An estimate of the autocorrelation matrix
of the error
is required. We could estimate
using equation 2 to obtain
the residuals
and then estimate the autocorrelation matrix of the residuals.
However, it can be shown that:
![]() |
For variance correction or colouring, an estimate of
can be calculated from the residuals after equation
2 is used to obtain the parameter estimates.
This estimate of
is
used in equation 3 to give the variance of the parameter estimates.
However, prewhitening requires an estimate of
before the BLUE can be computed and equation
3 used. To get round this an iterative procedure is used
(Bullmore et al., 1996).
Firstly, we obtain the residuals
using a GLM with
.
The autocorrelation
is then estimated for these residuals.
Given an estimate of
,
and hence
can be obtained by inverting in the spectral domain
(some autocorrelation models, e.g. autoregressive, have simple parametrised forms
for
, and hence inversion in the spectral domain is not
necessary).
Next, we use a second linear model with
, and the
process can then be repeated
to obtain new residuals from which
can be re-estimated and so on.
We use just one iteration and find that it performs
sufficiently well in practice. Further iterations either give no
further benefit or cause over-fitting, depending upon the autocorrelation
estimation technique used. Autoregressive model fitting procedures which determine
the order would do the former, nonparametric approaches(Tukey, multitapering etc.)
the latter.
Whether for use in prewhitening, or for correcting the variance and degrees of freedom of the test statistic, an accurate, robust estimate of the autocorrelation is necessary. This estimation could be carried out in either the spectral or temporal domain - they are interchangeable. Raw estimates (equation 11) can not be used since they are very noisy and introduce an unacceptably large bias. Hence some means of improving the estimate is required.
All approaches considered assume second-order stationarity - an assumption whose validity is helped by the use of the non-linear high pass filtering mentioned in the previous section. We consider standard windowing or tapering spectral analysis approaches, multitapering, parametric ARMA and a nonparametric technique which uses some simple constraints. The results of these different techniques applied to a typical grey-matter voxel in a rest/null data are shown for comparison in figure 2.