> Hi Sumathi,
>
> I would say this is still a somewhat open question, but the "usual"
> approach is to use a vesselness filter on the image to extract line-like
> structures. You might be interested to read:
>
> Frangi, Alejandro F., et al. "Multiscale vessel enhancement filtering." *Medical
> Image Computing and Computer-Assisted Interventation—MICCAI’98*. Springer
> Berlin Heidelberg, 1998. 130-137.
>
> That approach uses Hessian parameters to identify line-like structures in
> the image. There are some open implementations in ITK and I've seen some
> Matlab ones as well.
>
> However, you should be aware of some caveats:
> - The SWI high-pass filter size will impact the vessel width you find
> - On SWI in particular, veins are much more visible than arteries due to
> lack of deoxygenated hemoglobin in the latter. So, you will be more likely
> calculating vein density than blood vessel density.
> - Again, because of the contrast, it's difficult to distinguish on SWI
> whether veins are actually less dense or whether the related tissue is
> simply extracting less oxygen.
> - SWI is non-local and has a "blooming" effect, so your measures will be
> relative, and veins will generally appear wider than they are in reality.
>
> Best,
> Mike
>