artifact detect failing with list of 3d images.

[satra]

from bill @ nipy-user
The realigned_files is a list of 3D, not a single 4D. Looking at the source, it seemed like it was OK with either:

# read in functional image
if isinstance(imgfile, str):
    nim = load(imgfile)
elif isinstance(imgfile, list):
    if len(imgfile) == 1:
        nim = load(imgfile[0])
    else:
        images = [load(f) for f in imgfile]
        nim = funcs.concat_images(images)

# compute global intensity signal
(x, y, z, timepoints) = nim.get_shape()

When I jury-rigged it for testing (e.g. manually setting 'timepoints') it would fail a little further down with the same dimension mismatch at:

data = nim.get_data()
...
vol = data[:, :, :, t0]

I suspected that this may be a Nipype/Nibabel mismatch since it also complained that the get_shape() function was deprecated, but I'm somewhat new to this so I try not to assume. Anyway, thanks for taking a look. If anyone has a fix or a workaround (or sees an error in my setup) that would be great.

Thanks a ton. For completeness, I put the Original Inputs of the rapidart node at the bottom.

Cheers,

Bill

==================
* bound_by_brainmask : False
* global_threshold : 8.0
* ignore_exception : False
* intersect_mask : <undefined>
* mask_file : <undefined>
* mask_threshold : <undefined>
* mask_type : spm_global
* norm_threshold : 1.0
* parameter_source : SPM
* plot_type : png
* realigned_files : ['.../_session_id_02_subject_id_<SUBJID>/normalize/watime001.nii', ....]  {#120 files in sequence}
* realignment_parameters : ['..../_session_id_02_subject_id_<SUBJID>/realign/rp_time001.txt']
* rotation_threshold : <undefined>
* save_plot : True
* translation_threshold : <undefined>
* use_differences : [True, False]
* use_norm : True
* zintensity_threshold : 3.0

[satra]

could you please paste the code for constructing the ArtifactDetect node?

also since you are hacking at it, could you print nims.shape and nims.get_shape before it breaks?

[ckbehemoth]

Hello Satra,

Thanks for moving this thread to the right place. I slapped a couple of 'print's after the imgfile load function in rapidart.py (~line 364 v0.10) and here's the output:

        print "imgfile: {0}".format(imgfile)
        print "nim.shape: {0}".format(nim.shape)
        print "nim.get_shape(): {0}".format(nim.get_shape())

imgfile: ../Development/artifact_repair/l1_pipeline/preprocess/_session_id_02_subject_id_EQ051ZB/normalize/watime001.nii
nim.shape: (53, 63, 46)
nim.get_shape(): (53, 63, 46)

As for the ArtDetect node:

##############
#
# Artifact outlier check
#
art = pe.Node(interface=ra.ArtifactDetect(), name="art")
art.inputs.use_differences      = [True, False]
art.inputs.use_norm             = True
art.inputs.norm_threshold       = 1
art.inputs.zintensity_threshold = 3
art.inputs.mask_type            = 'spm_global'
art.inputs.parameter_source     = 'SPM'

#### Connections
#
preprocess.connect([
    (realign, slice_timing, [('realigned_files', 'in_files')]),
    (realign, normalize, [('mean_image', 'source')]),
    (slice_timing, normalize, [('timecorrected_files','apply_to_files')]),
    (normalize, smooth, [('normalized_files','in_files')]),
    (realign,art,[('realignment_parameters','realignment_parameters')]),
    (normalize,art,[('normalized_files','realigned_files')])
])

Is this the code you asked for? If not, I'm happy to provide more detail. Time permitting, I hope to get a version of rapidart working; I'll let you know how far I get. Thanks for the time and attention, and if I'm missing something silly or obvious, plz let me know. Nipype's been good to me ;-)

Cheers,

Bill