Efficient way for dataset creation of vasculature segmentation of fundus images

Hi Everyone,
I am trying to create a dataset for artery/vein segmentation of fundus images.Is there any time-efficient way/tool of creating it?
See below, what I am trying to achieve -

1 Like

Hi @jobs ,

Have a look at tools for filament/neurite tracing, that challenge is quite similar to your problem:
In Fiji there is Simple Neurite Tracer . Try with working on a single channel and inverting the image.
If you have Imaris the Filament Tracer works also well.

You could also try binarization (for example by thresholding) and then skeletonization but that way it would be very hard or impossible to separate the different veins I’d say.

3 Likes

Hi @jobs,

In this paper (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4583126/), see Table 4:

There are a few methods there for the analysis of similar looking data sets.

The paper is about the chicken chorioallantoic membrane experimental model - not fundus. But the images have many similarities so the approaches may work well too.

Cheers
Luciano

2 Likes

Thank You for answering and Sorry for the late response - I got engaged in some Covid stuff.
I’ll check it and will get back to you.

Thank You for answering and Sorry for the late response (I got engaged in some Covid stuff).
I’ll check it and will get back to you.

1 Like