Pulmonary blood flow using [15O]H2O PET

The lungs have two circulations, pulmonary and bronchial circulation. Pulmonary circulation receives all of the cardiac output; all venous blood is pumped from the right side of the heart via pulmonary arteries to the lungs for the gas exchange before returning via pulmonary veins into the left side of the heart. Bronchial circulation provides oxygenated blood to the walls of pulmonary arteries and veins, bronchi and bronchioles, nerves, lymph nodes, and visceral pleura. In healthy lungs, bronchial blood flow is low, only 1-5% of pulmonary circulation, but lung tumours are usually perfused via bronchial circulation.

Perfusion in normal lungs is extremely high: calculated from the cardiac output and the weight of lungs, blood flow via pulmonary circulation is ∼5-10 mL/(g min) at rest, and >30 mL/(g min) during exercise. The lung volume consists mostly of air (depending on the breathing cycle). Lung density can be assessed from PET transmission scan, and when combined with PET scan after inhalation of 11C or 15O labelled carbon monoxide, extravascular lung density can be measured (Rhodes et al., 1981). Peripheral lung density, excluding hilar region, is ∼0.28 g/mL, comprising 63% of blood and 37% extravascular tissue (Brudin et al., 1987). Lung water content can be estimated from [15O]H2O PET 5-5.5 min p.i., when equilibrium is reached (Schuster et al., 1985 and 2002; Velasquez et al., 1991; Naum et al., 2007), and could be used in assessment of pulmonary oedema (Fazio et al., 1976).

[15O]H2O model

The analysis method of pulmonary blood flow (PBF) is based on one-tissue compartment model. PBF can be estimated from regional time-activity concentration curves (TACs), or from dynamic PET image to produce perfusion map (Mintun et al., 1986; Schuster et al., 1990 and 1995; Serizawa et al., 1994; Richard et al., 2002; van der Veldt et al., 2010; Heinonen et al., 2013; Matsunaga et al., 2017). The method is validated against microsphere and macroaggregate measurements in animals and humans.

Radiowater concentration in the volume of interest, as measured with PET, (CPET(t)), is the volume fraction weighted sum of concentrations in the blood (CB(t)) and lung parenchyma (tissue, (CT(t)), considering also the volume fraction of the air (VA) but assuming that radiowater concentration in the air is negligible:

, in which 1-VA can be assessed from PET transmission scan or CT (Rhodes et al., 1981; Schuster et al., 1986a and 1986b; Matsunaga et al., 2017).

While the usual radiowater model, where f is the perfusion and p is the partition coefficient of water in tissue,

is applicable, the pulmonary blood flow (PBF) can be reported either per parenchymal volume, PBF=f, or per PET volume including the air and blood, PBF=(1-VA-VB)×f. Accordingly, the distribution volume is given either as VT=p or VT=(1-VA-VB)×p. The average PBF as measured using [15O]H2O PET is ∼1.4 mL/(min*mL lung) and ∼5.0 mL/(min*mL lung parenchyma) (Matsunaga et al., 2017). The VT for radiowater was 0.17±0.03 for the total lung and 0.60±0.08 for the lung parenchyma (Matsunaga et al., 2017).

In case of radiowater, the blood volume fraction represents only the precapillary fraction of the blood, which based on histological analysis could be estimated to be about 3% of lung volume (Singhal et al., 1973; Matsunaga et al., 2017). AIC analysis did not support including VB in the radiowater analysis of the lungs (Matsunaga et al., 2017); however, this analysis do not consider the time delay between the lungs and the input function.

Input function

Input function for the lungs must be taken from the pulmonary artery or RV cavity of the heart, since pulmonary circulation provides most of the blood input to the lungs. In contrast to the healthy lung tissue, arterial blood, either from LV cavity, ascending aorta, or arterial on-line sampling system should be used as input function for lung tumours (Hoekstra et al., 2002; van der Veldt et al., 2010), where bronchial circulation provides most of the blood input. Although the appearance time of [15O]H2O in the RV cavity and lungs does not usually differ more than 1-2 s, blood curve should still be corrected for the delay because of the very high blood flow. Due to the large size and heterogeneity of lung tissues, delay correction may need to be lung region specific, and implemented as an additional parameter in the model fitting for the lungs (Richard et al., 2002; Wellman et al., 2015).

Calculation of blood flow image

A parametric map of perfusion should be calculated if one is interested in flow values in individual image pixels, not only the regional averages. Transmission scan based correction for the air volume reduces apparent heterogeneity of PBF and transforms the statistical distribution of PBF values to Gaussian (Matsunaga et al., 2017).

If you have the PET images in DICOM format, convert them to ECAT 7 format.

Pulmonary blood flow (PBF) image can be calculated using imgflow or imgbfh2o; the latter uses basis function method (Matsunaga et al., 2017). These programs were developed for general use and do not take into account the air volume or correct blood curve for the delay time.

Calculation of PBF from regional tissue TACs

Transmission or CT images can be used to define the lung ROIs, possibly using a certain threshold value (Matsunaga et al., 2017).

Regional PBF can be estimated from lung tissue TACs using fit_wpul, which by default fits and corrects the delay time and dispersion for each region individually (see an example fit in Figure 1). Alternatively, general programs for radiowater model fitting can be used, either fit_h2o, which fits and corrects the delay time for each region individually, or bfmh2o, which requires pre-corrected blood curve. In these general programs the default constraints for model parameters may not be suitable for PBF estimation and may need to be changed.

Radiowater model fitted to pulmonary TTACs
Figure 1. Radiowater model was fitted using fit_wpul to lung TTACs using as input function the blood TAC from ROI drawn on heart LV cavity (black). In this example dispersion and VB were fixed to zero. Estimated delay time was between 0.5-2 s, K1 between 1.5 and 2.5 mL/(min*mL), and k2 between 10 and 13 min-1. The apparent VT was in the range 0.14-0.19 mL/mL.

See also:


Chen DL, Cheriyan J, Chilvers ER, Choudhury G, Coello C, Connell M, Fisk M, Groves AM, Gunn RN, Holman BF, Hutton BF, Lee S, MacNee W, Mohan D, Parr D, Subramanian D, Tal-Singer R, Thielemans K, van Beek EJ, Vass L, Wellen JW, Wilkinson I, Wilson FJ. Quantification of lung PET images: challenges and opportunities. J Nucl Med. 2017; 58(2): 201-207. doi: 10.2967/jnumed.116.184796.

Dupuis J, Harel F, Nguyen QT. Molecular imaging of the pulmonary circulation in health and disease. Clin Transl Imaging 2014; 2(5): 415-426. doi: 10.1007/s40336-014-0076-9.

Parthasarathi K (ed.): Molecular and Functional Insights Into the Pulmonary Vasculature. Springer, 2018. ISBN 978-3-319-68483-3. doi: 10.1007/978-3-319-68483-3.

Rhodes CG, Hughes JM. Pulmonary studies using positron emission tomography. Eur Respir J. 1995; 8(6): 1001-1007. PMID: 7589363.

Suresh K, Shimoda LA. Lung circulation. Compr Physiol. 2016; 6: 897-943. doi: 10.1002/cphy.c140049.

Thiriet M. Anatomy and Physiology of the Circulatory and Ventilatory Systems. Springer, 2014. doi: 10.1007/978-1-4614-9469-0.

Tags: , ,

Updated at: 2019-10-03
Created at: 2019-09-25
Written by: Vesa Oikonen