Hello ARTS forum,
I'm just beginning processing of a large (1000+ session) dataset through this pipeline. Within my first few runs I'm seeing quite a few failures. One that stands out appears to be trouble with ITK missing an image output (or trying to use an all-zeros image) during the diffusion anisotropy phase (see the screenshot as well):
*** Calculating diffusion anisotropy features...
Registering subject to IIT template (may take a few minutes)...
All_Command_lines_OK
Using double precision for computations.
Exception Object caught:
itk::ExceptionObject (0x558f230)
Location: "unknown"
File: /src/ants/build/ITKv5-install/include/ITK-5.0/itkImageMomentsCalculator.hxx
Line: 156
Description: itk::ERROR: ImageMomentsCalculator(0x55aed10): Compute(): Total Mass of the image was zero. Aborting here to prevent division by zero later on.
Based on other people's encounters with this particular ITK error, I'm guessing the DTI I'm feeding the pipeline is missing the "IIT template" somehow, resulting in a null image, but that's a total guess. Any suggestions here would be much appreciated!
Thanks,
Bill
Hello Bill,
Can you verify that the FA map was generated correctly? In the dataset folder, it should be at FA_processing/tbss/FA/reorient_input_FA.nii.gz
The IIT template is a file that is unmodified and in a fixed location throughout the run, so it is unlikely to be missing or empty.
Best,
Arnold
Yes, the file exists, but it's a completely black volume.
I'm attaching an annotated screenshot of the DTI directory from one of these failures, in case that helps with context. Green Arrows look like a normal DTI image, yellow arrows are "ok" (they're noise, so I'm not sure what I'm looking at, but I'm accepting it), and red arrows are featureless black volumes.
Thanks,
Bill
That context narrows it down to the possibility that the diffusion tensors were not calculated correctly. Can you share the processing log? I will be looking for an error or warning that happens earlier in the pipeline than when you screenshotted.
-Arnold
Attached. Thanks, Arnold!
Bill
Unfortunately, the processing log does not provide any more insight. Can you send the import_up_proc folder to me through email? My email is Arnold_Evia@rush.edu.
Does this error happen for all or some of your data?
-Arnold