Is there an input data dimension requirement?
Maxwell-X-Chen opened this issue · comments
I have a 3D whole brain dataset, compiled of 2D brain slices.
I am trying to run the program with two of these brain slice files of.Tiff format.
(I have changed the get_itk_image() function from utils.py to Tiff compatible, so the file format should not be the problem.
- OR does "SetImageIO("TIFFImageIO")" is not actually helping to read .Tiff file, and instead I should use packages like Tifffile?
However, I keep getting this error...
ValueError: Error when checking : expected main_input to have shape (2, None, None, None) but got array with shape (1, 60, 60, 60)
Any help would be appreciated! Could it be that the network only takes in **stacked Tiff files?**
Error:
Output:
Hi Maxwell-X-Chen,
Thanks for the feedback, I recommend running the VesSAP with the default parameters and inputs to see how it works and replace data loading modules afterwards. The easiest is to open the (ideally 16bit unsigned) 2D TIFFs as a stack in Fiji (from fiji.sc, with the Nifti IO plugin from https://imagej.nih.gov/ij/plugins/nifti.html) using File, Import, Image Sequence and then File, Save As, NifTI-1 so it is directly compatible with the image loading into VesSAP.