DCE analysis - Ktrans is calculated only in a few small points

How to get Ktrans from DCE multivolume?

At random ROI level Ktrans is calculated only in a single small points

In general, you should not expect to fit model to an arbitrary signal, since you may not have contrast uptake, or you may have significant artifacts, or high noise. If you try to fit model to the noise, you cannot expect more than noise as the output!

Looking at your specific dataset, it seems that you have very few time points (just 6?), which probably is insufficient for PK analysis.