Latest version 2.10.8
- [changed] Less log messages
- [added] export of masks from COCO annotation files in devel
- [Added] Read by url
- [Added] Export masks from COCO annotation files
- [Changed] Dicom files uint are stored directly. The int dtype in 3D can be stored with intercept in metadata.
- [Added] Nifti support for '.ni.gz' extension
- [Added] Orientation codes support
- [Added] Data can be accessed by the property (e.g.
datap.data3d
)
- [Added]
read_dataset()
allow to read different datasets in the same way - [Changed]
join_path
returnPath
object as default - [Added] Datasets are now stored in
datasets.csv
on github - [Fixed] Deprecation Warning on
get_slice_location()
- [Fixed] Shortest specific path now works
- [Changed] Relative dataset download path implementation
- [Added] URL can be used to download data
- [Added] Specific dataset can be stored in different path (download is in plan todo)
- [Changed] Some data are stored to
sample_data
instead ofsample-data
- [Changed] Internal representation of dataset dir is now just
root
(instead ofroot/medical/orig
) - [Changed] Yaml save cache with safe typ
- [Added] Crate backup cache file if the old is wrong
- [Changed] dependency on actual
imma
package
- [Changed] Python 3 compatibility only
- [Fixed] mode and order in
resize_to_shape
function - [Fixed] respect new functionality in skimage.transform
- [Added] New arguments in
resize_to_shape
function - [Added] New tests for seeds
- [Added] On resize check if all seeds are preserved
- [Added] New function io3d.datasets.get_labels() for obtaining list of available data
- [Changed] Checkum is computed on oreder directory (is less system dependent)
- [Changed] Compact checksumlog. Added hash for ircad
- [Added] Control over relative download directory for every dataset
- [Changed] Dataset SCP003 splited into image data and annotation
- [Added] New sample data SCP003
- [Changed] Save all types acceptable by json to hdf5
- [Changed] Accept None type i hdf5 export
- [Changed] Not-string like keys supported in hdf5
- [Added] File format h5 supported
- [Added] Reading patient name, id, age and sex
- [Added] Display patient's info in qt widget
- [Changed] If file exists the new file name is created which does not collide with Study ID
- [Added] Guess series number for automatic liver processing
- [Changed] Logging in dataset more verbatim
- [Added] Slice data added to datasets
- datasets.join_path can return datasets root path
- pydicom warnings are supressed
- New function for removing files
- New distance segmentation function
- New dataset generate function with synthetic liver and portal vein
- Rotation works now with interpolation order = 1
- Test and first development on File System Browser
- Logger for
dicom_expected
flag
- Added
dicom_expected
flag to prevent warning
- Directory information is returned in
pandas
dataframe - New sample generator - face
- Fixed warnings
- Updated requirements
- Datasets update
- New pklz write test
- Fixed SimpleITK version issues
- Datasets moved to other server
- Filter doubled files
- Memory error fixed
- PyQt4 QString compatibility with python 2 and python 3
- Better code climate with PEP8 fixes
- GUI Qt4 bytes type compatibility with python 2 and python 3
- New download function for datasets
- Object pickle algorithm compatible with python 2 and python 3