Scalable Multi-Task Gaussian Process Tensor Regression for Normative Modeling of Structured Variation in Neuroimaging Data

Most brain disorders are very heterogeneous in terms of their underlying biology and developing analysis methods to model such heterogeneity is a major challenge. A promising approach is to use probabilistic regression methods to estimate normative models of brain function using (f)MRI data then use these to map variation across individuals in clinical populations (e.g., via anomaly detection). To fully capture individual differences, it is crucial to statistically model the patterns of correlation across different brain regions and individuals. However, this is very challenging for neuroimaging data because of high-dimensionality and highly structured patterns of correlation across multiple axes. Here, we propose a general and flexible multi-task learning framework to address this problem. Our model uses a tensor-variate Gaussian process in a Bayesian mixed-effects model and makes use of Kronecker algebra and a low-rank approximation to scale efficiently to multi-way neuroimaging data at the whole brain level. On a publicly available clinical fMRI dataset, we show that our computationally affordable approach substantially improves detection sensitivity over both a mass-univariate normative model and a classifier that --unlike our approach-- has full access to the clinical labels.

[1]  Maja Pantic,et al.  TensorLy: Tensor Learning in Python , 2016, J. Mach. Learn. Res..

[2]  Timothy O. Laumann,et al.  Functional Brain Networks Are Dominated by Stable Group and Individual Factors, Not Cognitive or Daily Variation , 2018, Neuron.

[3]  S. Blakemore,et al.  Studying individual differences in human adolescent brain development , 2018, Nature Neuroscience.

[4]  Jonathan D. Cohen,et al.  Matrix-normal models for fMRI analysis , 2017, AISTATS.

[5]  Dmitry Kropotov,et al.  Scalable Gaussian Processes with Billions of Inducing Inputs via Tensor Train Decomposition , 2017, AISTATS.

[6]  Krzysztof J. Gorgolewski,et al.  Preprocessed Consortium for Neuropsychiatric Phenomics dataset , 2017, F1000Research.

[7]  Alex Kendall,et al.  What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? , 2017, NIPS.

[8]  I. Rezek,et al.  Understanding Heterogeneity in Clinical Cohorts Using Normative Models: Beyond Case-Control Studies , 2016, Biological Psychiatry.

[9]  Martin Styner,et al.  STGP: Spatio-temporal Gaussian process models for longitudinal neuroimaging data , 2016, NeuroImage.

[10]  Krzysztof J. Gorgolewski,et al.  A phenome-wide examination of neural and cognitive function , 2016, Scientific Data.

[11]  B. Franke,et al.  Quantifying patterns of brain activity: Distinguishing unaffected siblings from participants with ADHD and healthy individuals , 2016, NeuroImage: Clinical.

[12]  J. Harrison,et al.  Tensor Variate Normal Distribution for Stress Variability Analysis , 2016 .

[13]  B. Franke,et al.  From estimating activation locality to predicting disorder: A review of pattern recognition for neuroimaging-based psychiatric diagnostics , 2015, Neuroscience & Biobehavioral Reviews.

[14]  Andrew Gordon Wilson,et al.  Kernel Interpolation for Scalable Structured Gaussian Processes (KISS-GP) , 2015, ICML.

[15]  Elad Gilboa,et al.  Scaling Multidimensional Inference for Structured Gaussian Processes , 2012, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[16]  Andrew Gordon Wilson,et al.  Fast Kernel Learning for Multidimensional Pattern Extrapolation , 2014, NIPS.

[17]  Martin Styner,et al.  SGPP: spatial Gaussian predictive process models for neuroimaging data , 2014, NeuroImage.

[18]  Michael Eickenberg,et al.  Machine learning for neuroimaging with scikit-learn , 2014, Front. Neuroinform..

[19]  Oliver Stegle,et al.  It is all in the noise: Efficient multi-task Gaussian process inference with structured residuals , 2013, NIPS.

[20]  T. Insel,et al.  Why has it taken so long for biological psychiatry to develop clinical tests and what to do about it? , 2012, Molecular Psychiatry.

[21]  Ara Darzi,et al.  Preparing for precision medicine. , 2012, The New England journal of medicine.

[22]  Anthony C. Davison,et al.  Statistics of Extremes , 2015, International Encyclopedia of Statistical Science.

[23]  Neil D. Lawrence,et al.  Efficient inference in matrix-variate Gaussian models with \iid observation noise , 2011, NIPS.

[24]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

[25]  Neil D. Lawrence,et al.  Computationally Efficient Convolved Multiple Output Gaussian Processes , 2011, J. Mach. Learn. Res..

[26]  Morten Mørup,et al.  Applications of tensor (multiway array) factorizations and decompositions in data mining , 2011, WIREs Data Mining Knowl. Discov..

[27]  Neil D. Lawrence,et al.  Efficient Multioutput Gaussian Processes through Variational Inducing Kernels , 2010, AISTATS.

[28]  Tamara G. Kolda,et al.  Tensor Decompositions and Applications , 2009, SIAM Rev..

[29]  Neil D. Lawrence,et al.  Sparse Convolved Gaussian Processes for Multi-output Regression , 2008, NIPS.

[30]  Brian Caffo,et al.  A Bayesian hierarchical framework for spatial modeling of fMRI data , 2008, NeuroImage.

[31]  Edwin V. Bonilla,et al.  Multi-task Gaussian Process Prediction , 2007, NIPS.

[32]  Mark W. Woolrich,et al.  Mixture models with adaptive spatial regularization for segmentation with an application to FMRI data , 2005, IEEE Transactions on Medical Imaging.

[33]  C. Loan The ubiquitous Kronecker product , 2000 .

[34]  S. Roberts EXTREME VALUE STATISTICS FOR NOVELTY DETECTION IN BIOMEDICAL DATA PROCESSING , 2000 .

[35]  Karl J. Friston,et al.  Multisubject fMRI Studies and Conjunction Analyses , 1999, NeuroImage.

[36]  Carl E. Rasmussen,et al.  In Advances in Neural Information Processing Systems , 2011 .

[37]  M. Buchsbaum,et al.  Biologic heterogeneity and psychiatrict research. Platelet MAO activity as a case study. , 1979, Archives of general psychiatry.

[38]  L. Tucker,et al.  Some mathematical notes on three-mode factor analysis , 1966, Psychometrika.