Fetching contributors…
Cannot retrieve contributors at this time
70 lines (61 sloc) 3.33 KB

0.3.5 (March, 8, 2016)

  • Reworked now uses a pandas DataFrame to store activation data, and the Mappable class has been removed.
  • Fixed bug in peak-based search
  • Minor cleanup and doc improvements

0.3.4 (November 8, 2015)

  • Updated dependency list and moved dependencies into
  • Added ROI-based decoding method (generalization of voxel-based decoding)
  • Added new clustering module with high and low-level APIs
  • Python 3 compatibility
  • Much more extensive testing via Travis/Coveralls
  • Bug fixes/minor improvements:
    • Added thresholding options to Masker
    • Fix broken import statements
    • Removed seaborn dependency

0.3.3 (March 9, 2015)

  • Added Sphinx docs and readthedocs integration (
  • Improved study selection interface; get_ids_by_* methods replaced with a single get_studies() method
  • Added ability to download data files from within Neurosynth codebase
  • Base package now imports key modules into primary namespace (e.g., from Neurosynth import Dataset)
  • Expression parsing/study selection now handles N-gram features
  • Added deprecation warnings
  • Bug fixes/minor improvements:
    • Fixed loss-of-precision error in FDR-thresholded maps
    • Fixed unmasking of 4d volumes
    • Much faster feature count computation
    • Expanded test coverage; now includes real data in tests
    • More consistent output location arguments across codebase
    • Deleted legacy/unused code
    • Fixed naming issue when decoder is fed an array of images

0.3.2 (November 7, 2014)

  • Dataset, FeatureTable, and ImageTable all now use pandas everywhere
  • New datasets in data submodule
  • Improved test coverage of Dataset
  • Substantially improved Masker object that handles multiple masking layers
  • Added functionality to downsample whole brain to isometric grid
  • Initial stab at clustering pipeline (work in progress)
  • Improved interface for adding new features to Dataset
  • Bug fixes/minor improvements:
    • FeatureTable no longer breaks when reading in columns with N-gram names using pandas
    • get_feature_data() now takes reorder argument
    • convert pandas SparseDataFrame to CSR_matrix on data access as workaround for pandas bug
    • Fixed dataset loading/saving bug for datasets loaded/saved using previous versions
    • Fixed the above bug again, because it's a persistent little bugger
    • Updated dependency list
    • Fixed a variety of minor bugs caused by minor API changes in new pandas versions

0.3.1 (May 15, 2014)

  • Data matrices now stored internally as pandas DataFrames
  • Began implementation of classifier class (thanks to @zorro4)
  • Better/faster data accessor
  • Added several usage examples in /examples, including an IPython demo
  • Bug fixes/minor tweaks:
    • Random voxel selection routine works properly now
    • All IDs are now strings to avoid type mismatches for numerical IDs
    • Patched README and travis/coveralls definitions to reflect current dependencies

0.3.0 (May 26, 2013)

First-ever official release following the Neurosynth hackathon

  • Enabled and
  • Most of the current functionality was developed pre-0.3.0, so, really, just about everything
  • Added tests, thanks to @yarikoptic's heroic efforts
  • Testkraut integration

Neurosynth is a work in progress and we reserve the right to occasionally break the API.