Water diffusion in biological tissues is not Gaussian and signal attenuation is not monoexponential with b-value [1]. Approaches to deal with this behavior include the bi-exponential model [1,2], the Karger model [3], and Kurtosis approach [4]. We formulate an ODE model for diffusion MRI signal that is more general than Karger model, valid for more general diffusion gradient shapes and gives a good approximation to the ADC and Kurtosis. Given DMRI signals before and after cell swelling, we can estimate the amount of cell swelling after numerically solving an ODE system.