Dynamic gadoxetate contrast-enhanced MRI yields spatial hepatocellular function through hepatic extraction fraction map. This allows well-functioning hepatocyte sparing in radiotherapy to avoid radiation-induced liver toxicity. However, the major challenge of using this parametric map in a clinical practice for normal function sparing is the lack of standard method to determine liver function at a voxel level within the same patient. As such, population-based kernel density function was proposed to deal with this problem to predict voxel-based probability of liver function. This novel approach also allows derivation of functional probability map that could be used for radiation beam guidance in function-based radiation treatment planning.