Trouble making this work with other data

Zhiwei-Zhai / Lung-Vessel-Segmentation-Using-Graph-cuts

GNU General Public License v3.0

29 stars 19 forks source link

Trouble making this work with other data #3

Open arttuilmari opened 2 years ago

arttuilmari commented 2 years ago

Hello, I'm having a bit of trouble making this segmentation tool work with other data.

Are you using a binary mask of the airways for the distance transform? Do you think applying matlab's bwdist() on binary mask of airways would produce the correct transformation?
Does this tool work for one lung or both lungs at the same time?
Do I have to extract the lungs from CT image or can I use the whole throaic ct?

Zhiwei-Zhai commented 2 years ago

Hello, I'm having a bit of trouble making this segmentation tool work with other data.

Are you using a binary mask of the airways for the distance transform? Do you think applying matlab's bwdist() on binary mask of airways would produce the correct transformation?

Does this tool work for one lung or both lungs at the same time?

Do I have to extract the lungs from CT image or can I use the whole throaic ct?

The distance trainsform was used to calculate the distance map to airway. By reading the documentation of bwdist(), it seems that it could produce similar results.
It is possible to work with both lung, if your PC have enough memory. As constructing the graph will require a lot of memory, I do vessel segmentation for each lung.
I do think lung extraction is nessery, which can reduce the graph size and remove a lot of false positive outside lung.

dassagichnicht commented 2 years ago

Hello,

I try to recreate your work with own CT data. But I'm struggling to calculate the strain energy filter.

I tried using a 3D DICOM file as input and a TIF as output, but the output file is 2D and not 3D.
Using a single slice of the CT results in the error, this combination of pixel type and dimension is not supported! dimension = 2 pixel (component) type = float
Using a dicom 3D image as input and a DICOM as output results in the error itk::ERROR: GDCMImageIO(0000027A155E95A0): A Floating point buffer was passed but the stored pixel type was not specified.This is currently not supported
Using a 3D TIF as input results in the error The NumberOfComponents is larger than 1! Vector images are not supported for this filter.

What input and output format did you use to the script?

Thank you very much for your work and help!

Zhiwei-Zhai commented 2 years ago

I used mhd or mha format. Dicom can be converted to mhd/mha which saves 3D information. ITK or SimpleITK can be used for data fromat transfering.