As I understand it, this important question on "equal variance" is general and may not be just about about fMRI which is just signal repeated while subjects are at rest or active. Almost all MRI data we acquire for intensity-based, volumetry voxel, tract-based or volume-atals based or paremetric or qunatittaive MRI analysis (qMRI) at good regional signal to noise are expected to follow some statistical distribution based on what has been done in the preprocessing and processing stages.
Simply, if the scanner is stable and subjects and the scans are within short period and share comparable variables (age, gender, no caffeine, bo pills,.. no lesions,..). I would expect the signal- residuals to be normal and I also expect the variances between two serial sesions to be equivalent as deemed by many tests. I also think current wisdom in statistical analyses may still recommend using ANOVA-like analysis corrected for unqequal variances (Welch t-tetes or non parametric tests).
The best way I suggest based on my experience with randomization, bootstrap and MRI analysis, is to use the fMRI resting data (not activation) and obtain histogram distributions of the signal-to-noise from one regions at multiple time points using same subject(s). Subtract signals if you do not have a noise region. Pooling subjects with wide ages, genders and different regions may increase variance and may just non-random or physiological noise to the analysis.
Khader M Hasan, PhD
Associate Professor of Radiology
MSE 168, Tel 713 500 7690 (FAX 713 500 7684)
University of Texas Health Science Center at Houston
Medical School
Diagnostic and Interventional Imaging
Magnetic Resonance Imaging Research Division
Diffusion Tensor Imaging Lab, Tel 713 500 7683
http://www.uth.tmc.edu/radiology/faculty/khader-m-hasan/index.html<http://www.uth.tmc.edu/radiology/faculty/hasan.html>
________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] On Behalf Of Thierry Chaminade [[log in to unmask]]
Sent: Wednesday, April 10, 2013 8:48 AM
To: [log in to unmask]
Subject: [SPM] Heterostadicity in longitudinal fMRI study
Hi SPMers
I have an issue with second level flexible design. I have a longitudinal study with three time points (other factors are conditions and subjects). I usually use SPM flexible ANOVA, and usually find the independence and variance settings very straightforward, but not this time.
Is my factor "time (3 levels)" independent (because different scanning times) or not because (same subject)? And the same goes with the variance, equal (same condition, same subject, as a repeated measure) or unequal (different scanning conditions)?
If anyone has the theoretical and/or empirical response to this question I will greatly appreciate.
Many thanks in advance, Thierry C.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Thierry Chaminade, PhD
Institut de Neurosciences de la Timone
UMR 7289 CNRS-AMU
Faculté de Médecine - Bâtiment Neurosciences
27, Bd Jean Moulin - 13385 Marseille Cedex 05
FRANCE
Tél +33 (0)4 91 32 40 32
Fax +33 (0)4 91 32 40 56
Web http://int.univ-amu.fr
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
|