I posted this a little while back but ended up sticking with 5.0.8 for a while. Since I've upgraded to ubuntu 16.04 I've been forced to upgrade to eddy 5.0.9.
Anyway, my data are:
b0 - blip down
b0 + 96 direction 2500 bval - blip up
b0 + 96 direction 2000 bval - blip up
b0 + 96 direction 1500 bval - blip up
b0 + 96 direction 1000 bval - blip up
Acqparams:
0 1 0 0.05
-0 -1 -0 0.05
I used the b0 from the 2500 shell and concatenated the blip-down b0, making sure the b0 from the 2500 shell was first.
Index file:
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
I concatenated all four shells and ran them through eddy, making sure the 2500 shell was first
topup parameters:
--subsamp=1,1,1,1,1,1,1,1,1 --miter=10,10,10,10,10,20,20,30,30 --lambda=0.00033,0.000067,0.0000067,0.000001,0.00000033,0.000000033,0.0000000033,0.000000000033,0.00000000000067
eddy parameters:
none
I've added the subsamp option so I can use scans with odd dimensions. the miter and lambda was set as a response to: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1611&L=fsl&D=0&P=7720. It helped fix some issues I had with topup 5.0.9.
The biggest thing for me is the DWI appear not as well aligned to the b0. I've tried --dont_sep_offs_move with eddy and it actually improved the alignment. I've tried different options with --dont_peas and without it and nothing I've tried has appeared to give results which were as good as that from 5.0.8. There's either misalignment between the DWI and b0, or there's some misalignment between shells. The data itself looks to be low SNR, so is there some sort of configuration with eddy to help with this? I believe the low SNR config I used with topup helped tremendously so I'm thinking this might be the case.
Thanks,
-Justin
|