Thanks to recent developments, you can already use all these official packages on Windows. You need to implement your processing in Python CLI modules. Slicer writes all input data to files, launches your module with the active Python interpreter (it can be any Python on your computer), and when processing is completed read all the outputs. Process data in CLI module is useful anyway, as processing can run in the background, without blocking the user interface.
See White Matter Analysis extension as an example, which uses scipy, multiprocessing, xlrd, cython, joblib, … packages.
Within about 6 months, we plan to migrate Slicer to Python 3 and then we’ll be able to load any binary Python packages directly in Slicer’s interpreter. However, it’ll be still advisable to do most of the processing in Python CLI modules so that they can run in the background, easily interrupted, etc.