How can I find the FWHM of an imported non-symmetric matrix dataset?

I have imported different 300 x 300 uint16 datasets into matlab, consisting of 3 peaks and a little noise. I want to quantify the resolution and compare the datasets. The peaks have a better resolution in 1 dimension than the other, so it isn't symmetrical. I want to know the FWHM for each peak in 2 directions, the one with the highest value and the one with the lowest value (perpendicular to each other). How can I find the full width half maximum (FWHM) in a matrix, and in different directions?

[dpb]

That's an interesting question -- no otomh answer but a question. Is the desired orientation normal/parallel to the axes or are you looking at the general orientation?

Is there any theoretical distribution underlying the data as a clue for a functional form?

Any chance of attaching a smaller segment (say area around one of the peaks only for working with?