future lst-ai v1.2.0 release features

[twiltgen]

Here, I listed some of the features that will be included in a future release of lst-ai (v1.2.0)

• probability map: It is now possible to save the probability maps generated by the 3D-UNets
• seeding: lst_config.py was added, which initializes the seeding value for numpy, random, and tf
• fix two-step registration in the annotation process: If "--annotate_only" is used, the registration now also consists of rigid + deformable registration. So far, it has only applied deformable registration when "--annotate_only" was used.
• clipping: It is now possible to provide clipping values to standardize image intensities during preprocessing. So far, fixed default values have been used and it was not possible to use customized values.

[twiltgen]

Ideas for future releases:

• lesion filling: integrate an existing tool (like the one in FSL) or come up with our own solution if not compatible.