Hi Torsten,
I would also recommend following the advice of your Physicist. You are absolutely
right though to assume that under normal circumstances the b-value for the DW-directions
should be equal. Nevertheless there are three scenarios where this might possibly deviate
from:
1. you acquired your DW-images in an oblique direction. Then you need to apply
a coordinate transformation between your gradient coordinate system (x,y,z) and
your image coordinate system (AP, RL, FH).
Usually this transformation would only change the original directions of your gradient
vectors (x,y,z) but not their b-value.
2. additionally to your DW-gradients the imager applies gradients for the spatial
image encoding (phase encoding, read out) as well as additional gradients depending
on the sequence you are using (single shot EPI, STEAM etc.) with or without additional
flow and fat suppression. Those gradients can contribute to the diffusion weighting and
can be corrected for as well.
3. for whatever purpose the original DW-gradients have been implemented with non-equal
b-values (I wouldn't know why, though). Ask your Physicist why these b-values deviate
from the nominal 1450! And just a note on the unit: it is probably s/mm^2 not m^2/s (!).
The deviation of your generated b-values the from the nominal 1450 is at most 10% (e.g.
1568). That is not enough to easily decide from the colour coded FA-map (cFA) of your
scanned animal if either 1450 should be correct or the b-value scheme as listed.
If you have any doubt use a spherical water phantom and calculate the diffusion tensor
as well as the cFA based on the scheme with equal b-values as well as the modified
b-values. The cFA with the lowest sdev and mean close to zero will give the answer.
Evaluate the off-diagonal diffusion tensor values (Dxy, Dxz, Dyz) in terms of the
absolute value (at least one order of magnitude smaller than the trace/3) and check
the trace values (Dxx, Dyy, Dzz) - they should be as similar as possible and close
to the expected diffusion value of water for the given temperature (2.1xE-9 m^2/s at
21C and 2.95 at body temperature 37C).
I hope this helps,
Burkhard.
Burkhard Mädler, PhD
Tel.: (+49) 228 287 16186
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Torsten Ruest
Sent: Tuesday, March 09, 2010 11:48 AM
To: [log in to unmask]
Subject: [FSL] bvals
Hi there,
from our physicist I got the bvals:
10.209618 1460.209618 1568.308879 1568.308879 1460.209618 1568.308879
1352.110358
where the first one is the b"0", and the rest are the b values associated
with the 6 gradient directions. I am slightly confused, as these values are
termed "effective", while they seem to be calculated (automatically from the
scanner's software) from the set b value of 1450 m2/s. Note that these valus
have been taken from an experimental small animal MRI system, therefore
these data may not be accessible on a clinical scanner.
The question is whether I follow the advice from our physicist and take the
values given above, or shall I take the value of 1450 for each volume
(except the b 0 of course) acquired?
Thanks for clarification.
Best wishes,
Torsten
|