Wayne Lee

Optimizing Preprocessing and Analysis Pipelines for Single-Subject FMRI. I. Standard Temporal Motion and Physiological Noise Correction Methods

Subject‐specific artifacts caused by head motion and physiological noise are major confounds in BOLD fMRI analyses. However, there is little consensus on the optimal choice of data preprocessing steps to minimize these effects. To evaluate the effects of various preprocessing strategies, we present a framework which comprises a combination of (1) nonparametric testing including reproducibility and prediction metrics of the data‐driven NPAIRS framework (Strother et al. [2002]: NeuroImage 15:747–771), and (2) intersubject comparison of SPM effects, using DISTATIS (a three‐way version of metric multidimensional scaling (Abdi et al. [2009]: NeuroImage 45:89–95). It is shown that the quality of brain activation maps may be significantly limited by sub‐optimal choices of data preprocessing steps (or “pipeline”) in a clinical task‐design, an fMRI adaptation of the widely used Trail‐Making Test. The relative importance of motion correction, physiological noise correction, motion parameter regression, and temporal detrending were examined for fMRI data acquired in young, healthy adults. Analysis performance and the quality of activation maps were evaluated based on Penalized Discriminant Analysis (PDA). The relative importance of different preprocessing steps was assessed by (1) a nonparametric Friedman rank test for fixed sets of preprocessing steps, applied to all subjects; and (2) evaluating pipelines chosen specifically for each subject. Results demonstrate that preprocessing choices have significant, but subject‐dependant effects, and that individually‐optimized pipelines may significantly improve the reproducibility of fMRI results over fixed pipelines. This was demonstrated by the detection of a significant interaction with motion parameter regression and physiological noise correction, even though the range of subject head motion was small across the group (≪ 1 voxel). Optimizing pipelines on an individual‐subject basis also revealed brain activation patterns either weak or absent under fixed pipelines, which has implications for the overall interpretation of fMRI data, and the relative importance of preprocessing methods. Hum Brain Mapp, 2012.

Impaired development of intrinsic connectivity networks in children with medically intractable localization-related epilepsy

Typical childhood development is characterized by the emergence of intrinsic connectivity networks (ICNs) by way of internetwork segregation and intranetwork integration. The impact of childhood epilepsy on the maturation of ICNs is, however, poorly understood. The developmental trajectory of ICNs in 26 children (8–17 years) with localization‐related epilepsy and 28 propensity‐score matched controls was evaluated using graph theoretical analysis of whole brain connectomes from resting‐state functional magnetic resonance imaging (fMRI) data. Children with epilepsy demonstrated impaired development of regional hubs in nodes of the salience and default mode networks (DMN). Seed‐based connectivity and hierarchical clustering analysis revealed significantly decreased intranetwork connections, and greater internetwork connectivity in children with epilepsy compared to controls. Significant interactions were identified between epilepsy duration and the expected developmental trajectory of ICNs, indicating that prolonged epilepsy may cause progressive alternations in large‐scale networks throughout childhood. DMN integration was also associated with better working memory, whereas internetwork segregation was associated with higher full‐scale intelligence quotient scores. Furthermore, subgroup analyses revealed the thalamus, hippocampus, and caudate were weaker hubs in children with secondarily generalized seizures, relative to other patient subgroups. Our findings underscore that epilepsy interferes with the developmental trajectory of brain networks underlying cognition, providing evidence supporting the early treatment of affected children. Hum Brain Mapp 35:5686–5700, 2014.

PHYCAA: data-driven measurement and removal of physiological noise in BOLD fMRI.

The effects of physiological noise may significantly limit the reproducibility and accuracy of BOLD fMRI. However, physiological noise evidences a complex, undersampled temporal structure and is often non-orthogonal relative to the neuronally-linked BOLD response, which presents a significant challenge for identifying and removing such artifact. This paper presents a multivariate, data-driven method for the characterization and removal of physiological noise in fMRI data, termed PHYCAA (PHYsiological correction using Canonical Autocorrelation Analysis). The method identifies high frequency, autocorrelated physiological noise sources with reproducible spatial structure, using an adaptation of Canonical Correlation Analysis performed in a split-half resampling framework. The technique is able to identify physiological effects with vascular-linked spatial structure, and an intrinsic dimensionality that is task- and subject-dependent. We also demonstrate that increasing dimensionality of such physiological noise is correlated with increasing variability in externally-measured respiratory and cardiac processes. Using PHYCAA as a denoising technique significantly improves simulated signal detection with physiological noise, and real data-driven model prediction and reproducibility, for both block and event-related task designs. This is demonstrated compared to no physiological noise correction, and to the widely used RETROICOR (Glover et al., 2000) physiological denoising algorithm, which uses externally measured cardiac and respiration signals.

Atypical Functional Brain Connectivity during Rest in Autism Spectrum Disorders

Connectivity atypicalities in autism spectrum disorders (ASD) have been extensively proposed. The default mode network (DMN) is critical in this study, given the insight it provides for long‐distance connectivity, and the importance of regions in this network for introspection and social emotion processing, areas affected in ASD. However, study of this network has largely been limited to adults; research earlier in development is lacking. The objective of this study was to examine DMN connectivity in children/adolescents with ASD.

The neural correlates of visuo-spatial working memory in children with autism spectrum disorder: effects of cognitive load

BackgroundResearch on the neural bases of cognitive deficits in autism spectrum disorder (ASD) has shown that working memory (WM) difficulties are associated with abnormalities in the prefrontal cortex. However, cognitive load impacts these findings, and no studies have examined the relation between WM load and neural underpinnings in children with ASD. Thus, the current study determined the effects of cognitive load on WM, using a visuo-spatial WM capacity task in children with and without ASD with functional magnetic resonance imaging (fMRI).MethodsWe used fMRI and a 1-back colour matching task (CMT) task with four levels of difficulty to compare the cortical activation patterns associated with WM in children (7–13 years old) with high functioning autism (N = 19) and matched controls (N = 17) across cognitive load.ResultsPerformance on CMT was comparable between groups, with the exception of one difficulty level. Using linear trend analyses, the control group showed increasing activation as a function of difficulty level in frontal and parietal lobes, particularly between the highest difficulty levels, and decreasing activation as a function of difficulty level in the posterior cingulate and medial frontal gyri. In contrast, children with ASD showed increasing activation only in posterior brain regions and decreasing activation in the posterior cingulate and medial frontal gyri, as a function of difficulty level. Significant differences were found in the precuneus, dorsolateral prefrontal cortex and medial premotor cortex, where control children showed greater positive linear relations between cortical activity and task difficulty level, particularly at the highest difficulty levels, but children with ASD did not show these trends.ConclusionsChildren with ASD showed differences in activation in the frontal and parietal lobes—both critical substrates for visuo-spatial WM. Our data suggest that children with ASD rely mainly on posterior brain regions associated with visual and lower level processing, whereas controls showed activity in frontal lobes related to the classic WM network. Findings will help guide future work by localizing areas of vulnerability to developmental disturbances.

Exploring predictive and reproducible modeling with the single-subject FIAC dataset.

Predictive modeling of functional magnetic resonance imaging (fMRI) has the potential to expand the amount of information extracted and to enhance our understanding of brain systems by predicting brain states, rather than emphasizing the standard spatial mapping. Based on the block datasets of Functional Imaging Analysis Contest (FIAC) Subject 3, we demonstrate the potential and pitfalls of predictive modeling in fMRI analysis by investigating the performance of five models (linear discriminant analysis, logistic regression, linear support vector machine, Gaussian naive Bayes, and a variant) as a function of preprocessing steps and feature selection methods. We found that: (1) independent of the model, temporal detrending and feature selection assisted in building a more accurate predictive model; (2) the linear support vector machine and logistic regression often performed better than either of the Gaussian naive Bayes models in terms of the optimal prediction accuracy; and (3) the optimal prediction accuracy obtained in a feature space using principal components was typically lower than that obtained in a voxel space, given the same model and same preprocessing. We show that due to the existence of artifacts from different sources, high prediction accuracy alone does not guarantee that a classifier is learning a pattern of brain activity that might be usefully visualized, although cross‐validation methods do provide fairly unbiased estimates of true prediction accuracy. The trade‐off between the prediction accuracy and the reproducibility of the spatial pattern should be carefully considered in predictive modeling of fMRI. We suggest that unless the experimental goal is brain‐state classification of new scans on well‐defined spatial features, prediction alone should not be used as an optimization procedure in fMRI data analysis. Hum Brain Mapp, 2006.

Deep grey matter growth predicts neurodevelopmental outcomes in very preterm children.

We evaluated whether the volume and growth rate of critical brain structures measured by MRI in the first weeks of life following very preterm (<32/40 weeks) birth could predict subsequent neurodevelopmental outcomes at 4 years of age. A significant proportion of children born very prematurely have cognitive deficits, but these problems are often only detected at early school age. Structural T2-weighted magnetic resonance images were acquired in 96 very preterm neonates scanned within 2 weeks of birth and 70 of these at term-equivalent age. An automated 3D image analysis procedure was used to measure the volume of selected brain structures across all scans and time points. At 4 years of age, 53 children returned for neuropsychological assessments evaluating IQ, language and visual motor integration. Associations with maternal education and perinatal measures were also explored. Multiple regression analyses revealed that growth of the caudate and globus pallidus between preterm birth and term-equivalent age predicted visual motor integration scores after controlling for sex and gestational age. Further associations were found between caudate and putamen growth with IQ and language scores. Analyses at either preterm or term-equivalent age only found associations between normalized deep grey matter growth and visual motor integration scores at term-equivalent age. Maternal education levels were associated with measures of IQ and language, but not visual motor integration. Thalamic growth was additionally linked with perinatal measures and presence of white matter lesions. These results highlight deep grey matter growth rates as promising biomarkers of long-term outcomes following very preterm birth, and contribute to our understanding of the brain-behaviour relations in these children.

Visual functional magnetic resonance imaging of preterm infants

Aim  The aim of this study was to determine the feasibility of undertaking visual functional magnetic resonance imaging (fMRI) in very preterm children.

The autism puzzle: Diffuse but not pervasive neuroanatomical abnormalities in children with ASD

Autism Spectrum Disorder (ASD) is a clinically diagnosed, heterogeneous, neurodevelopmental condition, whose underlying causes have yet to be fully determined. A variety of studies have investigated either cortical, subcortical, or cerebellar anatomy in ASD, but none have conducted a complete examination of all neuroanatomical parameters on a single, large cohort. The current study provides a comprehensive examination of brain development of children with ASD between the ages of 4 and 18 years who are carefully matched for age and sex with typically developing controls at a ratio of one-to-two. Two hundred and ten magnetic resonance images were examined from 138 Control (116 males and 22 females) and 72 participants with ASD (61 males and 11 females). Cortical segmentation into 78 brain-regions and 81,924 vertices was conducted with CIVET which facilitated a region-of-interest- (ROI-) and vertex-based analysis, respectively. Volumes for the cerebellum, hippocampus, striatum, pallidum, and thalamus and many associated subregions were derived using the MAGeT Brain algorithm. The study reveals cortical, subcortical and cerebellar differences between ASD and Control group participants. Diagnosis, diagnosis-by-age, and diagnosis-by-sex interaction effects were found to significantly impact total brain volume but not total surface area or mean cortical thickness of the ASD participants. Localized (vertex-based) analysis of cortical thickness revealed no significant group differences, even when age, age-range, and sex were used as covariates. Nonetheless, the region-based cortical thickness analysis did reveal regional changes in the left orbitofrontal cortex and left posterior cingulate gyrus, both of which showed reduced age-related cortical thinning in ASD. Our finding of region-based differences without significant vertex-based results likely indicates non-focal effects spanning the entirety of these regions. The hippocampi, thalamus, and globus pallidus, were smaller in volume relative to total cerebrum in the ASD participants. Various sub-structures showed an interaction of diagnosis-by-age, diagnosis-by-sex, and diagnosis-by-age-range, in the case where age was divided into childhood (age < 12) and adolescence (12 < age < 18). This is the most comprehensive imaging-based neuro-anatomical pediatric and adolescent ASD study to date. These data highlight the neurodevelopmental differences between typically developing children and those with ASD, and support aspects of the hypothesis of abnormal neuro-developmental trajectory of the brain in ASD.

Hemodynamic effects of cholinesterase inhibition in mild Alzheimer's disease

To evaluate the spatiotemporal progression of perfusion changes in early stages of Alzheimers disease (AD), we imaged the perfusion response to pharmacological treatment in a group of mild AD patients and contrasted it to the perfusion of age‐, sex‐, and education‐matched healthy volunteers over the same time interval.