FADTTS: Functional Analysis of Diffusion Tensor Tract Statistics Hongtu Zhu, Ph.D. Department of Biostatistics and Biomedical Research Imaging Center, University of North Carolina at Chapel Hill The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Outline Motivation Multivariate Varying Coefficient Models Simulation Studies Real Data Analysis The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Functional Connectivity EEG, fMRI, resting fMRI Structural Connectivity Anatomical MRI, DTI (HARDI) group 1 group 2 The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Neonatal Brain Development PI: John H. Gilmore. www.google.com Knickmeyer RC, et al. J Neurosci, 2008 28: 12176-12182. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Early Brain Development 2 week 1 year 2 year Knickmeyer RC, et al. J Neurosci, 2008 28: 12176-12182. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Diffusion Tensor Tract Statistics FA 2 week 1 year Tensor 2 year 2 week 1 year 2 year The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Macaque Brain Development Casey, B.J. et al. TRENDS in Cognitive Sciences, 2005 9(3): 104-110. PI: Martin Styner & Marc Niethammer. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Casey, B.J. et al. TRENDS in Cognitive Sciences, 2005 9(3): 104-110. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Motivation Casey, B.J. et al. TRENDS in Cognitive Sciences, 2005 9(3): 104-110. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Functional Analysis of Diffusion Tensor Tract Statistics Data • Diffusion properties (e.g., FA, RA) Yi (s j ) (yi,1(s j ),L , y i,m (s j ))T • Grids {s1,L ,snG } (e) • Covariates (e.g., age, gender, diagnostic) FA x1,L , x n MD 1 2 3 The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL FADTTS The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Multivariate Varying Coefficient Model Decomposition: y i,k (s) x Ti Bk (s) i,k (s) i,k (s) Varying Coefficients x1,L , x n Low Frequency Signal i,k () ~ SP(0, ) High Frequency Noise i,k () ~ SP(0, ), (s,s') (s,s)1(s s') Covariance operator: y (s,s') (s,s') (s,s)1(s s') The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Weighted Least Squares Estimate n nG minBk (s) K h (s s j )[y i,k (s j ) x Ti Bk (s j )]2 i1 j 1 L n{vec(Bˆ (s) B(s) 0.5O(H 2 )): s [0,L0 ]} G(0, (s,s') 1 X ) Key Advantage Low Frequency Signal The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Functional Principal Component Analysis Smooth individual functions nG min i ,k (s) K h (s j s)[y i,k (s j ) x Ti Bˆ k (s j ) i,k (s j )]2 j 1 Estimated covariance operator n ˆ ( s,t) ˆ i (s) ˆ i (t)T i1 Estimated eigenfunctions ˆ , ˆ k,l (s)): l 1,L ,} {( k,l The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Statistical Inferences Testing Linear Hypotheses H 0 : Cvec( B(s)) = b0 (s) versus H1 : Cvec( B(s)) b0 (s) Grid Point Whole Tract T 1 Sn (s j ) nd(s j )T [C( (s j ,s j ) 1 )C ] d(s j ) Global Test Statistics X L0 T 1 Sn n d(s)T [C( (s,s) 1 )C ] d(s)ds X Local Test Statistics 0 K 2 Sn (s j ) (m) and S w n k k (1), 2 k k 1 The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Confidence Band Asymptotics n[bˆk,l (s) - bk,l (s) - bias(bˆk,l (s))] Gk,l () Critical point P(sups[0,L0 ] | Gk,l (s) | Ck,l ( )) =1- Confidence band Ck,l ( ) ˆ Ck,l ( ) ˆ (bk,l (s) , bk,l (s) + ) n n The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Comparisons Pros • • • • • Directly smooth varying coefficient functions Explicitly account for functional nature of tract statistics Characterize low frequency signal Drop high frequency noise Increase statistical power Cons • • Complicated asymptotic results Computationally intensive The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Simulation Studies Model Setting (13 (s), 23 (s)) c(ˆ13 (s), ˆ23 (s)) The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Simulation Studies Testing H 0 : (13 (s), 23 (s)) (0,0) The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Power Comparison between GLM and FADTTS n 64, 0.01 n 64, 0.05 n 128, 0.01 n 128, 0.05 The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Real Data Analysis Early Brain Development • Casey, B.J. et al. TRENDS in Cognitive Sciences, 2005 9(3): 104-110. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Real Data Analysis Splenium 2 1 128 subjects MD FA 3 Diffusion properties = Gender + Gestational age The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Real Data Analysis The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Local P-values FA MD The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Confidence Bands Intercept Gender Age FA MD 1 2 3 The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Functional Principal Component Analysis Eigenvalues 1 FA MD 2 3 The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL FADTTS GUI Toolbox The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL FADTTS GUI Toolbox Input: Raw data and test data. • Raw data include tract data, design data and diffusion data. • Test data include test matrix and vector. • All data is in .mat format. Output: Basic plots and P-value plots • Basic plots include diffusion plot, coefficient plot, eigenvalue and eigenfunction plot, confidence band plot. • P-value plot include local p-value (in –log10 scale) plot with global p-value. Download: FADTTS GUI Toolbox with related documents and sample data is free to download from http://www.nitrc.org/projects/fadtts/ The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Summary • From the statistical end, we have developed a new functional analysis pipeline for delineating the structure of the variability of multiple diffusion properties along major white matter fiber bundles and their association with a set of covariates of interest. • From the application end, FADTTS is demonstrated in a clinical study of neurodevelopment for revealing the complex inhomogeneous spatiotemporal maturation patterns as the apparent changes in fiber bundle diffusion properties. • We developed a GUI Tool box to facilitate the application of FADTTS. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL Future Research • extend FADTTS to the analysis of high angular resolution diffusion image (HARDI). • extend FADTTS to principal directions and full diffusion tensors on fiber bundles. • extend to more complex fiber structures, such as the medial manifolds of fiber tracts. • extend FADTTS to longitudinal studies and family studies. The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL References • • • • Zhu, H.T., Kong, L.L., Li, R.Z., Styner, M., Gerig, G., Lin, W.L., Gilmore, J. H. (2011). FADTTS: Functional Analysis of Diffiusion Tensor Tract Statistics varying coefficient models for DTI tract statistics. Neuroimage, in press. Zhu, H.T., Li, R. Z., Kong, L.L. (2011). Multivariate varying coefficient models for functional responses. Submitted. Zhu, H., Styner, M., Li, Y., Kong, L., Shi, Y., Lin, W., Coe, C., and Gilmore, J. (2010). Multivariate varying coefficient models for DTI tract statistics. In Jiang, T., Navab, N., Pluim, J., and Viergever, M., editors, Medical Image Computing and Computer-Assisted Intervention MICCAI 2010, volume 6361 of Lecture Notes in Computer Science, pages 690697. Springer Berlin / Heidelberg. NICTR Toolbox (2011). FADTTS: Functional Analysis of Diffusion Tensor Tract Statistics. http://www.nitrc.org/projects/fadtts/ The UNIVERSITY of NORTH CAROLINA at CHAPEL HILL