Releases: RSIP-Vision/medio
Releases · RSIP-Vision/medio
v0.4.0
Version 0.4.0 release of medio, install with:
pip install -U medio
https://pypi.org/project/medio/0.4.0/
General updates:
- Upgrade to ITK version 5.1.2+, including support for the latest ITK 5.2.
- Support for Python 3.9. The supported Python versions are now: 3.6, 3.7, 3.8, 3.9.
medio.read_img
:
- Reorientations with Pydicom backend. Use Pydicom backend with:
backend='pydicom'
(or equivalently'pdcm'
).
If you know that you have dicom data, you are encouraged to use Pydicom backend, as it is usually much faster
than the default ITK backend.
Reorientations are possible with the standarddesired_ornt
parameter, which can be'LPI'
for example. - New
coord_sys
optional argument inread_img
for stable metadata coordinate system (coord_sys='itk'
by default).
Thedesired_ornt
parameter and the returned metadata will be interpreted in the provided coordinate system.
coord_sys
is either'itk'
,'nib'
, or None - in which thecoord_sys
will be determined by the backend. - New
series
optional argument for selecting a specific dicom series out of a directory with multiple dicom series,
both ITK and Pydicom backends.
series
is either a series instance UID (str), or an int between 0 and n-1, where n is the number of series in the directory.
Ifseries
is None (default), a detailed error message is presented with a suggestion on how to use theseries
argument. - Taking care of a nonconventional planar configuration with Pydicom backend (relevant for channeled images, RGB for instance).
medio.save_dir
:
- Set uniform study and series timestamps across the saved slices for improved support in MicroDicom viewer.
- Add patient position tag (empty) to support dicompyler viewer.
- New
exist_ok
optional argument (exist_ok=False
by default). If True and the directory is not empty, an error will not be raised
v0.3.0
First fully open-source release of medio:
https://pypi.org/project/medio/0.3.0/
Changes from v0.2.0 (last version):
- Added Apache 2.0 license
affine.clone()
andmetadata.clone()
methodscopy=False
in numpy's astype conversions (better performance)- Better package's metadata and documentation
- Use dicom-numpy v0.4.0