Predict

Latest version: v3.1.17

Safety actively analyzes 621688 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 1 of 5

3.1.17

-------------------

Fixed
~~~~~~~
- Extraction mode was not properly taken into account in the shape features.

Removed
~~~~~~~
- Coliage features, as they were deprecated, not used in years.

3.1.16

-------------------

Fixed
~~~~~~~
- Bug in shape features when voxel volume could not be computed from metadata.
- Included assertionerror in shape feature computation.

Changed
~~~~~~~
- In shape feature computation, previously features had value np.NaN if for
a single slice the feature was not defined in 2D / 2.5D feature extraction.
Switched to ignoring NaNs.

3.1.15

-------------------

Fixed
~~~~~~~
- Raise error when contour consists of a single pixel. Previously, we
still computed features, but the majority was NaN, so it did not make sense.

Added
~~~~~~~
- Various other image types can now also be given to PREDICT to compute
features.

3.1.14

------------------

Fixed
~~~~~~~
- 2D volume was computed on labeled mask, not binary mask.
- Related to above, 3D shape features were not computed on all connected
components
- Bug in separation between 2D, 2.5D, and 3D computation of features

3.1.13

------------------

Added
~~~~~~~
- Manufacturer routine to DICOM features

3.1.12

------------------

Added
~~~~~~~
- In DICOM features, routines for several tags to convert the value
to a float. Also support for extracting any tag required, as long
as it contains a float.

Changed
~~~~~~~
- Patient features are now called dicom features. More accurate, as they
are extracted from the DICOM tags.

Page 1 of 5

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.