Is your feature request related to a problem? Please describe.
This is an initial feature request based on the discussion #4227. The goal is to create a thin wrapper with APIs similar to
|
class MedNISTDataset(Randomizable, CacheDataset): |
for easy access to TCIA data collections (as a starting point we are interested in a few predefined queries presented in the notebook).
Users should be able to fetch/cache image lists and images and randomly split training/validation/test sets. The APIs should be in python (not depending on the Linux command-line tools). Please see also example in the tutorial:
https://github.com/Project-MONAI/tutorials/blob/master/modules/tcia_csv_processing.ipynb Additional dicom reading and format converting functionalities could be organised as optional transforms provided as input arguments to the dataset.
cc @AHarouni @kirbyju
Is your feature request related to a problem? Please describe.
This is an initial feature request based on the discussion #4227. The goal is to create a thin wrapper with APIs similar to
MONAI/monai/apps/datasets.py
Line 33 in 56190dd
for easy access to TCIA data collections (as a starting point we are interested in a few predefined queries presented in the notebook).
Users should be able to fetch/cache image lists and images and randomly split training/validation/test sets. The APIs should be in python (not depending on the Linux command-line tools). Please see also example in the tutorial: https://github.com/Project-MONAI/tutorials/blob/master/modules/tcia_csv_processing.ipynb Additional dicom reading and format converting functionalities could be organised as optional transforms provided as input arguments to the dataset.
cc @AHarouni @kirbyju