Hi,
I assume that for a given subject all the different timepoint scans
are on the same scanner. Hence I would hope that the SIENA results
are reasonably compatible across scanners, though I would still put
in a scanner-confound-covariate in the randomise analysis. I hope
your subject groupings are matched across scanners!
However, I would be more concerned about running SIENAX across the
different scanners as that is more sensitive to pulse sequence/
hardware differences. If your groups are well matched you can still
try this (putting in scanner confound covariates) but as the effect
of scanner will probably be larger the assumption of linearly
additive confounding effect may be stretched too far.
Cheers, Steve.
On 3 Oct 2007, at 17:35, Antonios - Constantine Thanellas wrote:
> Dear fsl -users
>
> I'm gonna process with fsl (siena-sienax-randomise) MR T1 scans
> from two
> different scanners GE medical systems (1.5T and 3T) and Siemens
> (1.5T and 3T).
>
> These scanners use different pulse sequences (RM in case of GE
> scanner and
> GR/IR in case of Siemens).
>
> Among the data of the 1.5 T GE scanner there are some very slight
> differences concerning TE, TR.
> Among the data of the 3T GE scanner there were no differences in their
> characteristics.
> Comparing the differences between the scans of 1.5T and 3T GE
> scanner there
> are very small differences in the pixel resolution (0.08mm
> difference) and
> in TE (1msec), TR (2msec) and TI(100msec) parameters.
>
> As far as it concerns the Siemens scanner data, the 1.5T scans have
> among
> them slight differences in TE, TR(600msec) and image resolution and
> in some
> cases they use different coil.
> The 3T Siemens data scans have exactly the same characteristics
> among them
> and their differences with the 1.5 T Siemens scans are in matrix size
> (192x192x160 instead of 240x256x160), pixel resolution (0.25mm
> difference)
> TE(0.7msec in worst case), TR( 600msec ), TI(100msec), coil and
> flip angle
> (1 degree difference).
>
> The differences between GE scans and Siemens scans are mainly
> focused in
> Pulse sequence, TE (0.4msecs difference), TR (2400msec difference),
> matrix
> size (256x256x166 instead of 192x192x160) and pixel resolution. A
> table that
> follows gives a small example:
>
> PARAMETERS GE SCANNER SIEMENS SCANNER
> Pulse sequence RM IR/GR
> TE 3.924msec 3.54msec
> TR 8.91msec 2400msec
> Coil 8HRBRAIN PA
> Matrix size 256x256x166 192x192x160
> Resolution X,Y-axes 0.9375mm 1.25mm
>
> Do you consider wise to do longitudinal and cross sectional
> analysis with
> siena/sienax/randomise in all my data sets as one group? Or the
> differences
> in their acquisition characteristics require to separate the
> analysis in
> groups (for instance: Group 1: GE scans Group 2: Siemens scans ,
> or Group
> 1: GE 1.5T scans-Siemens 1.5T scans, Group 2: GE 3T scans -Siemens 3T
> scans, or a third case: Group 1: GE 1.5T scans, Group2:Siemens
> 1.5T, Group
> 3: GE 3T scans, Group4: Siemens 3T scans )???
>
> Thanks a lot once more
> Antonios-Constantine Thanellas
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|