities may show stable individual differences that evade the constraints of laboratory testing. The Survey of Autobiographical Memory (SAM) was designed to assess such trait mnemonics, or the dimensional characterization of self-reported mnemonic characteristics. The SAM comprises items assessing self-reported episodic autobiographical, semantic, and spatial memory, as well as future prospection. In a large sample of healthy young adults, the latent dimensional structure of the SAM was characterized with multiple correspondence analysis (MCA). This analysis revealed dimensions corresponding to general mnemonic abilities (i.e., good vs poor memory across subtypes), spatial memory, and future prospection. While episodic and semantic items did not separate in this data-driven analysis, these categories did show expected dissociations in relation to depression history and to laboratory-based measures of recollection. Remote spatial memory as assessed by the SAM showed the expected advantage for males over females. Spatial memory was also related to autobiographical memory performance. Brief versions of the SAM are provided for efficient research applications. Individual differences in memory function are likely related to other health-related factors, including personality, psychopathology, dementia risk, brain structure and function, and genotype. In conjunction with laboratory or performance based assessments, the SAM can provide a useful measure of naturalistic self-report trait mnemonics for probing these relationships.

Partial least squares methods: partial least squares correlation and partial least square regression.

Partial least square (PLS) methods (also sometimes called projection to latent structures) relate the information present in two data tables that collect measurements on the same set of observations. PLS methods proceed by deriving latent variables which are (optimal) linear combinations of the variables of a data table. When the goal is to find the shared information between two tables, the approach is equivalent to a correlation problem and the technique is then called partial least square correlation (PLSC) (also sometimes called PLS-SVD). In this case there are two sets of latent variables (one set per table), and these latent variables are required to have maximal covariance. When the goal is to predict one data table the other one, the technique is then called partial least square regression. In this case there is one set of latent variables (derived from the predictor table) and these latent variables are required to give the best possible prediction. In this paper we present and illustrate PLSC and PLSR and show how these descriptive multivariate analysis techniques can be extended to deal with inferential questions by using cross-validation techniques such as the bootstrap and permutation tests.

Analysis of regional cerebral blood flow data to discriminate among Alzheimer's disease, frontotemporal dementia, and elderly controls: a multi-block barycentric discriminant analysis (MUBADA) methodology.

We present a generalization of mean-centered partial least squares correlation called multiblock barycentric discriminant analysis (MUBADA) that integrates multiple regions of interest (ROIs) to analyze functional brain images of cerebral blood flow or metabolism obtained with SPECT or PET. To illustrate MUBADA we analyzed data from 104 participants comprising Alzheimer's disease (AD) patients, frontotemporal dementia (FTD) patients, and elderly normal controls. Brain images were analyzed via 28 ROIs (59,845 voxels) selected for clinical relevance. This is a discriminant analysis (DA) question with several blocks (one per ROI) and with more variables than observations, a configuration that precludes using DA. MUBADA revealed two factors explaining 74% and 26% of the total variance: Factor 1 isolated FTD, and Factor 2 isolated AD. A random effects model correctly classified 64% (chance = 33%) of "new" participants (p < 0.0001). MUBADA identified ROIs that best discriminated groups: ROIs separating FTD were bilateral inferior, middle frontal, left inferior, and middle temporal gyri, while ROIs separating AD were bilateral thalamus, inferior parietal gyrus, inferior temporal gyrus, left precuneus, middle frontal, and middle temporal gyri. MUBADA classified participants at levels comparable to standard methods (i.e., SVM, PCA-LDA, and PLS-DA) but provided information (e.g., discriminative ROIs and voxels) not easily accessible to these methods.

Effect of age on variability in the production of text-based global inferences.

As we age, our differences in cognitive skills become more visible, an effect especially true for memory and problem solving skills (i.e., fluid intelligence). However, by contrast with fluid intelligence, few studies have examined variability in measures that rely on one's world knowledge (i.e., crystallized intelligence). The current study investigated whether age increased the variability in text based global inference generation--a measure of crystallized intelligence. Global inference generation requires the integration of textual information and world knowledge and can be expressed as a gist or lesson. Variability in generating two global inferences for a single text was examined in young-old (62 to 69 years), middle-old (70 to 76 years) and old-old (77 to 94 years) adults. The older two groups showed greater variability, with the middle elderly group being most variable. These findings suggest that variability may be a characteristic of both fluid and crystallized intelligence in aging.

Multiple Subject Barycentric Discriminant Analysis (MUSUBADA): how to assign scans to categories without using spatial normalization.

We present a new discriminant analysis (DA) method called Multiple Subject Barycentric Discriminant Analysis (MUSUBADA) suited for analyzing fMRI data because it handles datasets with multiple participants that each provides different number of variables (i.e., voxels) that are themselves grouped into regions of interest (ROIs). Like DA, MUSUBADA (1) assigns observations to predefined categories, (2) gives factorial maps displaying observations and categories, and (3) optimally assigns observations to categories. MUSUBADA handles cases with more variables than observations and can project portions of the data table (e.g., subtables, which can represent participants or ROIs) on the factorial maps. Therefore MUSUBADA can analyze datasets with different voxel numbers per participant and, so does not require spatial normalization. MUSUBADA statistical inferences are implemented with cross-validation techniques (e.g., jackknife and bootstrap), its performance is evaluated with confusion matrices (for fixed and random models) and represented with prediction, tolerance, and confidence intervals. We present an example where we predict the image categories (houses, shoes, chairs, and human, monkey, dog, faces,) of images watched by participants whose brains were scanned. This example corresponds to a DA question in which the data table is made of subtables (one per subject) and with more variables than observations.

Partial Least Squares (PLS) methods for neuroimaging: a tutorial and review.

Partial Least Squares (PLS) methods are particularly suited to the analysis of relationships between measures of brain activity and of behavior or experimental design. In neuroimaging, PLS refers to two related methods: (1) symmetric PLS or Partial Least Squares Correlation (PLSC), and (2) asymmetric PLS or Partial Least Squares Regression (PLSR). The most popular (by far) version of PLS for neuroimaging is PLSC. It exists in several varieties based on the type of data that are related to brain activity: behavior PLSC analyzes the relationship between brain activity and behavioral data, task PLSC analyzes how brain activity relates to pre-defined categories or experimental design, seed PLSC analyzes the pattern of connectivity between brain regions, and multi-block or multi-table PLSC integrates one or more of these varieties in a common analysis. PLSR, in contrast to PLSC, is a predictive technique which, typically, predicts behavior (or design) from brain activity. For both PLS methods, statistical inferences are implemented using cross-validation techniques to identify significant patterns of voxel activation. This paper presents both PLS methods and illustrates them with small numerical examples and typical applications in neuroimaging.

A tutorial on multiblock discriminant correspondence analysis (MUDICA): a new method for analyzing discourse data from clinical populations.

PURPOSE: In communication disorders research, clinical groups are frequently described based on patterns of performance, but researchers often study only a few participants described by many quantitative and qualitative variables. These data are difficult to handle with standard inferential tools (e.g., analysis of variance or factor analysis) whose assumptions are unfit for these data. This article presents multiblock discriminant correspondence analysis (MUDICA), which is a recent method that can handle datasets not suited for standard inferential techniques. METHOD: MUDICA is illustrated with clinical data examining conversational trouble-source repair and topic maintenance in dementia of the Alzheimer's type (DAT). Seventeen DAT participant/spouse dyads (6 controls, 5 participants with early DAT, 6 participants with moderate DAT) produced spontaneous conversations analyzed for co-occurrence of trouble-source repair and topic maintenance variables. RESULTS: MUDICA found that trouble-source repair sequences and topic transitions are associated and that patterns of performance in the DAT groups differed significantly from those in the control group. CONCLUSION: MUDICA is ideally suited to analyze language and discourse data in communication disorders because it (a) can identify and predict clinical group membership based on patterns of performance, (b) can accommodate few participants and many variables, (c) can be used with categorical data, and (d) adds the rigor of inferential statistics.

How to compute reliability estimates and display confidence and tolerance intervals for pattern classifiers using the Bootstrap and 3-way multidimensional scaling (DISTATIS).

When used to analyze brain imaging data, pattern classifiers typically produce results that can be interpreted as a measure of discriminability or as a distance between some experimental categories. These results can be analyzed with techniques such as multidimensional scaling (MDS), which represent the experimental categories as points on a map. While such a map reveals the configuration of the categories, it does not provide a reliability estimate of the position of the experimental categories, and therefore cannot be used for inferential purposes. In this paper, we present a procedure that provides reliability estimates for pattern classifiers. This procedure combines bootstrap estimation (to estimate the variability of the experimental conditions) and a new 3-way extension of MDS, called DISTATIS, that can be used to integrate the distance matrices generated by the bootstrap procedure and to represent the results as MDS-like maps. Reliability estimates are expressed as (1) tolerance intervals which reflect the accuracy of the assignment of scans to experimental categories and as (2) confidence intervals which generalize standard hypothesis testing. When more than two categories are involved in the application of a pattern classifier, the use of confidence intervals for null hypothesis testing inflates Type I error. We address this problem with a Bonferonni-like correction. Our methodology is illustrated with the results of a pattern classifier described by O'Toole et al. (O'Toole, A., Jiang, F., Abdi, H., Haxby, J., 2005. Partially distributed representations of objects and faces in ventral temporal cortex. J. Cogn. Neurosci. 17, 580-590) who re-analyzed data originally collected by Haxby et al. (Haxby, J., Gobbini, M., Furey, M., Ishai, A., Schouten, J., Pietrini, P., 2001. Distributed and overlapping representation of faces and objects in ventral temporal cortex. Science 293, 2425-2430).