search

JonasLamy / LiverVesselness

Repository for Vesselness algorithms
Other
44 stars 9 forks source link

Issue with Using Sato and OOF Filters in MRI #7

Open camyj2010 opened 8 months ago

camyj2010 commented 8 months ago

Hello, I am currently facing challenges with the utilization of the Sato and OOF filters for segmenting perivascular spaces (PVS) in MRI images. While other filters are functioning correctly, the Sato and OOF filters exhibit inconsistent behavior. When attempting to implement these filters with specific parameters, the following issues arise:

Sato filter I used this parameters ./Sato --input /home/camyj2010/Images/T1.nii.gz --output /home/camyj2010/Images/outputSato.nii.gz --sigmaMin 1.4 --sigmaMax 2.8 --nbSigmaSteps 4 --alpha1 0.3 --alpha2 1.3 Error encountered: The Sato filter intermittently fails to produce the expected output. Sometimes, it executes successfully, while other times, it fails with this error. Using the same parameters Captura de pantalla 2024-03-08 161253 OOF filter I used this parameters ./OOF --input /home/camyj2010/Images/T1.nii.gz --output /home/camyj2010/Images/outputOOF.nii.gz --sigmaMin 0.6 --sigmaMax 1.6 --nbSigmaSteps 4 --sigma 0.2 Error encountered: The OOF filter generates different errors inconsistently. Sometimes it produces the same type of error as the Sato filter., and other times, it generates a different error upon execution. imagen_2024-03-08_162024339 I seek guidance to identify the underlying causes of these inconsistencies. Could there be overlooked steps or parameters contributing to this problem, or could it be a problem because of the computer capability on which it is being run? Any assistance in resolving these sporadic errors would be greatly appreciated.

Thank you

JonasLamy commented 8 months ago

It is strange indeed,

The first error seems to indicate that the ITK image reader fails to identify your .nii.gz file, and instead identify the file as a Folder, thus switching to dicom reading (jpeg files in a folder). Can you try extract the .nii.gz to .nii only and see if the issue arises ?

The second one is stranger...But the error states that two pieces of data have different coordinate system. In the images I've used, the coordinate system is an Identity matrix (modulo sign).

My guess is that some of your data are ill defined...

Also, there is a hidden "--inputIsDicom" option that you can add to read dicom folders, if the reading works with this flag, then your file format is definitely wrong.

Keep me updated with you progress.