If you want to determine noise type with kurtosis, skewness, do you calculate from background (since foregrounds may include tumors) ???

Hi,

I want to evaluate kurtosis and skewness features of images.

I want to calculate kurtosis and skewness to determine the noise type in medical images, which inlcude lesions.

Should I do the calculations from background ? Or, should the calculation make from randomly chosen regions ? or, what is your suggestion?