ABSTRACT
PURPOSE: The quantitative assessment of renal blood flow (RBF) may help to understand the physiological basis of kidney function and allow an evaluation of pathophysiological events leading to vascular damage, such as renal arterial stenosis and chronic allograft nephropathy. The RBF may be quantified using PET with H(2)(15)O, although RBF studies that have been performed without theoretical evaluation have assumed the partition coefficient of water (p, ml/g) to be uniform over the whole region of renal tissue, and/or radioactivity from the vascular space (V(A). ml/ml) to be negligible. The aim of this study was to develop a method for calculating parametric images of RBF (K(1), k(2)) as well as V(A) without fixing the partition coefficient by the basis function method (BFM). METHODS: The feasibility was tested in healthy subjects. A simulation study was performed to evaluate error sensitivities for possible error sources. RESULTS: The experimental study showed that the quantitative accuracy of the present method was consistent with nonlinear least-squares fitting, i.e. K(1,BFM)=0.93K(1,NLF)-0.11 ml/min/g (r=0.80, p<0.001), k(2,BFM)=0.96k(2,NLF)-0.13 ml/min/g (r=0.77, p<0.001), and V(A,BFM)=0.92V(A,NLF)-0.00 ml/ml (r=0.97, p<0.001). Values of the Akaike information criterion from this fitting were the smallest for all subjects except two. The quality of parametric images obtained was acceptable. CONCLUSION: The simulation study suggested that delay and dispersion time constants should be estimated within an accuracy of 2 s. V(A) and p cannot be neglected or fixed, and reliable measurement of even relative RBF values requires that V(A) is fitted. This study showed the feasibility of measurement of RBF using PET with H(2)(15)O.