UNCORRECTED PROOF 1 Support vector machine classification and characterization of age-related 2 reorganization of functional brain networks 3 Timothy B. Q1 Meier a, ⁎, Alok S. Desphande b , Svyatoslav Vergun c , Veena A. Nair d , Jie Song e , Bharat B. Biswal f , 4 Mary E. Meyerand c, e , Rasmus M. Birn g , Vivek Prabhakaran a, d 5 a Neuroscience Training Program, University of Wisconsin-Madison, Madison, WI 53705, USA 6 b Department of Elec. and Comp. Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA 7 c Department of Medical Physics, University of Wisconsin-Madison, Madison, WI 53705, USA 8 d Department of Radiology, University of Wisconsin-Madison, Madison, WI 53792, USA 9 e Department of Biomedical Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA 10 f Department of Radiology, University of Medicine and Dentistry of New Jersey, Newark, NJ 07103, USA 11 g Department of Psychiatry, University of Wisconsin-Madison, Madison, WI 53719, USA 12 13 abstract article info 14 Article history: 15 Received 8 June 2011 16 Revised 13 December 2011 17 Accepted 14 December 2011 18 Available online xxxx 19 20 21 22 Keywords: 23 Machine learning 24 Resting-state fMRI 25 Aging 26 Reorganization 27 Most of what is known about the reorganization of functional brain networks that accompanies normal aging 28 is based on neuroimaging studies in which participants perform specific tasks. In these studies, reorganiza- 29 tion is defined by the differences in task activation between young and old adults. However, task activation 30 differences could be the result of differences in task performance, strategy, or motivation, and not necessarily 31 reflect reorganization. Resting-state fMRI provides a method of investigating functional brain networks with- 32 out such confounds. Here, a support vector machine (SVM) classifier was used in an attempt to differentiate 33 older adults from younger adults based on their resting-state functional connectivity. In addition, the infor- 34 mation used by the SVM was investigated to see what functional connections best differentiated younger 35 adult brains from older adult brains. Three separate resting-state scans from 26 younger adults (18–35 yrs) 36 and 26 older adults (55–85) were obtained from the International Consortium for Brain Mapping (ICBM) 37 dataset made publically available in the 1000 Functional Connectomes project www.nitrc.org/projects/ 38 fcon_1000. 100 seed-regions from four functional networks with 5 mm 3 radius were defined based on a 39 recent study using machine learning classifiers on adolescent brains. Time-series for every seed-region 40 were averaged and three matrices of z-transformed correlation coefficients were created for each subject cor- 41 responding to each individual's three resting-state scans. SVM was then applied using leave-one-out cross- 42 validation. The SVM classifier was 84% accurate in classifying older and younger adult brains. The majority 43 of the connections used by the classifier to distinguish subjects by age came from seed-regions belonging 44 to the sensorimotor and cingulo-opercular networks. These results suggest that age-related decreases in pos- 45 itive correlations within the cingulo-opercular and default networks, and decreases in negative correlations 46 between the default and sensorimotor networks, are the distinguishing characteristics of age-related 47 reorganization. 48 © 2011 Published by Elsevier Inc. 49 50 51 52 53 Introduction 54 Until relatively recently, most of our knowledge regarding the 55 age-related reorganization of functional networks in the human 56 brain has been based on neuroimaging studies comparing differences 57 in task activity between younger and older adults. However, follow- 58 ing an explosion of research into resting-state functional connectivity, 59 it has been proposed that all functional networks that are utilized for 60 task performance are present during rest in the form of correlations 61 between low frequency fluctuations (Biswal et al., 1995; Smith et 62 al., 2009). Resting-state fMRI allows the investigation of age-related 63 changes in functional networks without confounds of task studies, 64 such as performance, motivation, and the use of divergent strategies. 65 Relatively few studies have attempted to characterize age-related 66 reorganization of functional networks at the brain-wide level using 67 resting-state functional connectivity (Biswal et al., 2010). In one 68 such study, Meunier et al. completed a graph theoretical analysis on 69 resting-state data investigating the effects of aging on the modular 70 organization of functional networks (2009). They found that large 71 modules, or networks, observed in young healthy adults were split 72 up into smaller modules in older adults. In addition, they observed a NeuroImage xxx (2012) xxx–xxx ⁎ Corresponding author at: University of Wisconsin-Madison, Neuroscience Training Program, 1310d Wisconsin Institutes for Medical Research, 1111 Highland Ave., Madi- son, WI 53705, USA Q4 . Fax: +1 608 265 4152. E-mail address: tbmeier@wisc.edu (T.B. Meier). YNIMG-09029; No. of pages: 13; 4C: 1053-8119/$ – see front matter © 2011 Published by Elsevier Inc. doi:10.1016/j.neuroimage.2011.12.052 Contents lists available at SciVerse ScienceDirect NeuroImage journal homepage: www.elsevier.com/locate/ynimg Please cite this article as: Meier, T.B., et al., Support vector machine classification and characterization of age-related reorganization of func- tional brain networks, NeuroImage (2012), doi:10.1016/j.neuroimage.2011.12.052