Auto-encoder pycox implementation for 3D images instead of tabular data

[nickel-04]

Hi Håvard Kvamme,

Thank you for your creating the pycox. it's been really helpful to implement survival model for different datasets.

I am working on survival analysis for liver cancer patients. I am having 3D CT data and clinical data. I have been trying to implement "https://github.com/havakv/pycox/blob/master/examples/03_network_architectures.ipynb" for the 3D CT dataset. Is it possible to achieve this with the same structure. I want to use Autoencoder model for extracting the encoded features and then pass it to the LogisticHazard model. I need your help in knowing whether can this be done for the 3D CT dataset (instead of the tabular dataset) or not?

Thanking you

Nikhil kumar