To Smooth or Not to Smooth? Bias and Efficiency in fMRI Time-Series Analysis

This paper concerns temporal filtering in fMRI time-series analysis. Whitening serially correlated data is the most efficient approach to parameter estimation. However, if there is a discrepancy between the assumed and the actual correlations, whitening can render the analysis exquisitely sensitive to bias when estimating the standard error of the ensuing parameter estimates. This bias, although not expressed in terms of the estimated responses, has profound effects on any statistic used for inference. The special constraints of fMRI analysis ensure that there will always be a misspecification of the assumed serial correlations. One resolution of this problem is to filter the data to minimize bias, while maintaining a reasonable degree of efficiency. In this paper we present expressions for efficiency (of parameter estimation) and bias (in estimating standard error) in terms of assumed and actual correlation structures in the context of the general linear model. We show that: (i) Whitening strategies can result in profound bias and are therefore probably precluded in parametric fMRI data analyses. (ii) Band-pass filtering, and implicitly smoothing, has an important role in protecting against inferential bias.

[1]  E. Parzen.  On Estimation of a Probability Density Function and Mode , 1962 .

[2]  C. Lawson,et al.  Solving least squares problems , 1976, Classics in applied mathematics.

[3]  B. Efron.  Bootstrap Methods: Another Look at the Jackknife , 1979 .

[4]  L. Fernholz.  von Mises Calculus For Statistical Functionals , 1983 .

[5]  C. Withers,et al.  Expansions for the Distribution and Quantiles of a Regular Functional of the Empirical Distribution with Applications to Nonparametric Confidence Intervals , 1983 .

[6]  D. Hinkley,et al.  Improvements of jackknife confidence limit methods , 1984 .

[7]  Edna Schechtman,et al.  Efficient bootstrap simulation , 1986 .

[8]  B. Efron.  The jackknife, the bootstrap, and other resampling plans , 1987 .

[9]  Karl J. Friston,et al.  Analysis of functional MRI time‐series , 1994, Human Brain Mapping.

[10]  Karl J. Friston,et al.  Statistical parametric maps in functional imaging: A general linear approach , 1994 .

[11]  Karl J. Friston,et al.  Analysis of fMRI Time-Series Revisited—Again , 1995, NeuroImage.

[12]  Charles L. Lawson,et al.  Solving least squares problems , 1976, Classics in applied mathematics.

[13]  Suzanne Corkin,et al.  A method of adjusting for temporal and spatial correlations in analysis of mean fMRI signal intensity changes , 1996, NeuroImage.

[14]  E. Bullmore,et al.  Statistical methods of estimation and inference for functional MR image analysis , 1996, Magnetic resonance in medicine.

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

[16]  Karl J. Friston,et al.  Event-related fMRI , 1997 .

[17]  M. D’Esposito,et al.  Empirical Analyses of BOLD fMRI Statistics , 1997, NeuroImage.

[18]  M. D’Esposito,et al.  Empirical analyses of BOLD fMRI statistics. I. Spatially unsmoothed data collected under null-hypothesis conditions. , 1997, NeuroImage.

[19]  A. -L. Paradis,et al.  Do high temporal frequencies of the event-related fMRI response have a more specific spatial localisation ? , 1998, NeuroImage.

[20]  Karl J. Friston,et al.  Nonlinear event‐related responses in fMRI , 1998, Magnetic resonance in medicine.

[21]  R. Weisskoff,et al.  Effect of temporal autocorrelation due to physiological noise and stimulus paradigm on voxel‐level false‐positive rates in fMRI , 1998, Human brain mapping.