Characterizing Functional Connectivity Differences in Aging Adults using Machine Learning on Resting State fMRI Data

Abstract

The brain at rest consists of spatially distributed but functionally connected regions, called intrinsic connectivity networks (ICNs). Resting state functional magnetic resonance imaging (rs-fMRI) has emerged as a way to characterize brain networks without confounds associated with task fMRI such as task difficulty and performance. Here we applied a Support Vector Machine (SVM) linear classifier as well as a support vector machine regressor to rs-fMRI data in order to compare age-related differences in four of the major functional brain networks: the default, cingulo-opercular, fronto-parietal, and sensorimotor. A linear SVM classifier discriminated between young and old subjects with 84% accuracy (p-value < 1 × 10−7). A linear SVR age predictor performed reasonably well in continuous age prediction (R2 = 0.419, p-value < 1 × 10−8). These findings reveal that differences in intrinsic connectivity as measured with rs-fMRI exist between subjects, and that SVM methods are capable of detecting and utilizing these differences for classification and prediction.

Prediction of brain maturity in infants using machine-learning algorithms

“…Model accuracy is improved when combining data obtained across complementary imaging modalities (Brown et al, 2012; Erus et al, 2014). The approach is well-suited to the study of increasingly diverse subject groups, ranging from infant (Pruett et al, 2015) to geriatric populations (Vergun et al, 2013). In addition, the ability to extract information regarding the relative importance of both connections and nodes via feature weights highlights changes within specific regions and/or networks attributable to age or disease.…”

“…This brain-wide approach provides a conceptual shift for the functional neuroimaging field, contextualizing results obtained through investigations focused on specific areas of the brain, while increasing the likelihood of identifying population differences by investigating much larger numbers of brain regions in comparison to currently employed univariate models. It has been employed to investigate brain-wide patterns of functional and structural connectivity, creating classifiers which differentiate groups with high accuracy and identifying connections critical for distinguishing clinical populations based upon age and diagnosis such as Alzheimer's disease, autism spectrum disorder, addiction, schizophrenia and obsessive–compulsive disorder (Brown et al, 2012; Ecker et al, 2010; Erus et al, 2014; Fair et al, 2012; Franke et al, 2012; Li et al, 2014; Magnin et al, 2009; Meier et al, 2012; Pariyadath et al, 2014; Robinson et al, 2010; Rosa et al, 2015; Shen et al, 2010; Vergun et al, 2013). Importantly, these methods can be extended to SVM regression (SVR) to enable quantitative predictions in individuals for variables such as chronological age or developmental status (Pereira et al, 2009; Smola, 2004).…”

“…Importantly, these methods can be extended to SVM regression (SVR) to enable quantitative predictions in individuals for variables such as chronological age or developmental status (Pereira et al, 2009; Smola, 2004). Recently, investigators have used this technique to develop predictive models of brain maturity in individual subjects across pediatric and adult populations (Franke et al, 2012; Satterthwaite et al, 2014; Vergun et al, 2013). …”

Transcallosal Connectivity Changes from Infancy to Late Adulthood: An Ex Vivo Diffusion Spectrum Imaging Study of Macaque Brains

“…Previous studies have demonstrated that interhemispheric connectivity is an important indicator of brain dysfunction in different stages of human lifespan (Luders et al, 2010;Ota et al, 2006). Earlier resting-state functional magnetic resonance imaging (rsfMRI) studies have also demonstrated that the interhemispheric functional connectivity changes with age in humans (Tomasi and Volkow, 2012;Vergun et al, 2013). However, most studies were based on a comparison between different age groups, or with the assumption of linear relationship with age in development or degeneration stage of the brain (Makris et al, 2007;Tomasi and Volkow, 2012;Vergun et al, 2013).…”

“…Earlier resting-state functional magnetic resonance imaging (rsfMRI) studies have also demonstrated that the interhemispheric functional connectivity changes with age in humans (Tomasi and Volkow, 2012;Vergun et al, 2013). However, most studies were based on a comparison between different age groups, or with the assumption of linear relationship with age in development or degeneration stage of the brain (Makris et al, 2007;Tomasi and Volkow, 2012;Vergun et al, 2013). Due to the limited age range in each study, the actual trend or change can be different from one study to another, depending on its specific age coverage in the entire lifespan.…”