Lognormal Distributions and Geometric Averages of Symmetric Positive Definite Matrices

This article gives a formal definition of a lognormal family of probability distributions on the set of symmetric positive definite (SPD) matrices, seen as a matrix-variate extension of the univariate lognormal family of distributions. Two forms of this distribution are obtained as the large sample limiting distribution via the central limit theorem of two types of geometric averages of i.i.d. SPD matrices: the log-Euclidean average and the canonical geometric average. These averages correspond to two different geometries imposed on the set of SPD matrices. The limiting distributions of these averages are used to provide large-sample confidence regions and two-sample tests for the corresponding population means. The methods are illustrated on a voxelwise analysis of diffusion tensor imaging data, permitting a comparison between the various average types from the point of view of their sampling variability.

[1]  Roland Bammer,et al.  Occipital‐Callosal Pathways in Children: Validation and Atlas Development , 2005, Annals of the New York Academy of Sciences.

[2]  P. Basser,et al.  Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. 1996. , 1996, Journal of magnetic resonance.

[3]  R. Bhattacharya,et al.  Statistics on Riemannian manifolds: asymptotic distribution and curvature , 2008 .

[4]  Ying Yao,et al.  An approximate degrees of freedom solution to the multivariate Behrens-Fisher problem* , 1965 .

[5]  Y. Benjamini,et al.  Controlling the false discovery rate: a practical and powerful approach to multiple testing , 1995 .

[6]  Jonathan E. Taylor,et al.  Inference for eigenvalues and eigenvectors of Gaussian symmetric matrices , 2008, 0901.3290.

[7]  Paul M. Thompson,et al.  Generalized Tensor-Based Morphometry of HIV/AIDS Using Multivariate Statistics on Deformation Tensors , 2008, IEEE Transactions on Medical Imaging.

[8]  Tom Leonard,et al.  Bayesian Inference for a Covariance Matrix , 1992 .

[9]  Xiuwen Liu,et al.  Extrinsic analysis on manifolds is computationally faster than intrinsic analysis with applications to quality control by machine vision , 2012 .

[10]  Morris Newman,et al.  Numerical values of Goldberg’s coefficients in the series for (^{}^{}) , 1987 .

[11]  D. Parker,et al.  A geometric analysis of diffusion tensor measurements of the human brain , 2000, Magnetic resonance in medicine.

[12]  Hongtu Zhu,et al.  Intrinsic Regression Models for Positive-Definite Matrices With Applications to Diffusion Tensor Imaging , 2009, Journal of the American Statistical Association.

[13]  S. R. Searle,et al.  Vec and vech operators for matrices, with some uses in jacobians and multivariate statistics , 1979 .

[14]  Karl Goldberg,et al.  The formal power series for $\log e^xe^y$ , 1956 .

[15]  R. Bhattacharya,et al.  LARGE SAMPLE THEORY OF INTRINSIC AND EXTRINSIC SAMPLE MEANS ON MANIFOLDS—II , 2003 .

[16]  I. Dryden,et al.  Non-Euclidean statistics for covariance matrices, with applications to diffusion tensor imaging , 2009, 0910.1656.

[17]  Xavier Pennec,et al.  A Riemannian Framework for Tensor Computing , 2005, International Journal of Computer Vision.

[18]  Roland Bammer,et al.  Diffusion tensor imaging using single‐shot SENSE‐EPI , 2002, Magnetic resonance in medicine.

[19]  S.T. Smith,et al.  Covariance, subspace, and intrinsic Crame/spl acute/r-Rao bounds , 2005, IEEE Transactions on Signal Processing.

[20]  Jimmie D. Lawson,et al.  The Geometric Mean, Matrices, Metrics, and More , 2001, Am. Math. Mon..

[21]  Tom Leonard,et al.  The Matrix-Logarithmic Covariance Model , 1996 .

[22]  Paul M. Thompson,et al.  Surface-Constrained Volumetric Brain Registration Using Harmonic Mappings , 2007, IEEE Transactions on Medical Imaging.

[23]  Ofer Pasternak,et al.  The effect of metric selection on the analysis of diffusion tensor MRI data , 2010, NeuroImage.

[24]  D. Le Bihan,et al.  Diffusion tensor imaging: Concepts and applications , 2001, Journal of magnetic resonance imaging : JMRI.

[25]  Maher Moakher,et al.  A rigorous framework for diffusion tensor calculus , 2005, Magnetic resonance in medicine.

[26]  T. O. Lewis,et al.  Quadratic Matrix Equations , 1974 .

[27]  Leif Ellingson,et al.  Nonparametric two-sample tests on homogeneous Riemannian manifolds, Cholesky decompositions and Diffusion Tensor Image analysis , 2013, J. Multivar. Anal..

[28]  Leif Ellingson,et al.  Nonparametric bootstrap of sample means of positive-definite matrices with an application to diffusion-tensor-imaging data analysis , 2017, Commun. Stat. Simul. Comput..

[29]  José Manuel Corcuera,et al.  INTRINSIC ANALYSIS OF STATISTICAL ESTIMATION , 1995 .

[30]  R. Dougherty,et al.  Group Comparison of Eigenvalues and Eigenvectors of Diffusion Tensors , 2010, Journal of the American Statistical Association.

[31]  Maher Moakher,et al.  A Differential Geometric Approach to the Geometric Mean of Symmetric Positive-Definite Matrices , 2005, SIAM J. Matrix Anal. Appl..

[32]  David Groisser.  Newton's method, zeroes of vector fields, and the Riemannian center of mass , 2004, Adv. Appl. Math..

[33]  Maher Moakher,et al.  The Riemannian Geometry of the Space of Positive-Definite Matrices and Its Application to the Regularization of Positive-Definite Matrix-Valued Data , 2011, Journal of Mathematical Imaging and Vision.

[34]  N. Ayache,et al.  Log‐Euclidean metrics for fast and simple calculus on diffusion tensors , 2006, Magnetic resonance in medicine.

[35]  N. Ayache,et al.  Clinical DT-MRI Estimation, Smoothing, and Fiber Tracking With Log-Euclidean Metrics , 2007 .

[36]  Jan R. Magnus,et al.  The Elimination Matrix: Some Lemmas and Applications , 1980, SIAM J. Algebraic Discret. Methods.

[37]  J. M. Oller,et al.  AN EXPLICIT SOLUTION OF INFORMATION GEODESIC EQUATIONS FOR THE MULTIVARIATE NORMAL MODEL , 1991 .

[38]  P. Thomas Fletcher,et al.  Riemannian geometry for the statistical analysis of diffusion tensor data , 2007, Signal Process..

[39]  Yongdo Lim,et al.  Weighted means and Karcher equations of positive operators , 2013, Proceedings of the National Academy of Sciences.

[40]  A. Terras.  Asymptotics of special functions and the central limit theorem on the space n of positive n × n matrices , 1987 .

[41]  Rachid Deriche,et al.  Statistics on the Manifold of Multivariate Normal Distributions: Theory and Application to Diffusion Tensor MRI Processing , 2006, Journal of Mathematical Imaging and Vision.

[42]  M. Fréchet.  Les éléments aléatoires de nature quelconque dans un espace distancié , 1948 .

[43]  J. Ashburner,et al.  Nonlinear spatial normalization using basis functions , 1999, Human brain mapping.

[44]  S. Cowin,et al.  Six-dimensional orthogonal tensorrepresentation of the rotation about an axis in three dimensions , 1995 .

[45]  L. Skovgaard.  A Riemannian geometry of the multivariate normal model , 1984 .

[46]  Brian A Wandell,et al.  Temporal-callosal pathway diffusivity predicts phonological skills in children , 2007, Proceedings of the National Academy of Sciences.

[47]  R. Bhatia,et al.  Riemannian geometry and matrix geometric means , 2006 .

[48]  Nicholas Ayache,et al.  Geometric Means in a Novel Vector Space Structure on Symmetric Positive-Definite Matrices , 2007, SIAM J. Matrix Anal. Appl..

[49]  Brandon Whitcher,et al.  Statistical group comparison of diffusion tensors via multivariate hypothesis testing , 2007, Magnetic resonance in medicine.

[50]  Xavier Pennec,et al.  Intrinsic Statistics on Riemannian Manifolds: Basic Tools for Geometric Measurements , 2006, Journal of Mathematical Imaging and Vision.

[51]  Morris Newman,et al.  Numerical values of Goldberg’s coefficients in the series for (^{}^{}) , 1987 .