reconstruction of the nifti mask image

[vergs98]

Hi, your code is really helpful, but I didn't understand how you can reconstruct the mask after your applied the model. Because you're taking just the images with a lesion, but then you should be able to reconstruct the entire MRI image. Also because the goal of this algorithm should be to predict the lesions for brain images without a mask.
So how can you transform your output (png) in a nifti image again?

[vergs98]

I have another question if I may: how can I predict an image in which I don't know the output? Because in the class dataset I have to give the mask, but in the case I have to predict the lesion in a new image, I don't have the mask.
How can I do it?