2d U-net on 3D nii CT data

[Yushen12sigma]

Hi, Niftynet team,

I'm trying to train a 2D unet with every single slice of a 3D CT scan.

According to your instruction,
https://niftynet.readthedocs.io/en/dev/window_sizes.html
"setting spatial_window_size = (h, w, 1) will generate a 2.5D windows, "

but in the instruction of configuration file
https://niftynet.readthedocs.io/en/dev/config_spec.html
"Array of three integers specifies the input window size. Setting it to single slice, e.g., spatial_window_size=64, 64, 1, yields a 2-D slice window."

So which is is correct? or should I chop the 3D scan into 2D images by myself?

Thank you!

[1160914483]

Hello, have you solved the problem?

[ericspod]

The spatial_window_size should have the dimensions of data you're working with, so for 2D it would (h, w). The unet_histology demo illustrates this using the 2D unet implementation.