galdude33 / fetal-mri-segmentation Goto Github PK

Thank you for your awesome work, I have a couple of questions and I appreciate if you can elaborate more on how to use your repository. as I am new to this world of medical image analysis, some questions may sound simple but I need a kick to be able to start using your code:

1. lets say I have a new data set of fetal MRI, what is the procedure of using your code so I could get a 3D model of the fetus structure as described in your readme.? e.g what is the order of using your .py files?
2. the data which I have is a DICOM directory, is there any changes necessary to the dataset? e.g. cropping, changing the data before interaction with NN? any specific cross section of fetus needed as input data?
3. I have followed the instructions for training and prediction, but it seems I cannot setup your repository in a manner that you have intended, especially, what is exactly the <config_dir> "containing the training configurations", or "config["split_dir"] = <split_dir_path> # The directory to load/save the train/val/test split files", any suggestions?

As my research project is going to build on-top of your work, I will appreciate it if you can help me. if there is a way to stay connected outside the issue page, please let me know.

Regards.

Make the function 'reconstruct_from_patches' parallel, currently single-threaded and slow

Currently the data is split to Train and Validation, we should add Test split (probably just one sample) for testing purpose, and check for over-fitting.

Add progress-bar like in 'patch_wise_prediction'

I'm doing some paper research on this. Can you tell me where I can find MRI datasets related to fetuses.thanks a million