[Neuroimaging] [DIPY] weird DKI maps

[Ariel Rokem]

Hi Ping,

Thanks for your email. I have not seen anything like this, so I can only
speculate. Is there anything else that is unusual about this dataset? Does
fitting DTI give you results that look reasonable? DKI is somewhat more
sensitive to certain kinds of noise, so I would worry about a single or a
few noisy volumes (e.g., a b0 volume that is corrupted by motion). Have you
tried looking at every volume of this data?

Cheers,

Ariel





On Mon, Feb 11, 2019 at 1:13 PM Ping-Hong Yeh <pinghongyeh at gmail.com> wrote:

> Hi DIPY users,
>
> Would you comment what could go wrong for DKI fitting of this mean
> kurtosis (MK) map using a three-shell DWI (b=3000,2000,1000)?
> The values of MK map are all zero except those in the sulci and
> ventricles.
>
> Here is the command I used,
> fwhm = 1.25
> gauss_std = fwhm / np.sqrt(8 * np.log(2))  # converting fwhm to Gaussian
> std
> data_smooth = np.zeros(data.shape)
> for v in range(data.shape[-1]):
>     data_smooth[..., v] = gaussian_filter(data[..., v], sigma=gauss_std)
>
>
> dkimodel = dki.DiffusionKurtosisModel(gtab)
> dkifit = dkimodel.fit(data_smooth, mask=mask)
> MK = dkifit.mk(0, 3)
> AK = dkifit.ak(0, 3)
> RK = dkifit.rk(0, 3)
>
> I have been using the same scripts for other 200ish data and the DKI maps
> were just fine.
>
> Thank you.
>
> Ping
> _______________________________________________
> Neuroimaging mailing list
> Neuroimaging at python.org
> https://mail.python.org/mailman/listinfo/neuroimaging
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20190211/57404dfc/attachment.html>