As with most NMR data, the transformation is usually accomplished either within the software provided by the spectrometer vendor or by third-party software such as MNova from Mestrelab. I also think that these vendors use some proprietary routines for getting a good 2D plot (for example, minimizing the "streaking" one encounters when two peaks are close in chemical shift but have different diffusion constants).
Is there a particular reason you want to do this manually? Do you have access to vendor-supplied software?
Thanks. - Josh
Thanks for answering. I have access to the Bruker topspin student version. I'm studying polymer behaviour in solution, in this case a polysaccharide in water. I've managed to obtain the DOSY 2D plot relating the chemical shift (i.e. the peaks) and their diffusion coeffcients, but what I want is to plot the diffusion coefficient vs intensity for a specific peak, thus obtaining some sort of distribution of diffusion coeffcients.
Thanks for your time.
Ah. This analysis is fairly sophisticated. The distribution of diffusion coefficients for a given resonance is related to the "linewidth" of its DOSY peak in the indirect/diffusion dimension. (A narrow distribution, i.e. a single species would result in a sharp DOSY peak, but a broad range of MW's would give rise to a broad DOSY peak in the D dimension).
A quick search for "dosy measurement of polydispersity" yields this article, which you may find handy:
Without trying to go into the nitty gritty detail of your processing, perhaps you should investigate whether the tools you need are part of the DOSY Toolbox, a free DOSY analysis package produced by the Nilsson lab at the University of Manchester:
The description on that page indicates their software handles multi exponential decays, so it may be able to handle what you seek.
I'm a firm believer in using software that other people have written to solve problems that many people face - no sense in rewriting software every time a question needs answering.
Good luck! - Josh