The objective of the study is to develop a method for non-invasive quantification of renal medullary blood flow in human subjects.

Methods: 

Nine healthy volunteers were placed on a standardized diet for four days. On the 5th day a low dose CT-scan was performed and blood flow determinations by PET scans were carried out with H215O (1 MBq) under three conditions: i) baseline, ii) glycerylnitrate infusion (0.3- 1.2 mg/kg body weight) and iii) L-NMMA bolus injection (3 mg/kg body weight). Data were analyzed the following way: By CT-scanning, the upper and lower kidney poles were identified, and selected as volumes of interest (VOI). Blood flow of the VOI was calculated based on the Kety-Schmidt one-compartment model. Using computer programming, tissue layers with a thickness of one voxel were eliminated from the outer cortical part of the VOI in a stepwise fashion starting with the most superficial layer. After removal of each layer, the blood flow of the remaining part of the VOI was calculated.

Results: 

The blood flow of the VOI decreased as a function of the number of cycles of voxel peeling. After peeling of 4–5 layers of (mostly) cortical tissue, the blood flow of the VOI became insensitive to removal of additional layers. This volume-insensitive flow was 1.96 ± 0.44 ml/g tissue/min under control conditions and increased (p< 0.05) to 2.52 ± 0.46 ml/g tissue/min during infusion of glycerylnitrate. Injection of L-NMMA caused a decrease (p< 0.05) to 1.33 ± 0.37 ml/g tissue/min.

Conclusion: 

The voxel peeling technique allows identification of an inner region of the kidney in which low blood flow values are insensitive to further volume reduction. It is reasonable to assume that the measured values reflect renal medullary blood flow.