vsoch/pybraincompare

brain image comparison and visualization methods for python!

#pybraincompare

Semantic and computational comparison methods for brain imaging data, and visualization of outputs.

Installation

manual Installation

git clone https://github.com/vsoch/pybraincompare
cd pybraincompare
sudo python setup.py install

pip Installation

pip install pybraincompare

The current version is for Python 3, the first of which is released on pip as 0.1.19. If you want Python 2 only, use 0.1.18:

# Python 2
pip install pybraincompare==0.1.18

# Python 3
pip install pybraincompare==0.1.19

For examples and demos, please see the documentation. Submit bugs and feature requests as issues.

pybraincompare's Issues

update mr utils for update to scipy

  >>> standard_mask=get_standard_mask(4)
  /home/vsochat/.local/lib/python2.6/site-packages/scipy/ndimage/interpolation.py:435: UserWarning: The behaviour of affine_transform with a one-dimensional array supplied for the matrix parameter will change in scipy 0.18.0.
    "The behaviour of affine_transform with a one-dimensional "

Add pearson correlation value

put next to region labels for now

Axis labels for NeuroVault

should be image name, and what kind of map (if available)

qa report should give option to skip mean image

For large datasets, it would be infeasible to calculate a mean image. The option should be provided to skip this all together, and the mean image removed from the report.

add "ensure_finite" option to mrtools resample_img_ref

The ensure_finite option will ensure that NaNs are not replace with zero (if ==False) and needs to be exposed.

Add dynamic ROI image generation for connectogram

user provides labeled nifti image
package generates some visualization of the roi

Scatterplot is displaying wrong labels (or datapoints)

In this example: http://www.neurovault.org/images/compare/1251/1266 when I chose Frontal Lobe and hover over points I sometimes get other labels than "Frontal Lobe" in the tooltip:

unsupported format character ';' (0x3b) at index 195

https://opbeat.com/neurovault/neurovault/errors/11/bcb0f3d749c24a26b0bef2010e86f411

Need to write more tests

To test all functions, and some way to test GUI output.

Add transformation functions

The goal of this enhancement is to allow for synchronizing different applications that both calculate transformations (eg, NeuroVault and NeuroSynth). Specifically, in the words of Chris:

1. creating, reading and writing reduced representations of a brain image. This is a vector of values created by downsampling the map to 4x4x4, reslicing it and picking only the voxels from within the brain mask. We use this representation for similarity calculations. It's close to what you do for decoding (using random sampling)
2. calculating similarity between two maps - it's fairly simple, (it's just a correlation with pairwise deletion), but we can also share the code for it

In summary - shared downsampling code and low-resolution image representation.

Add ontological comparison - reverse inference

methods are finished, just need to package-ize them.

Priors tables should be included in package and be extended to calculate a more probabilistic ontological comparison score

Scatterplots do not scale well for p maps

http://neurovault.org/images/compare/9631/3391

Add graph structure generation

We must be able to take a list of images defined with cognitive atlas tags, and output a data structure that can be used for reverse inference analysis and visualization of result.

Add ontological comparison - graph based "wang" metric

pybraincompare should be able to calculate ontological similarity for contrast ids defined in the Cognitive Atlas. This is currently done with an R package, but we need the same in python.

Fix up within-script documentation

Every function needs to have careful, detailed documentation. Use nilearn as an example.

ValueError: unsupported format character '<' (0x3c) at index 3

Hi i am getting error on running code . can you please help

HTML_BODY= """

<title></title>
<style>
table {color: #333;font-family: calibri; width: 80%%;border-collapse:collapse; border-spacing: 0;}
td, th { border: 1px solid #CCC;width:100px; height: 30px; }
th {background: #F3F3F3;font-weight: bold;font-size: 13px;}
td {background: #FAFAFA;text-align: center; font-size: 12px;}
</style>

S.No.	Scenario Name	Total Test Cases	Passed Test Cases	Failed Test Cases
1	%s%	%s%	%s%	%s%

 

Total Scenario	Total Test Cases	Passed Test Cases	Failed Test Cases
%s%	%s%	%s%	%s%

 

""" %(self.__flow,self.__totalflowtestcases,self.__flowtestpass,self.__flowtestfail,self.__alltestcases,self.__alltestcases,self.__testpass,self.__testfail)

Tests for ontological comparison

Move docs over to circle

no more travis.

Delete intermediate image files for qa

Intermediate images shouldn't be useful other than taking up space. I will include an option to keep them, and the default will be False.

Test for pybraincompare atlas

Need to test if saved atlas object (eg labels, unique values) corresponds to image. It's exactly the same for a non-resampled image, however small regions can be eliminated with resampling even using nearest neighbor approach. The atlas labels come from the xml file, so it could be the case at a much larger voxel dimension, for much smaller regions, that the two lists do not coincide.

A warning should be added to the creation atlas step if a region is eliminated with a resampling.

Clicking atlas region label does not restore points

BUG!

Make all template css and js scripts optional

A user is possibly wanting to embed the similarity search in his or her page, in which case two versions of something like bootstrap or jquery is not ideal. I've modified similarity search interface to take a custom "remove_scripts" argument that corresponds to a tag in the template for rows that include the scripts. For this TODO:

• I need to add these tags to every single template
• I need documentation to tell a user what tags are removable