You could try a Frangi filter before thresholding:

http://uk.mathworks.com/matlabcentral/fileexchange/24409-hessian-based-frangi-vesselness-filter

But it introduces artifacts where vessels branch.

You could also try the following (Su et al. 2014), then threshold; or using the following (Sun and Vallotton 2009) to obtain the centre line, then find the thickness of vessels.

R. Su, C. Sun, C. Zhang, and T. D. Pham, A New Method for Linear Feature and Junction Enhancement in 2D Images based on Morphological Operation, Oriented Anisotropic Gaussian Function and Hessian Information, Pattern Recognition, 47(10):3193-3208, October 2014.  http://dx.doi.org/10.1016/j.patcog.2014.04.024

C. Sun and P. Vallotton, Fast Linear Feature Detection Using Multiple Directional Non-Maximum Suppression. Journal of Microscopy, 234(2):147-157, May 2009. http://dx.doi.org/10.1111/j.1365-2818.2009.03156.x

Dear Muhammad,

I have a system which works very well with carotid arteries. If you can send me an example image, I can try and segment it. 

Kind regards,

Baris.

Thanks Baris

Pl. find attached image also reply me techniques you implemented in your system.

Best Regads

Muhammad Nadeem Ashraf

Dear Muhammad,

I have tried it and unfortunately it doesn't seem to work very well with this image. I could suggest looking at the waterfall/watershed segmentation.

Kind regards,

Baris.

I applied the Frangi filter to the green channel. Removing the speckles and the outer rim should be trivial, but note the artifacts introduced where the vessels bifurcate!

Changming, could you give it a try too?

http://www.isi.uu.nl/Research/Databases/DRIVE/results.php

Also check for papers by Chuck Stewart (CV Stewart) and Dual Align http://www.dualalign.com/index.php; although their tools focus on retinal registration.

Dear Baris,

would you pl. reply me techniques, even unsuccessful, to segment these vessels?

Thanks

Nadeem 

Lambert, The paper by Su et al. 2014 I mentioned earlier should be useful for junction/bifurcation regions. There are three retinal examples in the paper (Fig. 14) comparing several methods including the Frangi filter. I do not have a good grasp on the codes (of the paper to run on other images) which was done by my former student. Regards, Changming.

Dear Muhammad,

The technique I tried is the one described in the below paper. It was designed for segmenting carotid arteries from ultrasound image sequences:

http://downloads.hindawi.com/journals/isrn/2012/179087.pdf

I am not familiar with retinal scans. Changming's suggestion might be more useful for you.

Kind regards,

Baris.

Thank you for your reaction, Changming.

I also suggest to give a look to this paper:

http://www.sciencedirect.com/science/article/pii/S1361841514001364

and its Matlab implementation:

http://www.mathworks.com/matlabcentral/fileexchange/49172-trainable-cosfire-filters-for-vessel-delineation-with-application-to-retinal-images

 Hi! please, have a look at the method described in our recent publish paper: "Leveraging Multiscale Hessian-based Enhancement with a Novel Exudate Inpainting Technique for Retinal Vessel Segmentation"

http://vampire.computing.dundee.ac.uk/publications.html