Significant correlation between a set of genetic polymorphisms and a functional brain network revealed by feature selection and sparse Partial Least Squares

Floch, Édith Le; Guillemot, Vincent; Frouin, Vincent; Pinel, Philippe; Lalanne, Christophe; Trinchera, Laura; Tenenhaus, Arthur; Moreno, Antonio; Zilbovicius, Mônica; Bourgeron, Thomas; Dehaene, Stanislas; Thirion, Bertrand; Poline, Jean‐Baptiste; Duchesnay, Édouard

doi:10.1016/j.neuroimage.2012.06.061

Cited by 103 publications

(117 citation statements)

References 32 publications

Supporting

Mentioning

116

Contrasting

Unclassified

Order By: Relevance

“…We note the relationship between CCA and PLS coefficients in equations (6,9) and equations (7,10), respectively. PLS does not take into account the covariations between the responses Y explicitly.…”

Section: Partial Least Squaresmentioning

confidence: 99%

“…Many variants of RRR, CCA, and PLS have been proposed, such as using an 2 regularization on X and Y [13], or using an 1 regularization on the reduced-rank coefficients [6], [7]. Here, we propose to experiment low-rank regularizations of the whitening matrix Γ = (Y T Y) −1 .…”

Section: E Regularized Reduced-rank Modelsmentioning

confidence: 99%

“…CCA and PLS are variants of reduced-rank regression (RRR) models, a general formulation for low-rank linear models. These models have been used to link brain connectivity and behavior [4] [5], or brain and genetics [6] [7], or to characterize patients with schizophrenia [8] or Alzheimer's Disease [9]. Such studies evaluate the model accuracy by measuring correlations between the computed canonical modes.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Multi-output predictions from neuroimaging: assessing reduced-rank linear models

Rahim

Thirion

Varoquaux

2017

2017 International Workshop on Pattern Recognition in Neuroimaging (PRNI)

Self Cite

View full text Add to dashboard Cite

Abstract-Typical neuroimaging studies analyze associations between physiological or behavioral traits and brain structure or function. Some rely on predicting these scores from neuroimaging data. To explain association between brain features and multiple traits, reduced-rank regression (RRR) models are often used, such as canonical correlation analysis (CCA) and partial least squares (PLS). These methods estimate latent variables, or canonical modes, that maximize the covariations between neuroimaging features and behavioral scores. Here, we investigate theoretically and empirically the extent to which reduced-rank models predict out-of-sample clinical scores from functional connectivity. Experiments on a schizophrenia dataset show that i) significant correlations between canonical modes do not necessarily mean accurate generalization on unseen data, and ii) better accuracy is achieved when taking into account regularized covariance between scores.

show abstract

Section: Partial Least Squaresmentioning

confidence: 99%

Section: E Regularized Reduced-rank Modelsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi-output predictions from neuroimaging: assessing reduced-rank linear models

Rahim

Thirion

Varoquaux

2017

2017 International Workshop on Pattern Recognition in Neuroimaging (PRNI)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Another approach is to find linear combinations of features and responses that best correlate with each other using partial least square (PLS) or canonical correlation analysis (CCA), which can be cast as a reduced rank regression (RRR) problem [4]. In limited sample settings, especially when the number of features exceeds the number of samples, sparse variants of PLS, CCA, and RRR are often used [5], but these sparse variants in their raw forms suffer the same limitation as SMR in terms of false positives not being controlled.…”

Section: Introductionmentioning

confidence: 99%

Bootstrapped Permutation Test for Multiresponse Inference on Brain Behavior Associations

Poline

Thirion

et al. 2015

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

“…Activity at each time point was normalized to the first TR (labelled TR 0 in the figures). For all analyses, we ran 1000 permutations (Le Floch et al, 2012;McIntosh et al, 1996) to determine significant LVs at p < 0.001. In addition, we ran 100 bootstraps, estimating the standard errors of the salience for each voxel in order to assess the reliability and robustness of each voxel's contribution to a pattern of brain activity.…”

Section: Prs-fmri Analysismentioning

confidence: 99%

The neuroimaging and genetics of emotion perception in schizophrenia

Dzafic¹

View full text Add to dashboard Cite

Impaired emotion perception is a core deficit in schizophrenia (SCZ) that is robustly associated with poor functional outcome. Studies to date have failed to delineate the critical cognitive, neural and genetic factors that underlie 'naturalistic' emotion perception deficits in SCZ. Recently it has been proposed that in SCZ reduced precision in prior expectations impairs perception. Normal perception relies on predictive processing from precise prior expectations, which prioritise attention to context-relevant stimuli. However, no studies to date have explored the aberrancy of prior expectations during naturalistic emotion perception in SCZ and the underlying abnormal neural circuitry or involvement of SCZ genes. The present thesis aims to develop a novel dynamic emotion perception (DEP) task, in order to examine (1) the influence of prior expectations on naturalistic emotion perception in healthy individuals, and (2) emotion perception impairment in SCZ by investigating aberrancy in prior expectations, neural circuitry, and genetics.The aim of Study 1 was to create a novel functional magnetic resonance imaging (fMRI) paradigm to examine the mechanisms of prior expectations and delineate the spatiotemporal activity of brain regions that contribute to emotion perception. The newly developed and validated fMRI DEP task has improved ecological validity, using dynamic, audio-visual videos, instead of static pictures as stimuli. The improved method provided new behavioural and neuroimaging insights into the influence of prior expectations on naturalistic emotion perception. Prior expectations facilitated emotion perception after repeated exposure and learning over the course of the experiment, engaging prefrontal regions and right amygdala. In contrast, emotions incongruent with prior expectations engaged neural correlates of automatic change detection, which in a complex dynamic environment allows for adaptive behaviours in potentially advantageous or threatening situations.In Study 2, the DEP task was utilised to delineate, in the healthy brain, the functional networks that subserve rapid recognition of threatening emotions in situations either congruent or incongruent with prior expectations. The results showed that fast detection of threat congruent with prior expectations engaged a right amygdala subcortical network, which has been previously implicated in fear conditioning and coarse processing of emotion. In contrast, there was a trend association between fast detection of threat incongruent with prior expectations and activity in inferior frontoparietal network connected to the right temporoparietal junction; this has been previously implicated in rapid reorienting of attention to Finally, in Study 4, the association between SCZ polygenic risk score (PRS) and brain activity during the DEP task in healthy adults was investigated in order to assess emotion perception as an endophenotype for SCZ. The results demonstrate that PRS is associated with higher recruitment of regions involved in emotion regulat...

show abstract

Significant correlation between a set of genetic polymorphisms and a functional brain network revealed by feature selection and sparse Partial Least Squares

Cited by 103 publications

References 32 publications

Multi-output predictions from neuroimaging: assessing reduced-rank linear models

Multi-output predictions from neuroimaging: assessing reduced-rank linear models

Bootstrapped Permutation Test for Multiresponse Inference on Brain Behavior Associations

The neuroimaging and genetics of emotion perception in schizophrenia

Contact Info

Product

Resources

About