[Neuroimaging] failure using DIPY Free Water Elimination DTI

Ariel Rokem arokem at gmail.com
Mon Jan 14 00:20:04 EST 2019 

On Sat, Jan 12, 2019 at 2:56 PM Eleftherios Garyfallidis <elef at indiana.edu>
wrote:

> Must be magic :)
>
> On Sat, Jan 12, 2019 at 4:17 PM Ping-Hong Yeh <pinghongyeh at gmail.com>
> wrote:
>
>> HI Ariel,
>> Thank you so much for the code and it ran without any errors.
>> However, i am still puzzled what my scripts failed, which i do not see
>> the differences between yours and mine.
>>
>> Yeah - I have no idea. But happy to hear that it works for you :-)

Ariel


> Regards,
>>
>> Ping
>>
>> On Thu, Jan 10, 2019 at 6:27 PM Ariel Rokem <arokem at gmail.com> wrote:
>>
>>> Hi Ping,
>>>
>>> Sorry. Using your mask, I can't reproduce the error. Here is my code:
>>> https://gist.github.com/arokem/2b7963ad5bd625f24e60c4a9307e7b7c
>>>
>>> I'd make sure to get rid of all the nans first (although I didn't seem
>>> to need to do this in my implementation).
>>>
>>> Hope that helps,
>>>
>>> Ariel
>>>
>>> On Wed, Jan 9, 2019 at 4:47 AM Ping-Hong Yeh <pinghongyeh at gmail.com>
>>> wrote:
>>>
>>>> Hi Ariel
>>>>
>>>> Have you got the mask file?
>>>>
>>>> Any update?
>>>>
>>>> Thanks!
>>>> Ping
>>>>
>>>> On Fri, Jan 4, 2019 at 1:31 PM Ping-Hong Yeh <pinghongyeh at gmail.com>
>>>> wrote:
>>>>
>>>>> I have attached the mask file.  Thank you.
>>>>>
>>>>> maskimg = nib.load(join(dname, 'structural_mask.nii.gz'))
>>>>> mask = maskimg.get_data()
>>>>> affinemask = maskimg.affine
>>>>>
>>>>>
>>>>> On Fri, Jan 4, 2019 at 1:13 PM Ariel Rokem <arokem at gmail.com> wrote:
>>>>>
>>>>>>
>>>>>> On Fri, Jan 4, 2019 at 7:42 AM Ping-Hong Yeh <pinghongyeh at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Hi Ariel,
>>>>>>>
>>>>>>> Thank you for your time.
>>>>>>>  I've run fslmaths using the syntax of -nan  beforehand, i thought
>>>>>>> that has taken care of it.
>>>>>>>
>>>>>>
>>>>>> I believe it's still there. My script crashed on that.
>>>>>>
>>>>>>
>>>>>>> Here is the script i used,
>>>>>>>
>>>>>>> import dipy.reconst.fwdti as fwdti
>>>>>>> Delta = 22.536
>>>>>>> delta = 21.908
>>>>>>> _Delta = 0.022536
>>>>>>> _delta = 0.021908
>>>>>>>
>>>>>>> dname=os.path.join('/data/NEURO2/ping/MUX', id,
>>>>>>> 'SOA_N270_plus19b0_fwd','data_DIFFPREP_t2acpc_proc_DRBUDDI_proc')
>>>>>>> fdwi = join(dname,
>>>>>>> 'data_DIFFPREP_t2acpc_proc_DRBUDDI_up_final_no7b0_nan.nii.gz')
>>>>>>> fbvec = join(dname,
>>>>>>> 'data_DIFFPREP_t2acpc_proc_DRBUDDI_up_final_no7b0_invY.bvecs') #the final
>>>>>>> oriention after TORTOISE is RAS for earlier SOA289 data
>>>>>>> fbval = join(dname,
>>>>>>> 'data_DIFFPREP_t2acpc_proc_DRBUDDI_up_final_no7b0.bvals')
>>>>>>>
>>>>>>> from dipy.io import read_bvals_bvecs
>>>>>>> bvals, bvecs = read_bvals_bvecs(fbval, fbvec)
>>>>>>> from dipy.core.gradients import gradient_table
>>>>>>> gtab = gradient_table(bvals, bvecs, big_delta=Delta,
>>>>>>> small_delta=delta)
>>>>>>> img = nib.load(fdwi)
>>>>>>> data = img.get_data()
>>>>>>> affine = img.affine
>>>>>>>
>>>>>>> #free water
>>>>>>> fwdtimodel = fwdti.FreeWaterTensorModel(gtab, fit_method='NLS' )
>>>>>>> fwdtifit = fwdtimodel.fit(data, mask=mask)
>>>>>>> img = nib.Nifti1Image(fwdtifit.fa, affine)
>>>>>>> nib.save(img, join(dname, 'FWdti_FA.nii.gz'))
>>>>>>> img = nib.Nifti1Image(fwdtifit.md, affine)
>>>>>>> nib.save(img, join(dname, 'FWdti_MD.nii.gz'))
>>>>>>> img = nib.Nifti1Image(fwdtifit.ad, affine)
>>>>>>> nib.save(img, join(dname, 'FWdti_AD.nii.gz'))
>>>>>>> img = nib.Nifti1Image(fwdtifit.rd, affine)
>>>>>>> nib.save(img, join(dname, 'FWdti_RD.nii.gz'))
>>>>>>> img = nib.Nifti1Image(fwdtifit.f, affine)
>>>>>>> nib.save(img, join(dname, 'FWdti_fw.nii.gz'))
>>>>>>>
>>>>>>>
>>>>>> Where did `mask` come from?
>>>>>>
>>>>>> Cheers,
>>>>>>
>>>>>> Ariel
>>>>>>
>>>>>>>
>>>>>>> On Fri, Jan 4, 2019 at 9:23 AM Ariel Rokem <arokem at gmail.com> wrote:
>>>>>>>
>>>>>>>> Hi Ping,
>>>>>>>>
>>>>>>>> I can't quite reproduce your error, though I am getting similar
>>>>>>>> errors. Could you also share the code that you are running?
>>>>>>>>
>>>>>>>> For the time being, the one thing that I noticed is that there are
>>>>>>>> NaNs in your data. That might be causing this issue.
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>>
>>>>>>>> Ariel
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Mon, Dec 31, 2018 at 5:43 AM Ping-Hong Yeh <
>>>>>>>> pinghongyeh at gmail.com> wrote:
>>>>>>>>
>>>>>>>>> Hi Ariel,
>>>>>>>>>
>>>>>>>>>  Have you got the chance to look at the DWI data that was attached
>>>>>>>>> to the previous email?
>>>>>>>>>
>>>>>>>>> Thank you for your time.
>>>>>>>>>
>>>>>>>>> Ping
>>>>>>>>>
>>>>>>>>> On Sat, Dec 29, 2018 at 1:12 AM Ariel Rokem <arokem at gmail.com>
>>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>>> Hi Ping,
>>>>>>>>>>
>>>>>>>>>> Thanks for your email. Hard to say exactly what is going on
>>>>>>>>>> without further information. Could you tell us more about the data you are
>>>>>>>>>> using? How many b-values? How many directions? Any chance you could share
>>>>>>>>>> the data from the voxel that caused that error to be raised?
>>>>>>>>>>
>>>>>>>>>> Thanks!
>>>>>>>>>>
>>>>>>>>>> Ariel
>>>>>>>>>>
>>>>>>>>>> On Thu, Dec 27, 2018 at 6:55 AM Ping-Hong Yeh <
>>>>>>>>>> pinghongyeh at gmail.com> wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi Dipy Users,
>>>>>>>>>>>
>>>>>>>>>>>  I've ran into errors "Eigenvalues did not  converge" when using
>>>>>>>>>>> the free water elimination model,
>>>>>>>>>>> and here is the error message:
>>>>>>>>>>>
>>>>>>>>>>>   File
>>>>>>>>>>> "/Users/yehp/anaconda3/lib/python3.6/site-packages/dipy/reconst/dti.py",
>>>>>>>>>>> line 1615, in _decompose_tensor_nan
>>>>>>>>>>>     min_diffusivity=min_diffusivity)
>>>>>>>>>>>   File
>>>>>>>>>>> "/Users/yehp/anaconda3/lib/python3.6/site-packages/dipy/reconst/dti.py",
>>>>>>>>>>> line 1946, in decompose_tensor
>>>>>>>>>>>     eigenvals, eigenvecs = eigh(tensor)
>>>>>>>>>>>   File
>>>>>>>>>>> "/Users/yehp/anaconda3/lib/python3.6/site-packages/dipy/utils/arrfuncs.py",
>>>>>>>>>>> line 128, in eigh
>>>>>>>>>>>     return np.linalg.eigh(a, UPLO)
>>>>>>>>>>>   File
>>>>>>>>>>> "/Users/yehp/anaconda3/lib/python3.6/site-packages/numpy/linalg/linalg.py",
>>>>>>>>>>> line 1291, in eigh
>>>>>>>>>>>     w, vt = gufunc(a, signature=signature, extobj=extobj)
>>>>>>>>>>>   File
>>>>>>>>>>> "/Users/yehp/anaconda3/lib/python3.6/site-packages/numpy/linalg/linalg.py",
>>>>>>>>>>> line 95, in _raise_linalgerror_eigenvalues_nonconvergence
>>>>>>>>>>>     raise LinAlgError("Eigenvalues did not converge")
>>>>>>>>>>> numpy.linalg.linalg.LinAlgError: Eigenvalues did not converge
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Any suggestions?
>>>>>>>>>>>
>>>>>>>>>>> Thank you.
>>>>>>>>>>>
>>>>>>>>>>> Ping
>>>>>>>>>>>
>>>>>>>>>>> _______________________________________________
>>>>>>>>>>> Neuroimaging mailing list
>>>>>>>>>>> Neuroimaging at python.org
>>>>>>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>>>>>>
>>>>>>>>>> _______________________________________________
>>>>>>>>>> Neuroimaging mailing list
>>>>>>>>>> Neuroimaging at python.org
>>>>>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Neuroimaging mailing list
>>>>>>>>> Neuroimaging at python.org
>>>>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Neuroimaging mailing list
>>>>>>>> Neuroimaging at python.org
>>>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Neuroimaging mailing list
>>>>>>> Neuroimaging at python.org
>>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>>
>>>>>> _______________________________________________
>>>>>> Neuroimaging mailing list
>>>>>> Neuroimaging at python.org
>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>
>>>>> _______________________________________________
>>>> Neuroimaging mailing list
>>>> Neuroimaging at python.org
>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>
>>> _______________________________________________
>>> Neuroimaging mailing list
>>> Neuroimaging at python.org
>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>
>> _______________________________________________
>> Neuroimaging mailing list
>> Neuroimaging at python.org
>> https://mail.python.org/mailman/listinfo/neuroimaging
>>
> _______________________________________________
> 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/20190113/dda289ae/attachment.html>

More information about the Neuroimaging mailing list