A Model Selection Method for Nonlinear System Identification Based fMRI Effective Connectivity Analysis

In this paper a model selection algorithm for a nonlinear system identification method is proposed to study functional magnetic resonance imaging (fMRI) effective connectivity. Unlike most other methods, this method does not need a pre-defined structure/model for effective connectivity analysis. Instead, it relies on selecting significant nonlinear or linear covariates for the differential equations to describe the mapping relationship between brain output (fMRI response) and input (experiment design). These covariates, as well as their coefficients, are estimated based on a least angle regression (LARS) method. In the implementation of the LARS method, Akaike's information criterion corrected (AICc) algorithm and the leave-one-out (LOO) cross-validation method were employed and compared for model selection. Simulation comparison between the dynamic causal model (DCM), nonlinear identification method, and model selection method for modelling the single-input-single-output (SISO) and multiple-input multiple-output (MIMO) systems were conducted. Results show that the LARS model selection method is faster than DCM and achieves a compact and economic nonlinear model simultaneously. To verify the efficacy of the proposed approach, an analysis of the dorsal and ventral visual pathway networks was carried out based on three real datasets. The results show that LARS can be used for model selection in an fMRI effective connectivity study with phase-encoded, standard block, and random block designs. It is also shown that the LOO cross-validation method for nonlinear model selection has less residual sum squares than the AICc algorithm for the study.

[1]  C. Mallows Some Comments on Cp , 2000, Technometrics.

[2]  Torsten Söderström,et al.  Model-structure selection by cross-validation , 1986 .

[3]  Behzad Mansouri,et al.  Cortical deficits in human amblyopia: their regional distribution and their relationship to the contrast detection deficit. , 2007, Investigative ophthalmology & visual science.

[4]  Alan C. Evans,et al.  A general statistical analysis for fMRI data , 2000, NeuroImage.

[5]  M. Stone Cross‐Validatory Choice and Assessment of Statistical Predictions , 1976 .

[6]  Rainer Goebel,et al.  Mapping directed influence over the brain using Granger causality and fMRI , 2005, NeuroImage.

[7]  Robert F Hess,et al.  Deficient responses from the lateral geniculate nucleus in humans with amblyopia , 2009, The European journal of neuroscience.

[8]  D. Heeger,et al.  Linear Systems Analysis of Functional Magnetic Resonance Imaging in Human V1 , 1996, The Journal of Neuroscience.

[9]  Xingfeng Li,et al.  Effective connectivity anomalies in human amblyopia , 2011, NeuroImage.

[10]  C. Starmer,et al.  Indicator Transit Time Considered as a Gamma Variate , 1964, Circulation research.

[11]  Stephen A. Billings,et al.  An adaptive orthogonal search algorithm for model subset selection and non-linear system identification , 2008, Int. J. Control.

[12]  Colin L. Mallows,et al.  Some Comments on Cp , 2000, Technometrics.

[13]  Eva Möller,et al.  Fitting of one ARMA model to multiple trials increases the time resolution of instantaneous coherence , 2003, Biological Cybernetics.

[14]  Nan-Jung Hsu,et al.  Subset selection for vector autoregressive processes using Lasso , 2008, Comput. Stat. Data Anal..

[15]  Karl J. Friston,et al.  Bayesian model selection for group studies , 2009, NeuroImage.

[16]  J W Belliveau,et al.  Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. , 1995, Science.

[17]  Sheng Chen,et al.  Orthogonal least squares methods and their application to non-linear system identification , 1989 .

[18]  Lynne Kiorpes,et al.  Neural mechanisms underlying amblyopia , 1999, Current Opinion in Neurobiology.

[19]  Hans-Jochen Heinze,et al.  Causal visual interactions as revealed by an information theoretic measure and fMRI , 2006, NeuroImage.

[20]  Chantal Delon-Martin,et al.  fMRI Retinotopic Mapping—Step by Step , 2002, NeuroImage.

[21]  Paul V. McGraw,et al.  Spatial interactions reveal inhibitory cortical networks in human amblyopia , 2005, Vision Research.

[22]  Arnold Neumaier,et al.  Estimation of parameters and eigenmodes of multivariate autoregressive models , 2001, TOMS.

[23]  David M. Allen,et al.  The Relationship Between Variable Selection and Data Agumentation and a Method for Prediction , 1974 .

[24]  I. J. Leontaritis,et al.  Input-output parametric models for non-linear systems Part II: stochastic non-linear systems , 1985 .

[25]  Karl J. Friston,et al.  Nonlinear Responses in fMRI: The Balloon Model, Volterra Kernels, and Other Hemodynamics , 2000, NeuroImage.

[26]  Tim Hesterberg,et al.  Least Angle Regression and LASSO for Large Datasets , 2009, Stat. Anal. Data Min..

[27]  Anthony S. Bryk,et al.  Hierarchical Linear Models: Applications and Data Analysis Methods , 1992 .

[28]  M. Hubert,et al.  High-Breakdown Robust Multivariate Methods , 2008, 0808.0657.

[29]  A. Barron,et al.  Discussion: Multivariate Adaptive Regression Splines , 1991 .

[30]  Lester Melie-García,et al.  Estimating brain functional connectivity with sparse multivariate autoregression , 2005, Philosophical Transactions of the Royal Society B: Biological Sciences.

[31]  X. Huo,et al.  When do stepwise algorithms meet subset selection criteria , 2007, 0708.2149.

[32]  Behzad Mansouri,et al.  The fidelity of the cortical retinotopic map in human amblyopia , 2007, The European journal of neuroscience.

[33]  G. Glover,et al.  Retinotopic organization in human visual cortex and the spatial precision of functional MRI. , 1997, Cerebral cortex.

[34]  A. Cowey,et al.  Cortical area V4 and its role in the perception of color , 1992, The Journal of neuroscience : the official journal of the Society for Neuroscience.

[35]  J. Friedman,et al.  FLEXIBLE PARSIMONIOUS SMOOTHING AND ADDITIVE MODELING , 1989 .

[36]  J. Geweke,et al.  Measures of Conditional Linear Dependence and Feedback between Time Series , 1984 .

[37]  R. Tibshirani,et al.  Forward stagewise regression and the monotone lasso , 2007, 0705.0269.

[38]  T. Hesterberg,et al.  Least angle and ℓ1 penalized regression: A review , 2008, 0802.0964.

[39]  G. Glover Deconvolution of Impulse Response in Event-Related BOLD fMRI1 , 1999, NeuroImage.

[40]  M. Yuan,et al.  Model selection and estimation in regression with grouped variables , 2006 .

[41]  Serge O Dumoulin,et al.  Color responses of the human lateral geniculate nucleus: selective amplification of S-cone signals between the lateral geniculate nucleno and primary visual cortex measured with high-field fMRI , 2008, The European journal of neuroscience.

[42]  K. Kendrick,et al.  Partial Granger causality—Eliminating exogenous inputs and latent variables , 2008, Journal of Neuroscience Methods.

[43]  Jurgen A. Doornik,et al.  Testing Vector Error Autocorrelation and Heteroscedasticity , 1996 .

[44]  Clifford M. Hurvich,et al.  Regression and time series model selection in small samples , 1989 .

[45]  James Durbin,et al.  Testing for Serial Correlation in Least-Squares Regression When Some of the Regressors are Lagged Dependent Variables , 1970 .

[46]  Karl J. Friston,et al.  Nonlinear Dynamic Causal Models for Fmri Nonlinear Dynamic Causal Models for Fmri Nonlinear Dynamic Causal Models for Fmri , 2022 .

[47]  Lennart Ljung,et al.  Nonlinear black-box modeling in system identification: a unified overview , 1995, Autom..

[48]  K. Worden,et al.  Past, present and future of nonlinear system identification in structural dynamics , 2006 .

[49]  S. Billings,et al.  Fast orthogonal identification of nonlinear stochastic models and radial basis function neural networks , 1996 .

[50]  R. Tibshirani,et al.  Least angle regression , 2004, math/0406456.

[51]  D. J. Felleman,et al.  Distributed hierarchical processing in the primate cerebral cortex. , 1991, Cerebral cortex.

[52]  C. Granger Investigating causal relations by econometric models and cross-spectral methods , 1969 .

[53]  Olivier Faugeras,et al.  Using nonlinear models in fMRI data analysis: Model selection and activation detection , 2006, NeuroImage.

[54]  Daniele Marinazzo,et al.  Kernel method for nonlinear granger causality. , 2007, Physical review letters.

[55]  M. Korenberg Identifying nonlinear difference equation and functional expansion representations: The fast orthogonal algorithm , 2006, Annals of Biomedical Engineering.

[56]  Karl J. Friston,et al.  Dynamic causal modeling , 2010, Scholarpedia.

[57]  I. Hunter,et al.  The identification of nonlinear biological systems: Volterra kernel approaches , 1996, Annals of Biomedical Engineering.

[58]  Silke Dodel,et al.  Functional connectivity: studying nonlinear, delayed interactions between BOLD signals , 2003, NeuroImage.

[59]  Xue Wang,et al.  Estimating Granger causality after stimulus onset: A cautionary note , 2008, NeuroImage.

[60]  Peter J. Huber,et al.  Robust Statistics , 2005, Wiley Series in Probability and Statistics.

[61]  Ki H. Chon,et al.  Application of fast orthogonal search to linear and nonlinear stochastic systems , 2007, Annals of Biomedical Engineering.

[62]  Rodrigo Quian Quiroga,et al.  Nonlinear multivariate analysis of neurophysiological signals , 2005, Progress in Neurobiology.

[63]  T. Martin McGinnity,et al.  A least angle regression method for fMRI activation detection in phase-encoded experimental designs , 2010, NeuroImage.

[64]  Deborah A. Hall,et al.  Connectivity analysis with structural equation modelling: an example of the effects of voxel selection , 2003, NeuroImage.

[65]  Karl J. Friston,et al.  Dynamic causal modelling , 2003, NeuroImage.

[66]  Karl J. Friston,et al.  Comparing dynamic causal models , 2004, NeuroImage.

[67]  Xiaoping Hu,et al.  Effect of hemodynamic variability on Granger causality analysis of fMRI , 2010, NeuroImage.

[68]  Karl J. Friston,et al.  Multivariate Autoregressive Modelling of fMRI time series , 2003 .

[69]  W. Singer,et al.  Selection of intrinsic horizontal connections in the visual cortex by correlated neuronal activity. , 1992, Science.

[70]  Alan J. Miller Subset Selection in Regression , 1992 .

[71]  R F Hess,et al.  The cortical deficit in humans with strabismic amblyopia , 2001, The Journal of physiology.

[72]  Jafar A. Khan,et al.  Robust Linear Model Selection Based on Least Angle Regression , 2007 .

[73]  Peter J. Rousseeuw,et al.  Robust Regression and Outlier Detection , 2005, Wiley Series in Probability and Statistics.

[74]  Michael Elad,et al.  Optimally sparse representation in general (nonorthogonal) dictionaries via ℓ1 minimization , 2003, Proceedings of the National Academy of Sciences of the United States of America.

[75]  M. Korenberg,et al.  Iterative Fast Orthogonal Search for Modeling by a Sum of Exponentials or Sinusoids , 1998, Annals of Biomedical Engineering.

[76]  Alan J. Miller,et al.  Subset Selection in Regression , 1991 .

[77]  L. Faes,et al.  Assessment of Granger Causality by Nonlinear Model Identification: Application to Short-term Cardiovascular Variability , 2008, Annals of Biomedical Engineering.

[78]  Habib Benali,et al.  A nonlinear identification method to study effective connectivity in functional MRI , 2010, Medical Image Anal..

[79]  H. Akaike A new look at the statistical model identification , 1974 .

[80]  Wei Liao,et al.  Nonlinear connectivity by Granger causality , 2011, NeuroImage.

[81]  L Tychsen,et al.  Neuroanatomic abnormalities of primary visual cortex in macaque monkeys with infantile esotropia: preliminary results. , 1995, Journal of pediatric ophthalmology and strabismus.