In univariate processing of DOSY experiments, each signal is treated as independent, although all signals of a given component should, in the absence of exchange, show identical diffusion behavior. Multivariate methods exploit this covariance by fitting the entire spectra (or chosen parts thereof) simultaneously, with the aim of increasing the probability of successfully extracting component spectra.
Among the methods developed in this group are SCORE, LOCODOSY, T1-DOSY and Timecourse DOSY