Concatenate layer doesn't match shapes

[joaomamede]

Hi,I'm testing MIScnn for IF data (grayscale input) and the segmentation has 3 labels (background =0,1,2).

I'm having an error when running

model = Neural_Network(preprocessor=pp, loss=tversky_loss, metrics=[dice_soft, dice_crossentropy],
                       batch_queue_size=3, workers=3, learning_rate=0.0001)

ValueError: A Concatenate layer requires inputs with matching shapes except for the concatenation axis. Received: input_shape=[(None, 4, 128, 128, 128), (None, 5, 128, 128, 128)]

I tried all kinds of shapes such as the whole image (1024,1024) half (512,512), a quarter etc. with different overlaps and I always have this error. I also tried without swapping axis (keeping Z as axis=0).

The same code with nii transformed dataset has the same problem. Another set from CT scans that are (200,512,512) do not have this problem.
I'm attaching the html of the whole notebook. Any hints that could solve this problem?

LaminTest-Dictionary.pdf