Dear Justin,
>
> 1) 96 direction 3000 bvalue blip-up scan (first volume is b0 => 97 total volumes)
> 2) 96 direction 2500 bvalue blip-up scan (first volume is b0 => 97 total volumes)
> 3) 96 direction 2000 bvalue blip-up scan (first volume is b0 => 97 total volumes)
> 4) 96 direction 1500 bvalue blip-up scan (first volume is b0 => 97 total volumes)
> 5) 96 direction 1000 bvalue blip-up scan (first volume is b0 => 97 total volumes)
> 6) blip-down b0 (just 1 volume)
>
> What I did was grabbed the b0s from each scan:
>
> all_b0s = [blip-up b0 (from bvalue = 3000), blip-up b0 (from bvalue = 2500), blip-up b0 (from bvalue = 2000), blip-up b0 (from bvalue = 1500), blip-up b0 (from bvalue = 1000), blip-down b0]
>
> I fed these into topup with acqparams = [0 1 0 .1; 0 1 0 .1; 0 1 0 .1; 0 1 0 .1; 0 1 0 .1; 0 -1 0 .1]. The outputs looked correct when I checked the topup corrected b0s.
>
> Then, I concatenated all the dwmri (scans 1-5), made an index file with [1...1 2...2 3...3 4...4 5...5] (length = 96*5) and fed both of these into eddy.
>
> The problem is that the results looked incorrect. I actually reversed the scan order (1000 bvalue -> 1500 bvalue -> 2000 bvalue -> 2500 bvalue -> 3000 bvalue), reran topup and eddy, and the outputs looked correct.
>
> Is the b-value ordering important for multshell scans in eddy?
not the b-value ordering in itself. The “space/location” of the susceptibility field will be determined by the first volume in the input file to topup. The “space/location” of the eddy corrected diffusion data will be determined by the first diffusion weighted volume in the input data to eddy. You should therefore strive to have those images acquired as closely together in time as possible.
If your numbering above reflects the order in which they are acquired you should for example use the b0 from b=3000 as the first volume in the topup input and the 3000-series as the first part of the eddy input 4D file. If you do that I would have thought things should work well.
> Also, is there a best practices for this type of acquisition (multishell hemisphere and only a single reverse phase encoded b0) for topup/eddy?
Not really. What you might want to consider is to acquire the blip-down b0 a little closer in time to the first b0 of your diffusion time series. If you for example stick 6) on top of 1) you could achieve that without any disadvantages to your subject.
Jesper
|