Author ' s personal copy Partial Least Squares ( PLS ) methods for neuroimaging : A tutorial and review

By Anjali Krishnan, Lynne J. Williams, Anthony Randal McIntosh, Hervé Abdi. 2011

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Visual Feature Attribution Using Wasserstein GANs (2017, methodology): “The most common approach is to use regression analysis or machine learning tools to generate population average maps, which highlights features that are salient across the population [3, 32, 61, 16, 39, 45, 14].”
Mapping neurotransmitter systems to the structural and functional organization of the human neocortex (2021, methodology): “PLS is an unsupervised multivariate statistical technique that decomposes the two datasets into orthogonal sets of latent variables with maximum covariance 132 .”
Tensor decomposition of EEG signals: a brief review. (2015, unknown): “For brain research, the standard PLS seeks the common latent vectors from brain data X ∈ (cid:6) I × L and behavior data Y ∈ (cid:6) I × N and the constraint is that those latent vectors mostly explain the covariance between the brain and the behavior data (Krishnan et al., 2011).”
Bilingualism: Consequences for Mind and Brain (2012, methodology): “The fMRI data were analyzed using a multivariate statistical technique for neuroimaging data (Partial Least Squares; for review, see [105]) to identify integrated neural networks.”
Principal components of functional connectivity: A new approach to study dynamic brain connectivity during rest (2013, methodology): “This approach is similar to how significant latent dimensions are estimated in PLS (Krishnan et al., 2011; McIntosh et al., 1996) and in other fields studying time series (e.”
Models of the Aging Brain Structure and Individual Decline (2012, methodology): “Multivariate generative model: predictive patterns of decline The partial least squares correlation (PLSC) technique was initially introduced into the field of functional neuroimaging to relate brain activity data to experimental design matrices and it evolved to a powerful tool for various applicat”
A tutorial review: Metabolomics and partial least squares-discriminant analysis--a marriage of convenience or a shotgun wedding. (2015, background): “that can be used to identify the most important variables [78–80].”
Reconstructing subject‐specific effect maps (2017, unknown): “n used at the population level, where at each measurement, maps indicate statistical relationship between the condition and measurement across the entire population. Either a group analysis technique [2,22,40] or a machine learning based predictive model [1,14,26,31,13] is used to compare two cohort”
Imagine all the people: how the brain creates and uses personality models to predict behavior. (2014, methodology): “The functional connectivity analysis was performed using “seed” PLS, a multivariate task-related functional connectivity analysis technique used to investigate the relationship between the activity of a seed region and the activity in the rest of the brain (McIntosh 1999; Krishnan et al. 2011).”
Relevant Feature Set Estimation with a Knock-out Strategy and Random Forests (2015, unknown): “Bootstrapping (Krishnan et al., 2011) can provide a partial solution to this by quantifying the stability of weights, but it does not mitigate the relative assignment problem.”
Network-Level Structure-Function Relationships in Human Neocortex (2016, methodology): “PLS is a multivariate statistical technique used to relate two sets of variables to each other (Wold 1966; McIntosh and Lobaugh 2004; Abdi 2010; Krishnan et al. 2011; McIntosh and Mišic ́ 2013).”
Dynamic connectivity regression: Determining state-related changes in brain connectivity (2012, background): “In contrast, partial least squares (PLS; Krishnan et al., 2010; McIntosh et al., 1996) attempts to identify common patterns between brain activity and behavioral measures or experimental design.”
Structural foundations of resting-state and task-based functional connectivity in the human brain (2013, methodology): “Alternatively, a generalized linear model (GLM) can be used in combination with techniques such as partial least squares (PLS) or psychophysiological interactions (PPI) [45, 46] to identify task-related activations across cortical and subcortical structures.”
Constructing Subject- and Disease-Specific Effect Maps: Application to Neurodegenerative Diseases (2016, background): “Inference-based models, whether univariate [1] or multivariate [7,8], detect condition-related diﬀerences in image-based measurements between two groups of subjects, one composed of patients and the other composed of controls.”