Edge Spread Function Calucaltion

Hi everybody,

I have a DICOM CT image (now stored as a 512x512 matrix) containing a Polymer Gel dose distribution. Is there a way to calculate the edge spread function to estimate/quantify the degree of edge blurring in this image.

Thanks,

Basim

Answer 1

[Bjorn Gustavsson]

Yes, you can take a look at blind deconvolution, matlab's function deconvblind does this. In my experience the success of that method depends a bit on what your image looks like, its noise-level and what structures you have. If successful that'll give you an estimate of the point-spread-function which you then easily can transform to an edge-spread-function with the Abel transform.

HTH

[Basim Kakakhel]

Thanks, Bjorn

For the blind deconvolution what should be the size of the initial point spread function given that I have a 2D CT image.

Basim

[Bjorn Gustavsson]

Well, that depends on your image, so a bit tricky to give a good advice. In my experience, I've had images with typical point-spread-function sizes varying from 1 pixel wide to some 10 pixels wide. For that sort of images I've used something like a 2-D Gaussian with a few pixels wide in an 7-by-7 element matrix for the good images and perhaps a 15-by-15 or so for the images with the wider PSF...

Not much of a strict rule for you to follow - when it comes to this type of problem, I think you have to test/experiment and see what works best,

HTH