TorchIo to train large 2D images .

[zirid]

Is it possible to use TorchIo to train large 2D images as patches?
Are there any examples of using the library with 2D images ?

[fepegar]

Hi, @zirid. It should be possible. Maybe @MirenLurBarquin, @jmlipman, or @jxchen01 can give you some advice if you need it.

[MirenLurBarquin]

Hi, @zirid. I have not direct experience with 2D images. Nevertheless, as far as I remember, I went through a 3D volume taking 2D patches. Did you try to use the patch-based pipelines in the documentation, and try to set to 1 the last dimension in the patch_size tuple argument of the sampler?