Hypothetically speaking, if there is a fmri preprocessed ALFF nifti file of an individual with ASD (3 dimensional), if i applied noise ejection or random cropping or changing the sharpness or blur of the image to eventually create new augmented data, would that mean that the label of the newly created data is still valid to be labeled as ASD, or would it be labeled TD since we shifted with the information. What do you recommend as data augmentation methods and how would you label it?
Side question: nifti is a 3D image, whats the best way to cobvert it to 2D without losing a lot of info?
The ASD/TD label should only be associated with the subject the data came from, without anything to do with processing done to the image.
Also, ALFF is calculated after a lot of early preprocessing (e.g., denoising, motion correction), so what is the motivation for doing this augmentation?
What is application here? I suppose, if you have a i x j x k size image that you resize it to something like i x (j x k) so you would have a lot of brain slices represented along the y-axis, but it would help to know what you have in mind.