Renal blood flow using [15O]H2O
Renal blood flow (RBF) is very high: kidneys receive 20-25% of the resting cardiac output, but their weight is less than 1% of body mass. Renal perfusion, ∼400 mL/(min*dL), is higher than perfusion in the myocardium, liver, or brain. Cortical blood flow composes ∼80-90% of the total renal blood flow, because the mass and perfusion of medullary tissue is markedly smaller. However, quantitative measurement of regional distribution of blood flow within the kidneys with traditional methods, such as washout method, has been difficult and prone to errors (Grünfeld et al., 1974; Aukland, 1975; Blaufox, 1989). Effective renal plasma flow has been commonly assessed as urinary clearance of para-aminohippurate (PAH); conversion of that into quantitative RBF requires that the extraction ratio of PAH (∼0.9 in healthy men) is known. Without extraction correction, renal clearance provides only the minimum blood flow. 131I-hippuran has even lower extraction ratio than PAH.
[15O]H2O can be used to assess renal perfusion (Peters et al., 1972; Inaba et al., 1989; Nitzsche et al., 1993; Juillard et al., 2000; Alpert et al., 2002; Anderson et al., 2003; Kudomi et al., 2009; Damkjær et al., 2010; Koivuviita et al., 2012; Rahman et al., 2019; Normand et al., 2019), also in transplanted kidney (Päivärinta et al., 2019). Due to the short half-life of 15O, renal perfusion measurements can be repeated during one imaging session. This enables the assessment of microvascular function of the kidneys by measuring renal perfusion at baseline and after a vasodilator, such as enalapril, and calculating renal flow reserve (RFR) as the difference (Päivärinta et al., 2018).
Sympathetic nerve activity and even mild exercise reduces RBF, especially in heart failure patients (Middlekauff et al., 1997a, 1997b, 2000, 2001). MRI measurement has shown that hand-grip exercise reduced perfusion in renal cortex and medulla, and medullary oxygenation was increased because of decreased GFR (Haddock et al., 2018). Heat and cold stress activates sympathetic system, reducing renal blood flow and increasing renal vascular resistance. Stress during the PET scan should be avoided; Middlekauff et al (1997a) reported that induced mental stress reduced cortical renal blood flow by ∼30%. Sodium and water intake, and medications, must be standardized in renal studies to avoid confounding effects on GFR, perfusion, and oxygen consumption.
Renal perfusion decreases by age, and is higher in subjects on a high sodium intake, but the RBF response to salt intake is blunted in older individuals (Hollenberg et al., 1974).
Renal artery branches directly from abdominal aorta, and branches in renal pelvis into interlobal arteries. In 20-30% of normal individuals, additional arteries supply (usually) the lower pole of the kidney. Interlobar arteries branch into arcuate arteries, which reside at the border of cortex and medulla, and practically define the border of cortex and medulla. Arcuate arteries give rise to interlobular arteries that extend through the cortex toward the surface of the kidney. Interlobular arteries branch into afferent arterioles, which lead into glomeruli. A single efferent arteriole directs the blood from each glomerulus to the peritubular capillary network in the cortex and vasa recta in the medulla. Capillary network in the kidney is very dense, especially in the medulla: in mice, capillary density and capillary surface area are 7400/mm2 and 1400 cm2/cm3, respectively, in the renal medulla, which are clearly higher than in the mouse heart muscle (5300/mm2 and 1000 cm2/cm3) and liver (4200/mm2 and 800 cm2/cm3); in renal cortex the values are 4500/mm2 and 850 cm2/cm3, respectively (Kety, 1951).
All of the medullary blood flow is derived from the efferent arterioles of the juxtamedullary nephrons. Renal blood flow to the cortical glomeruli (cortical perfusion) and juxtamedullary glomeruli (cortical and medullary perfusion) are differentially regulated (Evans et al., 2004). Juxtamedullary afferent and efferent arterioles are larger than cortical arterioles. The efferent arterioles that descend from the juxtamedullary glomeruli to the medulla have also thicker endothelial layer, and the arterioles descend in parallel, with lateral branches forming capillary plexus. Medullary vasa recta has markedly larger diameter than cortical peritubular capillaries. Descending vasa recta capillaries have contractile smooth muscle-like pericytes, which constrict and dilate the capillaries and regulate medullary perfusion (Kennedy-Lydon et al., 2013). Vasoconstrictors (including angiotensin II, endothelin, and noradrenaline) mainly affect cortical blood flow (Evans et al., 2004). In healthy subjects administration of angiotensin II led to renal hypoxia (Schachinger et al., 2006). Vasodilators (including NO and bradykinin) increase medullary blood flow. Vasopressin reduces, and adenosine and acetylcholine increase medullary blood flow. Increased luminal flow, in physiological range, in medullary thick ascending limbs of Henle leads to increased production of ROS, which diffuse to vasa recta capillaries and reduce medullary blood flow (Cowley Jr et al., 2015).
Medullary venous vessels are larger and more numerous than arterial vessels.
Kidneys filter ∼100-200 L of fluid each day (or 1.0-1.2 L/min), but normally less than 1% (or ∼1 mL/min) of the filtered water is excreted in urine. About 20% of the plasma is filtered at the glomeruli from blood vessels into the Bowman capsule and proximal tubules. Proximal tubule reabsorbs 60-90% of the water and solutes. Diffusion of water is driven by osmotic pressure gradients, and is facilitated by Aquaporin AQP1 in the proximal tubules and by Aquaporin AQP2 in the connecting and collecting ducts. The osmotic gradient for water reabsorption is formed by active reabsorption of Na+, Cl-, and glucose. Hydrostatic pressure in the proximal tubular tissue, and high protein content in the peritubular capillaries, drives the water into the blood. The loop of Henle reabsorbs ∼15% of the filtered water; the descending limb has AQP1 channels, but the ascending limb is impermeable to water, as is the initial segment of the distal tubule. Since Na+, Cl-, and other solutes are actively reabsorbed in the ascending limb and the distal tubule, the remaining fluid is dilute, as compared to plasma. The late distal tubule and collecting duct cells possess AQP2, AQP3 and AQP4 channels, and reabsorb variable proportion (∼8-17%) of the water. The collecting ducts are responsible for maintaining the water balance of the body, since the osmolality of the tubular fluid that they receive is fairly constant in all conditions, but the produced urine can be very hypo- or hyperosmotic. Vasopressin level regulates the water reabsorption in these segments, as well as the medullary blood flow.
Ascending vasa recta capillaries in the medulla are highly permeable to water, and are responsible for removing the water that is reabsorbed from the tubular fluid in loops of Henle and in collecting ducts.
[15O]H2O bolus model
The analysis method of renal blood flow (RBF) is based on one-tissue compartment model. Model parameters for the kidney are RBF, p (ρ, partition coefficient, K1/k2), and RBV (renal arterial blood volume). Wash-out of [15O]H2O, injected into the renal artery, is mono-exponential during the first 60 s (Inaba et al., 1989). The first [15O]H2O kidney study was done using external detectors (Peters et al., 1972), and in that two compartments were detected, with compartmental distributions of 55% for the fast component (cortex) and 45% for the slow component (medulla) with perfusion 370 and 55 mL/(100g * min), respectively (Szabo & Mathews, 2006).
RBF can be estimated from regional time-activity concentration curves (TACs), or from dynamic PET image to produce perfusion image. Further, volumetrically normalized images could be used for statistical parametric mapping (Rahman et al., 2019). Arterial blood curve can be measured from arterial line as model input function, but input could also be measured from the image from ROI placed on ascending aorta.
Inaba et al. (1989) applied the radiowater model to assess renal perfusion from regional data in 8 subjects. They calculated renal plasma flow (RPF) as
, where HCT was the systemic hematocrit, because renal arterial hematocrit is not known. They apparently did not account for the RBV, which causes overestimation of RBF (Kudomi et al., 2009), although they had measured RBV using [15O]CO. The RPF was 171±61 mL/(min * 100 g); assuming that HCT=0.45, that means that RBF was about 3.1±1.1 mL/(min*g). The apparent partition coefficient (ρ) was 0.66±0.10, and it correlated with RPF. RPF correlated with glomerular filtration rate (Inaba et al., 1989). Nitzsche et al (1993) added RBV parameter to the model, and used it to validate renal perfusion assessment using [13N]ammonia; their radiowater results for cortical RBF and RBV were 4.70±0.28 mL/(min*g) and 0.16±0.11 mL/g, respectively. Middlekauff et al (1995) reported that cortical RBF was markedly lower in patients with advanced heart failure (2.4±0.1 mL/(min*mL)) than in healthy controls (4.3±0.2 mL/(min*mL)). Juillard et al (2000) validated the method, assuming K1/k2=1, against microspheres in pigs. Microspheres cannot be used to measure medullary blood flow, because they are trapped in the glomeruli (Ericson et al., 1979). Using the same method, Juillard et al. (2002) measured RBF of 2.2±0.2 mL/(min*g) in men with hypertension and moderate CKD; ACE inhibitor quinaprilat increased RBF by ∼20%. Alpert et al. (2002) obtained RBF of 3.4±0.4 mL/(min*g) in healthy subjects and 2.1±1.1 mL/(min*g) in patients with CKD; probenecid did not affect RBF. Rahman et al (2019) reported lower K1, k2, and VA in patients with chronic renal failure than in healthy subjects. Normand et al (2019) did renal radiowater PET in order to validate [1-11C]acetate for assessing renal perfusion in human subjects; delay and blood volume were accounted for in the model, and the measured cortical perfusion was 3.29±0.65 (range 2.00-4.70) mL/(g*min); after protein load, RBF increased ∼20%. The test-retest reproducibility of K1 in healthy subjects was 0.36 with an ICC of 0.64 (Normand et al., 2019).
Nalin et al (2014) reported renal perfusion values ∼2.3 mL/(mL*min) in healthy pigs and ∼1.3 mL/(mL*min) in diabetic pigs.
Alpert et al (2002) applied linearized method for calculation of RBF image, but ignored RBV; in our implementation of the linear method (see below) RBV can be included in the analysis. Not accounting for RBV causes overestimation of RBF Also time delay between arterial input and tissue TACs must be corrected, if it exceeds 2 s (Kudomi et al., 2009). Basis functions method is another option for computation of parametric RBF images (Kudomi et al., 2009).
Kötz et al (2009) reported range 0.71-0.82 for partition coefficient of water, and 1.30-1.85 mL/(min mL) for blood flow in the kidneys. Kudomi et al (2009) reported that apparent partition coefficient in healthy subjects was on average 0.35 mL/g and RBV was 0.15 mL/mL. Physiological partition coefficient (pphys) is 0.94 mL/mL (Kudomi et al., 2009). partial volume effect and tissue heterogeneity additionally cause the apparent p to be lower than pphys; this has been studied especially in [15O]H2O brain studies.
The total renal blood volume, including both arterial and venous volumes, can be measured using [15O]CO. Yamashita et al (1989) reported that the blood volume in 15 subjects was in the range ∼10-20 mL/100 g, and correlates with GFR. Renal blood volume is reduced markedly in murine models of CKD, already prior to fibrosis (Ehling et al., 2016).
Kudomi et al (2009) used pphys=0.94. RBF based on k2 is not affected by PVE and heterogeneity, but it is instead affected by glomerular filtration rate (GFR), estimated to be 9.6% of k2 (Kudomi et al., 2009). Similarly, RBF based on k2 will not be reduced by increased fraction of non-perfusing tissue volumes, for example scar tissue, but that will be represented as reduced K1 and K1/k2 (p). For perfusion maps, k2 is more stable than K1 (Rahman et al., 2019).
Cortical RBF has usually been the parameter of interest, but Damkjær et al (2010) have shown that calculation of medullary perfusion is feasible. Based on K1, corrected for extraction fraction 0.85, medullary perfusion was 2.30±0.17 mL/(g*min) at baseline, and 2.97±0.18 and 1.57±0.17 mL/(g*min) during glyceryl nitrate infusion and after L-NMMA bolus, respectively. Cortical perfusion values were 4.67±0.31, 4.66±0.53, and 3.48±0.23 mL/(g*min), respectively (Damkjær et al., 2010).
Due to the relatively low time-resolution of PET, the modelling of fast dynamics of reversible tracer uptake is technically demanding and error-prone.
In animal models of kidney diseases, including renal arterial stenosis, fibrosis, and unilateral ureteral obstruction (UUO), one of the kidneys is affected by the disease model and the other kidney serves as control organ. Because the tissue uptake of radiowater correlates with perfusion (although non-linearly), and arterial input function is the same for both kidneys, tissue uptake ratio between the affected and non-affected kidney provides perfusion ratio without arterial sampling.
In the analysis of pig model of renal artery stenosis Xia et al (2008) calculated perfusion ratio from AUCs of renal curves; AUC was limited to the first peak of the tissue curves to achieve better linearity between tissue uptake and perfusion. Stenosis was deemed successful only if reduction in perfusion rate was ≥10%. The same approach has been used in ischemia-reperfusion pig model (Gulaldi et al., 2013).
Perfusion ratio can be biased, as show in a simulation (Figure 1). However, the method is more robust than traditional compartmental model fitting. In rabbits, perfusion ratio separated UUO animals clearly from sham-operated animals, while there was overlap in the results of model fitting (Figure 2). The bias is mainly caused by the non-linear nature of the relationship between tissue uptake and perfusion; shortening integration time only slightly reduces the bias, and the effect of vascular blood is also relatively small. The effect of vascular volume increases with decreasing perfusion.
Blood data from online sampler
Arterial blood data, collected using on-line sampling system, must be calibrated and corrected for physical decay, dispersion, and time delay. It is recommended that regional tissue TAC from kidney is used in time delay correction, because aorta or even the heart, where high radioactivity concentration appears considerably sooner than in the kidneys, is also located in the PET image, preventing the use of count rate or head curve for time delay correction.
- Draw ROIs on kidneys for delay correction (ROIs need not be accurate) and calculate tissue TACs. Do not save aortic TACs in the same TAC file!
- On MS Windows PC in TPC network, do the corrections for blood data using water_input script. Alternatively, these corrections can also be done using a series of low-level commands.
- Verify visually that the corrected blood TAC is fine and that time delay correction has moved
it to start to rise at the same time as the tissue TACs. Previous
water_inputcommand made a graph of these curves. Alternatively you can create the plot by yourself.
Extraction of arterial blood data from PET image
Instead of using ABSS, blood TAC can be extracted from the dynamic image. With modern PET scanners the image resolution is good enough for using the blood TAC from small ROI drawn into abdominal aorta without any correction for partial volume effect. Rahman et al (2019) used profile fitting method to obtain blood TAC in their kidney radiowater study. Factor analysis could be used to derive the BTAC from ascending aorta and TTAC from cortical tissue (Ahn et al., 2000).
Delay correction is usually not needed for input that is extracted from abdominal aorta, and usually the time difference between it and renal tissue is less than 2 s.
Calculation of RBF image
If you have the PET images in DICOM format, convert them to ECAT format.
with basis function method (BFM):
RBF image can be calculated with basis function method
(Kudomi et al., 2009;
Rahman et al., 2019) using
rbf_bfm_h2o, version 0.01 or later, in Solaris terminal window on SUN or PC platform
(if the Solaris workstation where this software was installed happens to be working).
For example, if the dynamic PET image file name is us345dy1.v, and the pre-corrected arterial blood curve is us345ab.kbq, you would enter the following command:
rbf_bfm_h2o us345ab.kbq us345dy1.v 1. us345rbf.v us345vd.v us0345va.v 10.
This command will create RBF, p, and RBV images as us345rbf.v, us345vd.v, and us345va.v, respectively. All images of inputs and outputs are ECAT7 format. Unit of RBF is [ml blood * min-1 * (ml renal tissue)-1] and unit of RBV is [ml blood * (ml renal tissue)-1].
K1 image, and optionally k2, K1/k2 and RBV images, can be computed using imgbfh2o in Windows, Linux, or macOS command prompt window. Dynamic image data can be in ECAT, Analyze, or NIfTI format.
For example, if the dynamic PET image file name is us4044dy1.v, and the TAC from abdominal aorta is in file us4044aorta.tac, and you wish to calculate RBF from k2, with 240 s fit time, you would enter the following commands:
imgbfh2o -k2=us4044k2.v -k2max=8 -pmin=0.1 us4044aorta.tac us4044dy1.v 240 us4044k1.v imgcalc us4044k2.v x 0.94 us4044rbf.v
These commands will create an RBF image us4044rbf.v, where the unit of RBF is [ml blood * min-1 * (ml renal tissue)-1]. These parametric images can be processed further as needed.
BFM is based on a list curves calculated with a set of k2 values.
Program imgbfh2o can calculated the range of
k2s based on optional flow and p limits, or k2 limits can
be set directly using options
BFM will also be implemented in future release of Carimas; please ask Chunlei Han about the status of this project.
with linearized model:
RBF image can be calculated using imgflow in Windows, Linux, or macOS command prompt window. Dynamic image data can be in ECAT, Analyze, or NIfTI format.
For example, if the dynamic PET image file name is us345dy1.img, and the pre-corrected arterial blood curve is us345ab.kbq, you would enter the following command:
imgflow -va=us345rbv.img us345ab.kbq us345dy1.img 180 us345rbf.img
This command will create an RBF image (based on K1) us345rbf.img and
RBV image us345rbv.img. Unit of RBF is
[ml blood * min-1 * (ml renal tissue)-1] and
unit of RBV is [ml blood * (ml renal tissue)-1].
These parametric images can be processed further as needed.
RBF image based on k2 could be calculated with option
as in a previous example.
Calculation of regional RBF
with traditional non-linear model fitting
After ROIs have have been drawn and average time-activity concentration curves have been calculated from dynamic PET images, the regional RBF can be assessed using fit_h2o in Windows, Linux, or macOS command prompt window.
For example, if the renal TAC file name is us345dy1.dft, and the corrected arterial blood curve is us345ab.kbq, you would enter the following command:
fit_h2o -ml us345ab.kbq us345dy1.dft 240 us345rbf.res
This command will create a result file us345rbf.res, which
contains regional RBF, p, and RBV values.
If you remembered to put the option
-ml to the command line, the unit of RBF
is [ml blood * min-1 * (ml renal tissue)-1], otherwise
[ml blood * min-1 * (dl renal tissue)-1].
Sometimes you may need to specify lower and upper limits for the parameters. For this purpose a parameter file needs to be created. It is a text file, which has to be created/edited only once, using for example Notepad in MS Windows or from the command line with command:
Suitable contents for RBF calculation are:
K1_lower := 0 K1_upper := 600 K1k2_lower := 0.2 K1k2_upper := 1.2 Va_lower := 0 Va_upper := 40 Delay_lower := 0 Delay_upper := 0
After this, the program is called with following command:
fit_h2o -lim=rbf.lim -ml us345ab.kbq us345dy1.dft 240 us345rbf.res
with basis functions method (BFM):
BFM can be applied to estimate regional RBF using bfmh2o in Windows, Linux, or macOS command prompt window.
For example, if the renal TAC file name is us345dy1.dft, and the corrected arterial blood curve is us345ab.kbq, you would enter the following command:
bfmh2o -ml us345ab.kbq us345dy1.dft 240 us345rbf.res
Ahn JY, Seo K, Lee JS, Lee DS. Factor analysis for the quantification of renal cortical blood flow using O-15 water dynamic PET. Nuclear Science Symposium Conference Record, 2000 IEEE. doi: 10.1109/NSSMIC.2000.949260.
Alpert NM, Rabito CA, Correia DJA, Babich JW, Littman BH, Tompkins RG, Rubin NT, Rubin RH, Fischman AJ. Mapping of local renal blood flow with PET and H215O. J Nucl Med. 2002; 43: 470-475. PMID: 11937589.
Damkjær M, Vafaee M, Møller ML, Braad PE, Petersen H, Høilund-Carlsen PF, Bie P. Renal cortical and medullary blood flow responses to altered NO availability in humans. Am J Physiol Regul Integr Comp Physiol. 2010; 299: R1449-R1455. doi: 10.1152/ajpregu.00440.2010.
Effros RM, Lowenstein J, Baldwin DS, Chinard FP. Vascular and extravascular volumes of the kidney of man. Circulation Res. 1967; XX: 162-173. doi: 10.1161/01.RES.20.2.162.
Green MA, Hutchins GD. Positron emission tomography (PET) assessment of renal perfusion. Semin Nephrol. 2011; 31(3): 291-299. 10.1016/j.semnephrol.2011.05.008.
Inaba T, Yamashita M, Kawase Y, Nakahashi H, Watanabe H. Quantitative measurement of renal plasma flow by positron emission tomography with oxygen-15 water. Tohoku J Exp Med. 1989; 159: 283-289. doi: 10.1620/tjem.159.283.
Juillard L, Janier MF, Fouque D, Lionnet M, Le Bars D, Cinotti L, Barthez P, Gharib C, Laville M. Renal blood flow measurement by positron emission tomography using 15O-labeled water. Kidney Int. 2000; 57(6): 2511-2508. doi: 10.1046/j.1523-1755.2000.00110.x.
Juillard L, Janier MF, Foucue D, Cinotti L, Maakel N, Le Bars D, Barthez PY, Pozet N, Laville M. Dynamic renal blood flow measurement by positron emission tomography in patients with CRF. Am J Kidney Dis. 2002; 40(5): 947-954. doi: 10.1053/ajkd.2002.36325.
Kety SS. The theory and applications of the exchange of inert gas at the lungs and tissues. Pharmacol Rev. 1951; 3(1): 1-41. PMID: 14833874.
Koivuviita N, Liukko K, Kudomi N, Oikonen V, Tertti R, Manner I, Vahlberg T, Nuutila P, Metsärinne K. The effect of revascularization of renal artery stenosis on renal perfusion in patients with atherosclerotic renovascular disease. Nephrol Dial Transplant. 2012; 27: 3843-3848. doi: 10.1093/ndt/gfs301.
Kudomi N, Koivuviita N, Liukko KE, Oikonen VJ, Tolvanen T, Iida H, Tertti R, Metsärinne K, Iozzo P, Nuutila P. Parametric renal blood flow imaging using [15O]H2O and PET. Eur J Nucl Med Mol Imaging 2009; 36: 683-691. doi: 10.1007/s00259-008-0994-8.
Kötz B, West C, Saleem A, Jones T, Price P. Blood flow and Vd (water): both biomarkers required for interpreting the effects of vascular targeting agents on tumor and normal tissue. Mol Cancer Ther. 2009; 8(2): 303-309. 10.1158/1535-7163.MCT-08-1016.
Middlekauff HR, Nitzsche EU, Hamilton MA, Schelbert HR, Fonarow GC, Moriguchi JD, Hage A, Saleh S, Giggs GG. Evidence for preserved cardiopulmonary baroreflex control of renal cortical blood flow in humans with advanced heart failure. A positron emission tomography study. Circulation 1995; 92: 395-401. doi: 10.1161/01.CIR.92.3.395.
Middlekauff HR, Nguyen AH, Negrao CE, Nitzsche EU, Hoh CK, Natterson BA, Hamilton MA, Fonarow GC, Hage A, Moriguchi JD. Impact of acute mental stress on sympathetic nerve activity and regional blood flow in advanced heart failure. Implication for ‘triggering’ adverse cardiac events. Circulation 1997a; 96: 1835-1842. 10.1161/01.CIR.96.6.1835.
Middlekauff HR, Nitzsche EU, Nguyen AH, Hoh CK, Gibbs GG. Modulation of renal cortical blood flow during static exercise in humans. Circulation Res. 1997b; 80: 62-68. doi: 10.1161/01.RES.80.1.62.
Middlekauff HR, Nitzsche EU, Hoh CK, Hamilton MA, Fonarow GC, Hage A, Moriguchi JD. Exaggerated renal vasoconstriction during exercise in heart failure patients. Circulation 2000; 101: 784-789. doi: 10.1161/01.CIR.101.7.784.
Middlekauff MR, Nitzsche EU, Hoh CK, Hamilton MA, Fonarow GC, hage A, Moriguchi JD. Exaggerated muscle mechanoreflex control of reflex renal vasoconstriction in heart failure. J Appl Physiol. 2001; 90: 1714-1719. doi: 10.1152/jappl.2001.90.5.1714.
Nitzsche EU, Choi Y, Killion D, Hoh CK, Hawkins RA, Rosenthal JT, Buxton DB, Huang SC, Phelps ME, Schelbert HR. Quantification and parametric imaging of renal cortical blood flow in vivo based on Patlak graphical analysis. Kidney Int. 1993; 44: 985-996. doi: 10.1038/ki.1993.340.
Peters JE, Ter-Pogossian MM, Rockoff ML, Metzger JM, Koehler PR. Measurement of renal blood flow by means of radioactive water labeled with oxygen 15. In: Radionuclides in Nephrology (Blaufox MD, Funck-Brentano J-L, eds.), pp. 27-36. Grune & Stratton, 1972.
Päivärinta J, Koivuviita N, Oikonen V, Iida H, Liukko K, Manner I, Löyttyniemi E, Nuutila P, Metsärinne K. The renal blood flow reserve in healthy humans and patients with atherosclerotic renovascular disease measured by positron emission tomography using [15O]H2O. EJNMMI Res. 2018; 8:45. doi: 10.1186/s13550-018-0395-3.
Szabo Z, Mathews WB. Positron emission tomography imaging of the kidneys. In: Functional Imaging in Nephro-Urology. Taylor & Francis, 2006, pp 197-210. ISBN 9781841845111.
Turku PET Centre Modelling report: Analysis of renal perfusion from [15O]H2O PET studies. TPCMOD0032.
Updated at: 2019-11-30
Created at: 2008-03-18
Written by: Vesa Oikonen