Sparse logistic regression for whole brain classification of fMRI data

By Srikanth Ryali, Kaustubh Supekar, Daniel A. Abrams, Vinod Menon. NeuroImage, 2010

302 citations recorded by the source corpus.

Multivariate pattern recognition methods are increasingly being used to identify multiregional brain activity patterns that collectively discriminate one cognitive condition or experimental group from another, using fMRI data. The performance of these methods is often limited because the number of regions considered in the analysis of fMRI data is large compared to the number of observations (trials or participants). Existing methods that aim to tackle this dimensionality problem are less than optimal because they either over-fit the data or are computationally intractable. Here, we describe a novel method based on logistic regression using a combination of L1 and L2 norm regularization that more accurately estimates discriminative brain regions across multiple conditions or groups. The L1 norm, computed using a fast estimation procedure, ensures a fast, sparse and generalizable solution; the L2 norm ensures that correlated brain regions are included in the resulting solution, a critical aspect of fMRI data analysis often overlooked by existing methods. We first evaluate the performance of our method on simulated data and then examine its effectiveness in discriminating between well-matched music and speech stimuli. We also compared our procedures with other methods which use either L1-norm regularization alone or support vector machine based feature elimination. On simulated data, our methods performed significantly better than existing methods across a wide-range of contrast-to-noise ratios and feature prevalence rates. On experimental fMRI data, our methods were more effective in selectively isolating a distributed fronto-temporal network that distinguished between brain regions known to be involved in speech and music processing. These findings suggest that our method is not only computationally efficient, but it also achieves the twin objectives of identifying relevant discriminative brain regions and accurately classifying fMRI data.

Subfields: Cognitive Neuroscience

View Paper: https://www.semanticscholar.org/paper/20b4832e5f6c04816c6083e0e6ee402917f00c5d

How researchers see this paper

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

In Cognitive Neuroscience

Citations in the corpus

Penalized Least Squares Regression Methods and Applications to Neuroimaging (2010, unknown): “Cross-validation is perhaps the most popular way for parameter tuning in the literature see, for instance, Tibshirani (1996); Zou and Hastie (2005); Gunn et al. (2002); Carroll et al. (2009); Ryali et al. (2010), although it may be computationally expensive for largescale data.”
Model sparsity and brain pattern interpretation of classification models in neuroimaging (2012, methodology): “Methods that impose sparsity in the feature/voxel dimension have been introduced for neuroimaging applications [26–29].”
Analyzing Neuroimaging Data Through Recurrent Deep Learning Models (2018, unknown): “Over the recent years, the whole-brain lasso, as well as closely related decoding approaches (e.g., McIntosh and Lobaugh, 2004; Ryali et al., 2010; Gramfort et al., 2013), have found widespread application throughout the neuroscience literature (e.”
Sparse Overlapping Sets Lasso for Multitask Learning and its Application to fMRI Analysis (2013, methodology): “While logistic Glasso [19], lasso [15], and the elastic net penalty [14] have been applied to neuroimaging data, these methods do not exclusively take into account both the common macrostructure and the differences in microstructure across brains.”
Analytic estimation of statistical significance maps for support vector machine based multi-variate image analysis and classification (2013, unknown): “For instance, methods described in Batmanghelich et al. (2012), Gaonkar et al. (2011), Ryali et al. (2010) and Sabuncu and Van Leemput (2011)) attempt to apply sparsity to make interpretations/inferences.”
The structural-functional connectome and the default mode network of the human brain (2014, unknown): “Similar approaches have been used on fMRI-data (Huang et al., 2010; Ryali et al., 2010; Varoquaux and Craddock, 2013; Varoquaux et al., 2010; Wee et al., in press) and in other domains, e.g. to analyze genetic datasets (Stifanelli et al., 2011; Witten and Tibshirani, 2009).”
Estimation of functional connectivity in fMRI data using stability selection-based sparse partial correlation with elastic net penalty (2011, unknown): “, 2005; Shevade and Keerthi, 2003) and elastic net optimization problems (Friedman et al., 2010; Ryali et al., 2010; Tibshirani et al., 2010).”
Does feature selection improve classification accuracy? Impact of sample size and feature selection on classification using anatomical magnetic resonance images (2012, background): “These include lasso regression (Tibshirani, 1996), elastic net (Zou and Hastie, 2005), automatic relevance determination (Chu et al., 2010; MacKay, 1995; Tipping, 2001), and other sparse models (Carroll et al., 2009; Grosenick et al., 2008; Ryali et al., 2010; Yamashita et al., 2008).”
Assessing and tuning brain decoders: Cross‐validation, caveats, and guidelines (2016, methodology): “For sparse models, using the l1 penalty, sparsity is often seen as a means to select relevant voxels for prediction (Carroll et al., 2009; Ryali et al., 2010).”
Parallel Random Coordinate Descent Method for Composite Minimization: Convergence Analysis and Error Bounds (2016, background): “One application is found in data mining or machine learning [27], where we must solve a sparse problem:”
Machine learning framework for early MRI-based Alzheimer’s conversion prediction in MCI subjects (2014, unknown): “…(called regularizationpath ),and has been used widely in previousworks (Huttunen et al., 2012, 2013; Ryali et al., 2010) for the multi-voxel pattern analyses of functional neuroimaging data as well as for AD related studies using structural MRI data…”
Tensor Regression with Applications in Neuroimaging Data Analysis (2012, background): “…data to predict cognitive outcome, to classify disease status, and to identify brain regions associated with clinical response, which have received increasing interest in recent years (Lindquist, 2008; Lazar, 2008; Martino et al., 2008; Friston, 2009; Ryali et al., 2010; Hinrichs et al., 2009).”