Prediction of clinical depression scores and detection of changes in whole-brain using resting-state functional MRI data with partial least squares regression.
ABSTRACT: In diagnostic applications of statistical machine learning methods to brain imaging data, common problems include data high-dimensionality and co-linearity, which often cause over-fitting and instability. To overcome these problems, we applied partial least squares (PLS) regression to resting-state functional magnetic resonance imaging (rs-fMRI) data, creating a low-dimensional representation that relates symptoms to brain activity and that predicts clinical measures. Our experimental results, based upon data from clinically depressed patients and healthy controls, demonstrated that PLS and its kernel variants provided significantly better prediction of clinical measures than ordinary linear regression. Subsequent classification using predicted clinical scores distinguished depressed patients from healthy controls with 80% accuracy. Moreover, loading vectors for latent variables enabled us to identify brain regions relevant to depression, including the default mode network, the right superior frontal gyrus, and the superior motor area.
Project description:Supervised classification machine learning algorithms may have limitations when studying brain diseases with heterogeneous populations, as the labels might be unreliable. More exploratory approaches, such as Sparse Partial Least Squares (SPLS), may provide insights into the brain's mechanisms by finding relationships between neuroimaging and clinical/demographic data. The identification of these relationships has the potential to improve the current understanding of disease mechanisms, refine clinical assessment tools, and stratify patients. SPLS finds multivariate associative effects in the data by computing pairs of sparse weight vectors, where each pair is used to remove its corresponding associative effect from the data by matrix deflation, before computing additional pairs.We propose a novel SPLS framework which selects the adequate number of voxels and clinical variables to describe each associative effect, and tests their reliability by fitting the model to different splits of the data. As a proof of concept, the approach was applied to find associations between grey matter probability maps and individual items of the Mini-Mental State Examination (MMSE) in a clinical sample with various degrees of dementia.The framework found two statistically significant associative effects between subsets of brain voxels and subsets of the questions/tasks.SPLS was compared with its non-sparse version (PLS). The use of projection deflation versus a classical PLS deflation was also tested in both PLS and SPLS.SPLS outperformed PLS, finding statistically significant effects and providing higher correlation values in hold-out data. Moreover, projection deflation provided better results.
Project description:PURPOSE: Epilepsy is a common chronic neurological disorder. We aim to investigate the underlying mechanism of epilepsy with partial least squares- (PLS-) based gene expression analysis, which is more sensitive than routine variance/regression analysis. METHODS: Two microarray data sets were downloaded from the Gene Expression Omnibus (GEO) database. PLS analysis was used to identify differentially expressed genes. Gene ontology and network analysis were also implemented. RESULTS: A total of 752 genes were identified to be differentially expressed, including 575 depressed and 177 overexpressed genes in patients. For GO enrichment analysis, except for processes related to the nervous system, we also identified overrepresentation of dysregulated genes in angiogenesis. Network analysis revealed two hub genes, CUL3 and EP300, which may serve as potential targets in further therapeutic studies. CONCLUSION: Our results here may provide new understanding for the underlying mechanisms of epilepsy pathogenesis and will offer potential targets for producing new treatments.
Project description:Magnetic resonance imaging measures have been proposed as objective markers to study upper motor neuron loss in motor neuron disorders. Cross-sectional studies have identified imaging differences between groups of healthy controls and patients with amyotrophic lateral sclerosis (ALS) or primary lateral sclerosis (PLS) that correlate with disease severity, but it is not known whether imaging measures change as disease progresses. Additionally, whether imaging measures change in a similar fashion with disease progression in PLS and ALS is unclear. To address these questions, clinical and imaging evaluations were first carried out in a prospective cross-sectional study of 23 ALS and 22 PLS patients with similar motor impairment and 19 age-matched healthy controls. Clinical evaluations consisted of a neurological examination, the ALS Functional rating scale-revised, and measures of finger tapping, gait, and timed speech. Age and ALSFRS score were not different, but PLS patients had longer duration of symptoms. Imaging measures examined were cortical thickness, regional brain volumes, and diffusion tensor imaging of the corticospinal tract and callosum. Imaging measures that differed from controls in a cross-sectional vertex-wise analysis were used as regions of interest for longitudinal analysis, which was carried out in 9 of the ALS patients (interval 1.26 ± 0.72 years) and 12 PLS patients (interval 2.08 ± 0.93 years). In the cross-sectional study both groups had areas of cortical thinning, which was more extensive in motor regions in PLS patients. At follow-up, clinical measures declined more in ALS than PLS patients. Cortical thinning and grey matter volume loss of the precentral gyri progressed over the follow-up interval. Fractional anisotropy of the corticospinal tracts remained stable, but the cross-sectional area declined in ALS patients. Changes in clinical measures correlated with changes in precentral cortical thickness and grey matter volume. The rate of cortical thinning was greater in ALS patients with shorter disease durations, suggesting that thickness decreases in a non-linear fashion. Thus, cortical thickness changes are a potential imaging marker for disease progression in individual patients, but the magnitude of change likely depends on disease duration and progression rate. Differences between PLS and ALS patients in the magnitude of thinning in cross-sectional studies are likely to reflect longer disease duration. We conclude that there is an evolution of structural imaging changes with disease progression in motor neuron disorders. Some changes, such as diffusion properties of the corticospinal tract, occur early while cortical thinning and volume loss occur later.
Project description:Partial least squares (PLS) is a well known dimension reduction method which has been recently adapted for high dimensional classification problems in genome biology. We develop sparse versions of the recently proposed two PLS-based classification methods using sparse partial least squares (SPLS). These sparse versions aim to achieve variable selection and dimension reduction simultaneously. We consider both binary and multicategory classification. We provide analytical and simulation-based insights about the variable selection properties of these approaches and benchmark them on well known publicly available datasets that involve tumor classification with high dimensional gene expression data. We show that incorporation of SPLS into a generalized linear model (GLM) framework provides higher sensitivity in variable selection for multicategory classification with unbalanced sample sizes between classes. As the sample size increases, the two-stage approach provides comparable sensitivity with better specificity in variable selection. In binary classification and multicategory classification with balanced sample sizes, the two-stage approach provides comparable variable selection and prediction accuracy as the GLM version and is computationally more efficient.
Project description:In this article, we evaluate Parallel Level Sets (PLS) and Bowsher's method as segmentation-free anatomical priors for regularized brain positron emission tomography (PET) reconstruction. We derive the proximity operators for two PLS priors and use the EM-TV algorithm in combination with the first order primal-dual algorithm by Chambolle and Pock to solve the non-smooth optimization problem for PET reconstruction with PLS regularization. In addition, we compare the performance of two PLS versions against the symmetric and asymmetric Bowsher priors with quadratic and relative difference penalty function. For this aim, we first evaluate reconstructions of 30 noise realizations of simulated PET data derived from a real hybrid positron emission tomography/magnetic resonance imaging (PET/MR) acquisition in terms of regional bias and noise. Second, we evaluate reconstructions of a real brain PET/MR data set acquired on a GE Signa time-of-flight PET/MR in a similar way. The reconstructions of simulated and real 3D PET/MR data show that all priors were superior to post-smoothed maximum likelihood expectation maximization with ordered subsets (OSEM) in terms of bias-noise characteristics in different regions of interest where the PET uptake follows anatomical boundaries. Our implementation of the asymmetric Bowsher prior showed slightly superior performance compared with the two versions of PLS and the symmetric Bowsher prior. At very high regularization weights, all investigated anatomical priors suffer from the transfer of non-shared gradients.
Project description:OBJECTIVES To assess cognitive impairment and depression in aging former professional football (National Football League [NFL]) players and to identify neuroimaging correlates of these dysfunctions. DESIGN We compared former NFL players with cognitive impairment and depression, cognitively normal retired players who were not depressed, and matched healthy control subjects. SETTING Research center in the North Texas region of the United States. PATIENTS Cross-sectional sample of former NFL players with and without a history of concussion recruited from the North Texas region and age-, education-, and IQ-matched controls. Thirty-four retired NFL players (mean age, 61.8 years) underwent neurological and neuropsychological assessment. A subset of 26 players also underwent detailed neuroimaging; imaging data in this subset were compared with imaging data acquired in 26 healthy matched controls. MAIN OUTCOME MEASURES Neuropsychological measures, clinical diagnoses of depression, neuroimaging mea-sures of white matter pathology, and a measure of cerebral blood flow. RESULTS Of the 34 former NFL players, 20 were cognitively normal. Four were diagnosed as having a fixed cognitive deficit; 8, mild cognitive impairment; 2, dementia; and 8, depression. Of the subgroup in whom neuroimaging data were acquired, cognitively impaired participants showed the greatest deficits on tests of naming, word finding, and visual/verbal episodic memory. We found significant differences in white matter abnormalities in cognitively impaired and depressed retired players compared with their respective controls. Regional blood flow differences in the cognitively impaired group (left temporal pole, inferior parietal lobule, and superior temporal gyrus) corresponded to regions associated with impaired neurocognitive performance (problems with memory, naming, and word finding). CONCLUSIONS Cognitive deficits and depression appear to be more common in aging former NFL players compared with healthy controls. These deficits are correlated with white matter abnormalities and changes in regional cerebral blood flow.
Project description:Birth weight, length and circumferences of the head, chest and arm are key measures of newborn size and health in developing countries. We assessed maternal socio-demographic factors associated with multiple measures of newborn size in a large rural population in Bangladesh using partial least squares (PLS) regression method. PLS regression, combining features from principal component analysis and multiple linear regression, is a multivariate technique with an ability to handle multicollinearity while simultaneously handling multiple dependent variables. We analyzed maternal and infant data from singletons (n = 14,506) born during a double-masked, cluster-randomized, placebo-controlled maternal vitamin A or ?-carotene supplementation trial in rural northwest Bangladesh. PLS regression results identified numerous maternal factors (parity, age, early pregnancy MUAC, living standard index, years of education, number of antenatal care visits, preterm delivery and infant sex) significantly (p<0.001) associated with newborn size. Among them, preterm delivery had the largest negative influence on newborn size (Standardized ? = -0.29 - -0.19; p<0.001). Scatter plots of the scores of first two PLS components also revealed an interaction between newborn sex and preterm delivery on birth size. PLS regression was found to be more parsimonious than both ordinary least squares regression and principal component regression. It also provided more stable estimates than the ordinary least squares regression and provided the effect measure of the covariates with greater accuracy as it accounts for the correlation among the covariates and outcomes. Therefore, PLS regression is recommended when either there are multiple outcome measurements in the same study, or the covariates are correlated, or both situations exist in a dataset.
Project description:Neuroanatomic features associated with antidepressant treatment outcomes in older depressed individuals are not well established. This study used diffusion tensor imaging to examine frontal white matter structure in depressed subjects undergoing a 12-week trial of sertraline. We hypothesized that remission would be associated with higher frontal anisotropy measures, and failure to remit with lower anisotropy.74 subjects with Major Depressive Disorder and age 60 years or older were enrolled in a twelve-week open-label trial of sertraline and completed clinical assessments and 1.5T magnetic resonance brain imaging. The apparent diffusion coefficient (ADC) and fractional anisotropy (FA) were measured in regions of interest placed in the white matter of the dorsolateral prefrontal cortex, anterior cingulate cortex, and corpus callosum. Differences in ADC and FA values between subjects who did and did not remit to treatment over the study period were assessed using generalized estimating equations, controlling for age, sex, medical comorbidity and baseline depression severity.Subjects who did not remit to sertraline exhibited higher FA values in the superior frontal gyri and anterior cingulate cortices bilaterally. There were no statistically significant associations between ADC measures and remission.Failure to remit to sertraline is associated with higher frontal FA values. Functional imaging studies demonstrate that depression is characterized by functional disconnection between frontal and limbic regions. Those individuals where this disconnection is related to structural changes as detected by DTI may be more likely to respond to antidepressants.ClinicalTrials.gov NCT00339066.
Project description:Glia activation is thought to contribute to neuronal damage in several neurodegenerative diseases based on preclinical and human post-mortem studies, but its role in primary lateral sclerosis (PLS) is unknown.To localize and measure glia activation in people with PLS compared to healthy controls (HC).Ten participants with PLS and ten age-matched HCs underwent simultaneous magnetic resonance (MR) and proton emission tomography (PET). The radiotracer [11C]-PBR28 was used to obtain PET-based measures of 18 kDa translocator protein (TSPO) expression, a marker of activated glial cells. MR techniques included a structural sequence to measure cortical thickness and diffusion tensor imaging (DTI) to assess white matter integrity.PET data showed increased [11C]-PBR28 uptake in anatomically-relevant motor regions which co-localized with areas of regional gray matter atrophy and decreased subcortical fractional anisotropy.This study supports a link between glia activation and neuronal degeneration in PLS, and suggests that these disease mechanisms can be measured in vivo in PLS. Future studies are needed to determine the longitudinal changes of these imaging measures and to clarify if MR-PET with [11C]-PBR28 can be used as a biomarker for drug development in the context of clinical trials for PLS.
Project description:OBJECTIVE:To characterize [11 C]-PBR28 brain uptake using positron emission tomography (PET) in people with amyotrophic lateral sclerosis (ALS) and primary lateral sclerosis (PLS). We have previously shown increased [11 C]-PBR28 uptake in the precentral gyrus in a small group of ALS patients. Herein, we confirm our initial finding, study the longitudinal changes, and characterize the gray versus white matter distribution of [11 C]-PBR28 uptake in a larger cohort of patients with ALS and PLS. METHODS:Eighty-five participants including 53 with ALS, 11 with PLS, and 21 healthy controls underwent integrated [11 C]-PBR28 PET-magnetic resonance brain imaging. Patients were clinically assessed using the Upper Motor Neuron Burden (UMNB) and the Amyotrophic Lateral Sclerosis Functional Rating Scale-Revised (ALSFRS-R). [11 C]-PBR28 uptake was quantified as standardized uptake value ratio (SUVR) and compared between groups. Cortical thickness and fractional anisotropy were compared between groups and correlated with SUVR and the clinical data. [11 C]-PBR28 uptake and ALSFRS-R were compared longitudinally over 6 months in 10 ALS individuals. RESULTS:Whole brain voxelwise, surface-based, and region of interest analyses revealed increased [11 C]-PBR28 uptake in the precentral and paracentral gyri in ALS, and in the subcortical white matter for the same regions in PLS, compared to controls. The increase in [11 C]-PBR28 uptake colocalized and correlated with cortical thinning, reduced fractional anisotropy, and increased mean diffusivity, and correlated with higher UMNB score. No significant changes were detected in [11 C]-PBR28 uptake over 6 months despite clinical progression. INTERPRETATION:Glial activation measured by in vivo [11 C]-PBR28 PET is increased in pathologically relevant regions in people with ALS and correlates with clinical measures. Ann Neurol 2018;83:1186-1197.