NF: Functionality for reading AFNI BRIK/HEAD #561

rmarkello · 2017-10-02T02:19:03Z

In partial response to #310, this is a work-in-progress for providing functionality to read AFNI BRIK/HEAD files.

For example, the following should work with these changes:

>>> import nibabel as nib
>>> brik = nib.load('example_4d+orig.BRIK')
>>> brik.shape
(33, 41, 25, 3)

I tried to model the framework off other interfaces (e.g., copying functions from PARREC and MINC—hopefully that's okay!), only modifying what was necessary to suit the specifics of BRIK/HEAD format. As of now this is pretty bare bones, but it does seem to work on a variety of manual tests and a few automated ones adopted from nibabel/tests/test_minc1.py.

I'd appreciate any feedback or insight, including on what functionality/tests should be added.

coveralls · 2017-10-02T02:42:29Z

Coverage decreased (-0.09%) to 96.176% when pulling 8efba75 on rmarkello:brikhead into dbe74e1 on nipy:master.

effigies · 2017-10-06T19:18:30Z

Hi, thanks for this. We've fixed some tests in master, if you want to rebase or merge.

codecov-io · 2017-10-09T00:05:56Z

Codecov Report

Merging #561 into master will increase coverage by 0.11%.
The diff coverage is 97.59%.

@@            Coverage Diff             @@
##           master     #561      +/-   ##
==========================================
+ Coverage   90.31%   90.42%   +0.11%     
==========================================
  Files          87       88       +1     
  Lines       10816    10982     +166     
  Branches     1794     1814      +20     
==========================================
+ Hits         9768     9930     +162     
- Misses        719      720       +1     
- Partials      329      332       +3

Impacted Files	Coverage Δ
nibabel/imageclasses.py	`100% <100%> (ø)`	⬆️
nibabel/brikhead.py	`97.57% <97.57%> (ø)`

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 636d5d2...c5f0ecf. Read the comment docs.

coveralls · 2017-10-09T00:30:17Z

Coverage increased (+0.09%) to 93.444% when pulling c5f0ecf on rmarkello:brikhead into 636d5d2 on nipy:master.

matthew-brett · 2017-11-10T14:44:46Z

Sorry to get to this so slowly - I've just started a new job ...

I noticed the reference to https://github.com/florisvanvugt/afnipy/blob/master/afni.py - but that project is GPL licensed - at a superficial first glance, you haven't used that code - is that right? If you have, we have to get permission to use it without the GPL.

Just at a superficial glance, it looks like a very good job of a first pass, and pretty close to ready. For my taste (and my interpretation of PEP8) you have too many blank lines, and I'd like to see some doctests, and eventually, we could negotiate some Image API tests for this image type (tests/test_image_api.py), but generally - excellent - thanks.

rmarkello · 2017-11-10T19:32:21Z

No need to apologize -- I've been meaning to add some more tests to get it out of its current failing state. I assumed that was the reason behind any delays!

As for the reference to afni.py: I didn't use any of that code, but rather used it for some insight into how AFNI data was stored in the BRIK file. I left the reference there to indicate the source of that insight, but it can certainly be removed since the official AFNI documentation is just as valid.

I'll work on paring down some of the blank lines and adding some doctests, but do you have a format you'd like followed for the latter? I (sort of) based this on an interpretation of parrec.py and minc1.py, neither of which have doctests. I see some minimal ones in nifti1.py that I can follow, but even there most of the functions aren't tested. Thanks for the guidance!

Shotgunosine · 2017-11-29T18:34:55Z

I emailed Floris and got him to update the license to MIT. I'm glad you had already found his code.

Shotgunosine · 2018-01-22T19:02:26Z

Hey, we'd love to see this PR merged. What needs to be done? Anything we can do to help?

effigies · 2018-01-22T19:37:27Z

Doctests that generally mirror nifti1.py (when appropriate) seem like a good start. And from a quick glance through, I wonder if we can subclass AFNIArrayProxy from ArrayProxy, to avoid rewriting more than absolutely necessary.

Nothing else really stuck out to me from a cursory read through. Merging master into this branch should make it easier for the coverage checks to identify which lines aren't being tested, and we can see about improving coverage to make sure everything's behaving as it should.

Examples of invalid BRIK/HEAD images would also be useful for checking corner cases.

rmarkello · 2018-01-22T20:52:14Z

Apologies for my neglect on this!

I was able to successfully subclass AFNIArrayProxy from ArrayProxy, getting rid of a few unnecessary functions in the process, and am currently working on some doc-tests. I'll take a look around for some invalid images, but might ask for some help from @Shotgunosine on that front.

Regarding coverage: would you prefer a merge or a rebase from master? I only ask given the explicit guidelines in the documentation.

effigies · 2018-01-22T21:10:14Z

Merges and rebases are equally acceptable. If you're comfortable with rebases, they're often cleaner to review on a commit-by-commit basis.

effigies · 2018-01-23T17:37:00Z

nibabel/brikhead.py

+           [  0.    ,   0.    ,   3.    , -52.3511],
+           [  0.    ,   0.    ,   0.    ,   1.    ]])
+    >>> head = load('example4d+orig.HEAD')
+    >>> (head.get_data() == brik.get_data()).all()


Breaking on memmap for old numpy. The following should work, though you may need to add an import numpy as np at the top of the doctest.

>>> np.array_equal(head.get_data(), brik.get_data()) True

rmarkello · 2018-01-24T13:32:56Z

Pretty sure this is broken due to the new testing file I added. Will reduce its size, hopefully seeing the Appveyor fixed, later today!

effigies · 2018-01-24T13:42:58Z

Seems to be specific to Python 3.5+, so I suspect a pretty small change will fix it.

effigies · 2018-01-24T14:19:50Z

Oh, I see. You meant the test BRIK file, not the Python test files. Carry on. :-)

Shotgunosine · 2018-01-24T17:47:41Z

I've asked around here and it doesn't look like anyone's got any bad brik-head files handy that we could test on.

matthew-brett · 2018-01-24T17:53:37Z

Could you make some bad ones by setting bad values into the header? Maybe you could save space by building and testing the bad BRIK files during the tests?

rmarkello · 2018-01-29T01:30:39Z

I've added a few broken HEAD files where I manually changed the attributes, as suggested. Thankfully, since the HEAD file gets read in first, there's no need for an accompanying BRIK file since it will error out before getting there.

However, taking a closer look at the AppVeyor error, I'm a bit confused. From some digging, I think it's specific to the example4d+orig.BRIK.gz file (the doc-test that uses that file is failing, too). The only change made to that file prior to the AppVeyor failing was that I gzipped it (87873ee) in order to test capabilities for detecting compressed files. I'm unclear why that would have screwed things up (and only on Python 3.5+, as @effigies pointed out). I'll keep digging, but if you have any ideas they'd be appreciated!

effigies · 2018-01-29T16:34:29Z

From a quick search, it seems others have run up against this on AppVeyor when using mmaped files. It's resolved by explicitly closing the file handles.

This seems like probably any easy fix to attempt before we dig deeper.

effigies

Had a quick look through for unclosed mmaps. I may have found it, or may be declaring a premature victory.

effigies · 2018-01-30T14:04:54Z

nibabel/brikhead.py

+                hdr_fobj)
+        brik_fobj = file_map['image'].get_prepare_fileobj()
+        data = klass.ImageArrayProxy(brik_fobj, hdr,
+                                     mmap=mmap)


I think the issue might be here. I don't think you need to get_prepare_fileobj in this method.

data = klass.ImageArrayProxy(file_map['image'], hdr, mmap=mmap)

I think this will help release the filehandles between reads. Please see analyze.py for probably the most exercised (as the Nifti1Image superclass) variant of from_file_map.

effigies · 2018-01-30T14:06:35Z

nibabel/brikhead.py

+        """
+        file_map = klass.filespec_to_file_map(filename)
+        # only BRIK can be compressed, but `filespec_to_file_map` doesn't
+        # handle that case; remove potential compression suffixes from HEAD


As a point of personal preference, I would rather improve filespec_to_filemap to handle the case where header may not be compressed. (But if that makes more sense as a separate PR, just to get this in, that's fine by me.)

I think it'd be good to see this PR merged, first, if that's all right! Once this is in I'll be happy to raise an issue to discuss this change, since, as far as I can tell filespec_to_file_map is used by most image classes in nibabel.

How about a little wrapper around filespec_to_filemap in this PR, to point the way, and keep that part of the logic separate?

effigies

Memmap issue cleared up with that last commit, and the test coverage is looking good. I've got no more comments at this time.

effigies · 2018-02-09T14:07:04Z

How's everybody feeling about this PR?

@rmarkello @Shotgunosine are you using this regularly and it's operating as you'd hope? Any further additions you're thinking of making?

@matthew-brett Have you had a chance to look over this? A lot of this plumbing is stuff you're much more familiar with than I am.

rmarkello · 2018-02-12T01:42:16Z

I haven't run into any issues with my own use of this, but, admittedly, I've been utilizing it in some pretty pedestrian workflows (though I suppose that will be the normal use case!).

In terms of further additions: I think, as mentioned above, modifying the filespec_to_file_map will be the next step.

matthew-brett

Some thoughts reading through.

Generally - thanks - you did a good job of working out the general structure in nibabel.

At some point, we will need some API tests in test_image_api.py.

matthew-brett · 2018-02-20T12:58:57Z

nibabel/brikhead.py

+               'check .HEAD file and try again.')
+    # get data type and key; if error, bad attribute/HEAD file
+    try:
+        atype = TYPE_RE.findall(var)[0]


Is OK to find more than one TYPE?

Consider more specific error messages?

There should never be more than one TYPE associated with an attribute in a HEAD file. If there is, the HEAD file is corrupted and an error should be raised. I'll add that check and make the error messages more explicit.

matthew-brett · 2018-02-20T13:01:26Z

nibabel/brikhead.py

+See https://afni.nimh.nih.gov/pub/dist/doc/program_help/README.attributes.html
+for more information on required information to have a valid BRIK/HEAD dataset.
+
+Examples


This is going to appear as part of a main module docstring - I guess that's a bit ugly.

An alternative might be to define data_dir and use explicit os.path.join(datadir, 'my_file.HEAD') in the doctests, assuming the user can work out what datadir is.

matthew-brett · 2018-02-20T13:05:52Z

nibabel/brikhead.py

+    (4, 'mni', 'MNI')), fields=('code', 'label', 'space'))
+
+
+class AFNIError(Exception):


You could also use HeaderDataError (and ImageDataError) from spatialimages.py, or inherit from these. I realize that the other image types don't do that, but I think it's more correct.

matthew-brett · 2018-02-20T13:07:59Z

nibabel/brikhead.py

+    Parameters
+    ----------
+    info : dict
+        As obtained by `parse_AFNI_header()`


Nicer Sphinx output from

As obtained by :func:`parse_AFNI_header`

matthew-brett · 2018-02-20T13:09:55Z

nibabel/brikhead.py

+
+    Returns
+    -------
+    (key, value)


As for other docstrings, this line should have a name and type, for example:

Returns -------- name : str Name of attribute value : object Value of attribute.

matthew-brett · 2018-02-20T13:19:51Z

nibabel/brikhead.py

+        shape = tuple(self.info['DATASET_DIMENSIONS'][:dset_rank[0]])
+        n_vols = dset_rank[1]
+
+        return shape + (n_vols,) if n_vols > 1 else shape


Is that correct, that you can't have a shape (x, y, z, 1) AFNI dataset, it is forced back to (x, y, z) shape?

Well, no. By default, AFNI datasets are actually always ndim=4 of shape (x, y, z, >=1), because of how the dimensions are stored in the HEAD file (see response below). You'll never get an (x, y, z) dataset because that would preclude you from subsetting along the "time" axis, and AFNI programs feature heavily in permitting you to do this. I'll make sure to remedy this in my next update.

matthew-brett · 2018-02-20T15:53:02Z

nibabel/brikhead.py

+        floatfacs = self.info.get('BRICK_FLOAT_FACS', None)
+        if floatfacs is None or not np.any(floatfacs):
+            return None
+        scale = np.ones(self.info['DATASET_RANK'][1])


Is DATASET_RANK[1] the number of time points always? So when the AFNI docs say time axis, they mean DATASET_RANK[1] > 1?

Correct. DATASET_RANK[0] must be set to 3, denoting a 3D image, and DATASET_RANK[1] will always determine how many "sub-bricks" (in AFNI parlance) / volumes there are along the fourth (traditionally, but not exclusively) time axis. As in my above response, DATASET_RANK[1] will (at least as far as I'm aware) always be >=1.

Thanks - would you mind putting this in comments somewhere? Maybe in an extra extended docstring at the top? As in:

""" The current module docstring Some comments about this module for public consumption """ """ More notes on the AFNI format DATASET_RANK[0] must be set to 3, denoting a 3D image, and DATASET_RANK[1] will always determine how many "sub-bricks" (in AFNI parlance) / volumes there are along the fourth (traditionally, but not exclusively) time axis. As in my above response, DATASET_RANK[1] will (at least as far as I (RM) am aware) always be >=1. """

matthew-brett · 2018-02-20T15:54:39Z

nibabel/brikhead.py

+
+        return scale
+
+    def get_slope_inter(self):


Did you need this? Maybe NotImplemented if it doesn't make sense for this format? Or just remove it?

Because AFNIArrayProxy inherits from ArrayProxy, this needs to be included. ArrayProxy demands that the provided "spec" (i.e., AFNIHeader in this case) implement this (see: arrayproxy.py:140). I added the inheritance in response to @effigies suggestion, above.

Maybe add comment to this effect?

matthew-brett · 2018-02-20T16:00:03Z

nibabel/brikhead.py

+        """
+        file_map = klass.filespec_to_file_map(filename)
+        # only BRIK can be compressed, but `filespec_to_file_map` doesn't
+        # handle that case; remove potential compression suffixes from HEAD


How about a little wrapper around filespec_to_filemap in this PR, to point the way, and keep that part of the logic separate?

matthew-brett · 2018-02-20T16:04:37Z

nibabel/brikhead.py

+        head_fname = file_map['header'].filename
+        if not os.path.exists(head_fname):
+            for ext in klass._compressed_suffixes:
+                head_fname = re.sub(ext, '', head_fname)


I think this is going to remove gz or whatever inside filenames - for example - consider head_fname = 'foogz_bar.nii.gz'.

I think you want something like:

for ext in klass._compressed_suffixes: head_fname = head_fname[:-len(ext)] if head_fname.endswith(ext) else head_fname

@matthew-brett

Edited brikhead.py in response to @matthew-brett reviews in nipy#561. Cleaned up doc-strings, added/clarified error mesages to be more specific, and modified how AFNI files are read in to ensure that they accurately represent underlying data (i.e., are always 4D). Added `filespec_to_file_map` class method for `AFNIImage` to separate logic of reading in files with compression; likely need to edit `filename_parser.types_filenames` instead.

rmarkello · 2018-02-26T18:28:31Z

To do:

Add relevant API tests to test_image_api.py
Wrap filename_parser.types_filenames to account for BRIK/HEAD idiosyncracies currently being handled by AFNIImage.filespec_to_file_map

rmarkello · 2018-03-06T14:26:34Z

I think this might be ready for another look! Let me know if there's anything else I can do.

matthew-brett

Some comments on the main file.

matthew-brett · 2018-03-09T11:20:35Z

nibabel/brikhead.py

+
+class AFNIImageError(ImageDataError):
+    """Error when reading AFNI BRIK files"""
+    pass


You actually don't need the pass, given you have the comment.

matthew-brett · 2018-03-09T11:26:29Z

nibabel/brikhead.py

+        raise AFNIHeaderError('Invalid attribute name entry in HEAD file. '
+                              '%s' % err_msg)
+    atype = _attr_dic.get(atype[0], str)
+    attr = ' '.join(var.strip().split('\n')[3:])


Maybe attr = ' '.join(var.strip().splitlines()[3:]) - deals with Windows line endings too, just in case.

matthew-brett · 2018-03-09T11:28:41Z

nibabel/brikhead.py

+        except ValueError:
+            raise AFNIHeaderError('Failed to read variable from HEAD file due '
+                                  'to improper type casting. %s' % err_msg)
+        if len(attr) == 1:


Maybe return aname, attr[0] if len(attr) == 1 else attr . If you prefer your version, that's also fine.

matthew-brett · 2018-03-09T11:31:49Z

nibabel/brikhead.py

+    >>> print(info['BRICK_TYPES'])
+    [1, 1, 1]
+    """
+    from six import string_types


Import better at top of file?

matthew-brett · 2018-03-09T11:33:49Z

nibabel/brikhead.py

+        self._scaling = header.get_data_scaling()
+
+    @property
+    def scaling(self):


Maybe note this new attribute in the class docstring.

matthew-brett · 2018-03-09T11:44:45Z

nibabel/brikhead.py

+            fake_data = strided_scalar(self._shape)
+            _, scaling = np.broadcast_arrays(fake_data, scaling)
+            raw_data = raw_data * scaling[slicer]
+


I have a general preference for avoiding blank lines, probably from contact with PEP8 "Use blank lines in functions, sparingly, to indicate logical sections.". Your call though.

matthew-brett · 2018-03-09T11:46:01Z

nibabel/brikhead.py

+        Get image zooms from header data
+
+        Spatial axes are first three indices, time axis is last index. If
+        dataset is not a time series the last index will be zero.


I think you mean that the last value will be zero, rather than the last index.

matthew-brett · 2018-03-09T11:53:35Z

nibabel/brikhead.py

+
+    Examples
+    --------
+    >>> brik = load(os.path.join(datadir, 'example4d+orig.BRIK.gz'))


Better to use nib.load in example (they might not realize where load comes from).

matthew-brett · 2018-03-09T12:03:59Z

nibabel/brikhead.py

+            If `filespec` is not recognizable as being a filename for this
+            image type.
+        """
+        # copied from filebasedimages.py


Can you post-process instead? As in, inherit from FileBasedImage (sorry - why aren't you doing that? - I'm sure there's a good reason), and post-process the result?

file_map = super(AFNIImage, klass).filespec_to_file_map(filespec)

and then use file_map['image'].filename and file_map['header'].filename?

Do you agree that it seems cleaner to raise an error for my_image.HEAD.gz, even if my_image.BRIK.gz exists?

matthew-brett · 2018-03-09T12:20:39Z

nibabel/brikhead.py

+        """
+        Make `file_map` from filename `filespec`
+
+        Deals with idiosyncracies of AFNI BRIK / HEAD formats


Maybe add something like:

AFNI BRIK files can be compressed, but HEAD files cannot - see https://afni.nimh.nih.gov/pub/dist/doc/program_help/README.compression.html. Let's say you have an AFNI file you want to load, with files my_image.HEAD and my_image.BRIK.gz. To identify the AFNI BRIK / HEAD pair, you can specify:

The HEAD filename - e.g. my_image.HEAD

The BRIK filename without compressed extension, e.g. my_image.BRIK

The full BRIK filename - e.g. my_image.BRIK.gz.

As a side note, I guess you're not proposing to support .Z compressed files. I bet they are rare.

matthew-brett

Sorry to be slow - was moving house, things got hectic.

Here a few small comments on brikhead.py. I also made some small edits, doc extensions in rmarkello#1

matthew-brett · 2018-05-04T08:54:21Z

nibabel/brikhead.py

+        raw_data = super(AFNIArrayProxy, self).__getitem__(slicer)
+        # apply volume specific scaling (may change datatype!)
+        if self._scaling is not None:
+            scaling = self._scaling.copy()


Why the copy?

I don't remember anymore! It doesn't seem necessary so I'll fix it asap.

matthew-brett · 2018-05-04T11:11:58Z

nibabel/brikhead.py

+    """
+    bo = info['BYTEORDER_STRING']
+    bt = info['BRICK_TYPES']
+    if isinstance(bt, list):


Is it ever not a list?

Yes; for a single volume image, this would be an integer value.

For PR nipy#561

Includes new test file `example_4d+orig.BRIK/HEAD`

Changed AFNIArrayProxy to subclass from ArrayProxy. Removed whitespace.

Created new tests for brikhead.py and added/modified datafiles to support them. Notably, added scaled+tlrc.BRIK/HEAD to test inclusion of data scaling factors and gzipped exampled4d+orig.BRIK to accommodate loading gzipped files.

Added a few "broken" .HEAD files to the testing suite for brikhead.py and implemented a few try-except blocks to catch the resultant errors. Maybe (?) resolved the issues with the AppVeyor OSError on Python 3 for numpy memmap in loading the .BRIK files. Fixed a doc-test in brikhead.py.

`brikhead.AFNIImage.from_file_map()` was needlessly opening the 'image' file before passing it to ImageArrayProxy, potentially causing issues seen on AppVeyor with mmaps. This commit modifies the call to be more in line with how this is handled in analyze.py.

@matthew-brett

Edited brikhead.py in response to @matthew-brett reviews in nipy#561. Cleaned up doc-strings, added/clarified error mesages to be more specific, and modified how AFNI files are read in to ensure that they accurately represent underlying data (i.e., are always 4D). Added `filespec_to_file_map` class method for `AFNIImage` to separate logic of reading in files with compression; likely need to edit `filename_parser.types_filenames` instead.

Created minimal test_image_api test for brikhead, modeled on PARREC image tests.

Edited filespec_to_file_map in AFNIImage to handle BRIK/HEAD compression business more smoothly.

Mostly edited ``AFNIImage.filespec_to_file_map``, but also made minor changes in response to review comments. Added doc-strings describing relevant AFNI-specific features.

Forgot to import nibabel in doctests...

Minor updates to tests to check error handling

Building on PR nipy#561.

For PR nipy#561

rmarkello · 2018-05-22T15:49:51Z

This should be ready to go, @matthew-brett ! Thanks so much for the helpful feedback.

effigies · 2018-06-01T14:43:47Z

@matthew-brett Sorry to keep bothering you, but is this ready to go? I see you made some contributions... not sure if that was final touch-ups or not.

matthew-brett

Nice - thanks - and very sorry for the delay. Let's put this in, and see what happens :)

matthew-brett · 2018-06-01T18:58:54Z

Chris - if you're happy for this to go in the release, please merge away.

effigies · 2018-06-01T18:59:28Z

So. Happy.

rmarkello force-pushed the brikhead branch from 8efba75 to e1e95c4 Compare October 9, 2017 00:05

rmarkello force-pushed the brikhead branch from e1e95c4 to 2f4b69f Compare January 23, 2018 02:36

effigies reviewed Jan 23, 2018

View reviewed changes

effigies reviewed Jan 30, 2018

View reviewed changes

effigies approved these changes Jan 31, 2018

View reviewed changes

matthew-brett reviewed Feb 20, 2018

View reviewed changes

matthew-brett reviewed Mar 9, 2018

View reviewed changes

matthew-brett reviewed May 4, 2018

View reviewed changes

matthew-brett mentioned this pull request May 4, 2018

Any interest in having a look at nibabel PR 561? florisvanvugt/afnipy#1

Open

rmarkello added a commit to rmarkello/nibabel that referenced this pull request May 7, 2018

RF: Fixed needless list copy in brikhead.py

f70793e

For PR nipy#561

rmarkello and others added 16 commits May 22, 2018 10:45

NF: Basic functionality for reading AFNI BRIK/HEAD files

1890839

TEST: Minimal tests for AFNI BRIK/HEAD files

f088f51

Includes new test file `example_4d+orig.BRIK/HEAD`

PL: Cleaning up brikhead flake8 errors

c418a1f

RF: ArrayProxy and whitespace

c78c786

Changed AFNIArrayProxy to subclass from ArrayProxy. Removed whitespace.

TEST: Added BRIK/HEAD doc-tests

8d9e414

TEST: Added tests/files for brikhead.py

aee4f91

Created new tests for brikhead.py and added/modified datafiles to support them. Notably, added scaled+tlrc.BRIK/HEAD to test inclusion of data scaling factors and gzipped exampled4d+orig.BRIK to accommodate loading gzipped files.

TEST: Added brikhead to test_image_api.py

6bd34fe

Created minimal test_image_api test for brikhead, modeled on PARREC image tests.

RF: Added brikhead.filespec_to_file_map

a674003

Edited filespec_to_file_map in AFNIImage to handle BRIK/HEAD compression business more smoothly.

RF: Minor brikhead.py refactoring

477c2ec

Mostly edited ``AFNIImage.filespec_to_file_map``, but also made minor changes in response to review comments. Added doc-strings describing relevant AFNI-specific features.

FIX: Brikhead doctests

04df389

Forgot to import nibabel in doctests...

TEST: Added testing for errors

8c7e1eb

Minor updates to tests to check error handling

RF: some edits / doc extensions to brikhead.py

6b308bf

Building on PR nipy#561.

RF: Fixed needless list copy in brikhead.py

c5f0ecf

For PR nipy#561

rmarkello force-pushed the brikhead branch from f70793e to c5f0ecf Compare May 22, 2018 14:46

This was referenced May 29, 2018

ETA for next nibabel release? #631

Closed

REL: 2.3.0 #632

Merged

matthew-brett approved these changes Jun 1, 2018

View reviewed changes

effigies merged commit 4c2bca7 into nipy:master Jun 1, 2018

nibotmi mentioned this pull request Jun 1, 2018

NF nib-diff #617

Merged

1 task

effigies mentioned this pull request Apr 25, 2019

Provide at least reading capability for AFNI's (BRIK/HEAD) format #310

Closed

		(4, 'mni', 'MNI')), fields=('code', 'label', 'space'))


		class AFNIError(Exception):

NF: Functionality for reading AFNI BRIK/HEAD #561

NF: Functionality for reading AFNI BRIK/HEAD #561

Uh oh!

Conversation

rmarkello commented Oct 2, 2017

Uh oh!

coveralls commented Oct 2, 2017

Uh oh!

effigies commented Oct 6, 2017

Uh oh!

codecov-io commented Oct 9, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

coveralls commented Oct 9, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

matthew-brett commented Nov 10, 2017

Uh oh!

rmarkello commented Nov 10, 2017

Uh oh!

Shotgunosine commented Nov 29, 2017

Uh oh!

Shotgunosine commented Jan 22, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

effigies commented Jan 22, 2018

Uh oh!

rmarkello commented Jan 22, 2018

Uh oh!

effigies commented Jan 22, 2018

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rmarkello commented Jan 24, 2018

Uh oh!

effigies commented Jan 24, 2018

Uh oh!

effigies commented Jan 24, 2018

Uh oh!

Shotgunosine commented Jan 24, 2018

Uh oh!

matthew-brett commented Jan 24, 2018

Uh oh!

rmarkello commented Jan 29, 2018

Uh oh!

effigies commented Jan 29, 2018

Uh oh!

effigies left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

effigies left a comment

Choose a reason for hiding this comment

Uh oh!

effigies commented Feb 9, 2018

Uh oh!

rmarkello commented Feb 12, 2018

Uh oh!

matthew-brett left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

codecov-io commented Oct 9, 2017 •

edited

Loading

coveralls commented Oct 9, 2017 •

edited

Loading

Shotgunosine commented Jan 22, 2018 •

edited

Loading

rmarkello commented Feb 26, 2018 •

edited

Loading