Print

Print
Hi Jesper,

Thanks, I know from previous messages DSI acquisitions are not supported, and I think it's very good there's a check for that in eddy. As you say though, it's a bit too strict - at least for our protocol. Our protocol is to acquire a 11 b=0-images, and then 8/32/64 DWIs at b=300/700/2500 s/mm2, respectively. I know this is a protocol used in more locations, not only throughout UCL but in other universities as well. I'd be happy to send through and example dataset for you to check. Is there a way to update eddy to ensure these checks are less strict? Or to estimate the distortions on a densely sampled subset (e.g., the b=700 and b=2500 shells) and then apply to everything? Given that the eddy source code is not available I can't change it myself and recompile so I hope this is something that can be changed from your end.

Cheers,
Sjoerd

On 20 October 2015 at 11:24, Jesper Andersson <[log in to unmask]> wrote:
Dear Sjoerd,

this is a consequence of a test in eddy to ensure that data is indeed shelled (i.e. not DSI). It looks for things like number of different b-values (shells), minimum number of scans with a given b-value (shell), minimum and maximum proportion of scans within shells etc. If the data doesn’t fulfil these criteria it will be assumed that it is DSI and give the error message you saw.

However, I have come to realise that the way we design our multi shell protocols isn’t universal and that the test might be too harsh sometimes. The more examples of protocols I get from different people the better, because then I can modify my criteria to allow for these.

Jesper


On 20 Oct 2015, at 10:12, Sjoerd Vos <[log in to unmask]> wrote:

> Hi,
>
> I was trying to run the new eddy_openmp on my multi-shell diffusion data, and I got the exception copied below. eddy in v5.0.8 had no problem correcting the same dataset, but I'd like to get the rotated bvecs that v5.0.9 outputs. This exception was irrespective of how my data/bvalues were ordered. For simple single-shell data (extracting the b=0 and b=2500 data from my multi-shell data) the new version works fine. Any suggestions on this issue?
>
> Thanks,
> Sjoerd
>
> Reading images
> Running Register
> Loading prediction maker
> Evaluating prediction maker model
> Calculating parameter updates
> Iter: 0, Total mss = 257.21
> Loading prediction maker
> Evaluating prediction maker model
> Calculating parameter updates
> Iter: 1, Total mss = 229.195
> Loading prediction maker
> Evaluating prediction maker model
> Calculating parameter updates
> Iter: 2, Total mss = 228.106
> Loading prediction maker
> Evaluating prediction maker model
> Calculating parameter updates
> Iter: 3, Total mss = 228.035
> Loading prediction maker
> Evaluating prediction maker model
> Calculating parameter updates
> Iter: 4, Total mss = 227.975
> Running sm.Setb0Reference
> Running sm.PolateB0MovPar
> Running Register
> Loading prediction maker
> Evaluating prediction maker model
> I'm thrown
> terminate called after throwing an instance of 'EDDY::KMatrixException'
>  what():  KMatrixException: msg=MultiShellKMatrix::SetDiffusionPar: Data not shelled
> Abort