A deep learning based approach identifies regions more relevant than resting-state networks to the prediction of general intelligence from resting-state fMRI

被引:13
作者
Vieira, Bruno Hebling [1 ,2 ]
Dubois, Julien [3 ,4 ]
Calhoun, Vince D. [2 ,5 ,6 ]
Salmon, Carlos Ernesto Garrido [1 ]
机构
[1] Univ Sao Paulo, InBrain Lab, Dept Fis, Ribeirao Preto, Brazil
[2] Emory Univ, Triinst Ctr Translat Res Neuroimaging & Data Sci, Georgia State Univ, Georgia Inst Technol, Atlanta, GA 30322 USA
[3] Cedars Sinai Med Ctr, Los Angeles, CA 90048 USA
[4] CALTECH, Pasadena, CA 91125 USA
[5] Mind Res Network, Albuquerque, NM USA
[6] Georgia Inst Technol, Sch Elect & Comp Engn, Atlanta, GA 30332 USA
基金
巴西圣保罗研究基金会;
关键词
brain-behavior; deep learning; fMRI; intelligence; resting-state; FUNCTIONAL CONNECTIVITY; CORTEX;
D O I
10.1002/hbm.25656
中图分类号
Q189 [神经科学];
学科分类号
071006 ;
摘要
Prediction of cognitive ability latent factors such as general intelligence from neuroimaging has elucidated questions pertaining to their neural origins. However, predicting general intelligence from functional connectivity limit hypotheses to that specific domain, being agnostic to time-distributed features and dynamics. We used an ensemble of recurrent neural networks to circumvent this limitation, bypassing feature extraction, to predict general intelligence from resting-state functional magnetic resonance imaging regional signals of a large sample (n = 873) of Human Connectome Project adult subjects. Ablating common resting-state networks (RSNs) and measuring degradation in performance, we show that model reliance can be mostly explained by network size. Using our approach based on the temporal variance of saliencies, that is, gradients of outputs with regards to inputs, we identify a candidate set of networks that more reliably affect performance in the prediction of general intelligence than similarly sized RSNs. Our approach allows us to further test the effect of local alterations on data and the expected changes in derived metrics such as functional connectivity and instantaneous innovations.
引用
收藏
页码:5873 / 5887
页数:15
相关论文
共 45 条
[1]  
Abrol A., 2020, BIORXIV PREPRINT, DOI 10.1101/2020.04.14.041582
[2]   Deep learning encodes robust discriminative neuroimaging representations to outperform standard machine learning [J].
Abrol, Anees ;
Fu, Zening ;
Salman, Mustafa ;
Silva, Rogers ;
Du, Yuhui ;
Plis, Sergey ;
Calhoun, Vince .
NATURE COMMUNICATIONS, 2021, 12 (01)
[3]  
[Anonymous], 2015, ACS SYM SER
[4]   The role of default network deactivation in cognition and disease [J].
Anticevic, Alan ;
Cole, Michael W. ;
Murray, John D. ;
Corlett, Philip R. ;
Wang, Xiao-Jing ;
Krystal, John H. .
TRENDS IN COGNITIVE SCIENCES, 2012, 16 (12) :584-592
[5]   Network Neuroscience Theory of Human Intelligence [J].
Barbey, Aron K. .
TRENDS IN COGNITIVE SCIENCES, 2018, 22 (01) :8-20
[6]  
Belkin M., 2018, Advances in neural information processing systems, V31
[7]   Julia: A Fresh Approach to Numerical Computing [J].
Bezanson, Jeff ;
Edelman, Alan ;
Karpinski, Stefan ;
Shah, Viral B. .
SIAM REVIEW, 2017, 59 (01) :65-98
[8]   Investigating the electrophysiological basis of resting state networks using magnetoencephalography [J].
Brookes, Matthew J. ;
Woolrich, Mark ;
Luckhoo, Henry ;
Price, Darren ;
Hale, Joanne R. ;
Stephenson, Mary C. ;
Barnes, Gareth R. ;
Smith, Stephen M. ;
Morris, Peter G. .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2011, 108 (40) :16783-16788
[9]   A method for making group inferences from functional MRI data using independent component analysis [J].
Calhoun, VD ;
Adali, T ;
Pearlson, GD ;
Pekar, JJ .
HUMAN BRAIN MAPPING, 2001, 14 (03) :140-151
[10]   Multisubject independent component analysis of fMRI: A decade of intrinsic networks, default mode, and neurodiagnostic discovery [J].
Calhoun, Vince D. ;
Adali, Tülay .
IEEE Reviews in Biomedical Engineering, 2012, 5 :60-73