AbstractIntravoxel incoherent motion (IVIM) theory in MRI was proposed to account for the effect of vessel/capillary perfusion on the aggregate diffusion weighted MR signal. The prevalent IVIM modeling is based on equation-1: SI(b)/SI(0) = (1 -PF) × exp(-b × Dslow) + PF × exp(-b × Dfast) [1] where SI(b) and SI(0) denote the signal intensity of images acquired with the b-factor value of b and b=0 s/mm2, respectively. We recently reported that, for the liver and likely for other organs as well, IVIM modeling of the perfusion component is constrained by the diffusion component, with a reduced Dslow measure leading to artificially higher PF and Dfast measures. With higher b-value associated lower image signal of the targeted tissue, Euqation-1 is focused on describing the signal decay pattern along increasingly higher b-values by three IVIM parameters. Signal intensity at each b-value (i.e., SI(b)) is normalised by the signal intensity of b=0 image (i.e., SI(0)). We noted an apparent problem for Euqation-1. For example, if we want to compare the IVIM parameters of the normal liver parenchyma and a liver tumor, following Euqation-1 we will take the assumption that the SI(0) of the normal parenchyma and the tumor are the same and considered equally as 1 (or 100) for the biexponential decay modelling. However, this assumption is invalid for many scenarios. From our liver IVIM database of 27 healthy female subjects, we chose six of the youngest subjects (20-27 yrs) and six of the oldest subjects (58-71 yrs) and measured the signals of the liver and left erector spinae muscle on b=0 and 2 s/mm2 images. The results show, while there was no apparent difference of left erector spinae muscle signal among the young and elderly groups, the elderly group’s liver SI(0) is approximately 20 % lower than that of young group. This difference skewed the ratios of various SI(b)/SI(0) and the followed IVIM parameter determination. The general trend is that lower liver SI(0) is associated with lower Dslow and higher PF and Dfast. If IVIM bi-exponential decay fitting starts from a very low non-zero b images (such as b=2 s/mm2 images), this problem persists. We performed an additional analysis of our IVIM database of five cirrhotic livers and the results show SI(b=2) of cirrhotic right liver is positively associated Dslow (Pearson r=0.687), and negatively associated with PF (Pearson r=-0.733). Though the examples we used in this letter are on liver aging and liver fibrosis, the points discussed are expected to be generalisable to other pathologies and to other organs.