I'm getting an error with dipy DSI reconstruction on HCP WU-MINN data.
It's a bit of a mystery as I've definitely got successful reconstructions using dipy with code very similar to this. But now I am getting the error below every time.
All the files are included in this gist to reproduce the error, apart from the DWI data (too big); that was simply data.nii.gz from subject 100307.
The error:
In [5]: run track_dipy_hcp_test.py
0dit-`/alexandra/mcintosh_lab/john/Software/miniconda2/envs/_ipython2.4b/lib/python2.7/site-packages/dipy/reconst/dsi.py:173: RuntimeWarning: invalid value encountered in divide
Pr /= Pr.sum()
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
/alexandra/mcintosh_lab/john/Data/dipy_hcp_test_gist/track_dipy_hcp_test.py in <module>()
93 relative_peak_threshold=.5,
94 min_separation_angle=25,
---> 95 parallel=False) # True
96
97 # get generalized fractional anisotropy (GFA)
/alexandra/mcintosh_lab/john/Software/miniconda2/envs/_ipython2.4b/lib/python2.7/site-packages/dipy/direction/peaks.pyc in peaks_from_model(model, data, sphere, relative_peak_threshold, min_separation_angle, mask, return_odf, return_sh, gfa_thr, normalize_peaks, sh_order, sh_basis_type, npeaks, B, invB, parallel, nbr_processes)
453 direction, pk, ind = peak_directions(odf, sphere,
454 relative_peak_threshold,
--> 455 min_separation_angle)
456
457 # Calculate peak metrics
/alexandra/mcintosh_lab/john/Software/miniconda2/envs/_ipython2.4b/lib/python2.7/site-packages/dipy/direction/peaks.pyc in peak_directions(odf, sphere, relative_peak_threshold, min_separation_angle, minmax_norm)
128
129 """
--> 130 values, indices = local_maxima(odf, sphere.edges)
131
132 # If there is only one peak return
/alexandra/mcintosh_lab/john/Software/miniconda2/envs/_ipython2.4b/lib/python2.7/site-packages/dipy/reconst/recspeed.so in dipy.reconst.recspeed.local_maxima (dipy/reconst/recspeed.c:4598)()
ValueError: odf can not have nans