While visualizing the k-space data, I found that many volumes had zeroed-out regions at the edges of k-space.
While this would make the image size larger, this adds no additional information.
I have verified that there are several such cases in the knee data set.
I have not yet checked the brain data set.
Also, all the blanks are in the phase encoding dimension (the last dimension, which has variable length).
Is there a good reason for doing this?
Thank you in advance for helping out.
This padding is an artifact of the way the kspace is acquired. I think you can find the exact dimensions that were padded in the ismrmrd_header e.g.:
hdr = ismrmrd.xsd.CreateFromDocument(data['ismrmrd_header'][()])
enc = hdr.encoding
enc_size = (enc.encodedSpace.matrixSize.x,
enc_limits_center = enc.encodingLimits.kspace_encoding_step_1.center
enc_limits_max = enc.encodingLimits.kspace_encoding_step_1.maximum + 1
padding_left = enc_size // 2 - enc_limits_center
padding_right = padding_left + enc_limits_max