Validation of fractional moving blood volume ... - Wiley Online Library

Ultrasound Obstet Gynecol 2004; 23: 363–368 Published online 5 March 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/uog.1002

Validation of fractional moving blood volume measurement with power Doppler ultrasound in an experimental sheep model E. HERNANDEZ-ANDRADE*, T. JANSSON†, D. LEY‡, M. BELLANDER‡, M. PERSSON†, ´ G. LINGMAN* and K. MARSˇ AL* *Departments of Obstetrics and Gynecology, †Electrical Measurements and ‡Pediatrics, University of Lund, Lund, Sweden

K E Y W O R D S: animal; fetus; fractional moving blood volume; mean pixel intensity; microspheres; power Doppler ultrasound

ABSTRACT Objective To compare fractional moving blood volume (FMBV) estimation using power Doppler ultrasound (PDU) with blood flow estimation using radioactive microspheres (RMS) for evaluation of fetal organ blood perfusion. Methods Blood flow was measured in the adrenal gland of nine exteriorized fetal lambs. Five fetal lambs underwent total umbilical cord occlusion in order to induce changes in the adrenal blood flow (asphyxia group). Four lambs were used as sham controls (control group). Three RMS injections, with coincident PDU recordings of the adrenal gland, were performed in each lamb. In the asphyxia group, measurements were taken before the cord occlusion, 5 min later and when the mean blood pressure decreased below 25 mmHg. In the control group, the measurements were done with an interval of 5 min. FMBV normalized for attenuation of PDU signals, and mean pixel intensity (MPI) were estimated offline. After completion of the study, adrenal blood perfusion was calculated according to the reference sample microsphere technique, using the isotope activity and expressed in mL/min/100 g. The correlation between RMS and FMBV and MPI, respectively, was analyzed individually for each lamb. Results In the asphyxia group, all lambs showed a marked reduction in the adrenal blood perfusion towards the third RMS injection. In the control group, the adrenal perfusion showed small variations throughout the experiment. In the total material, there was a higher correlation between FMBV and RMS (median, r = 0.90; range, 0.43–0.99) than between MPI and RMS (median, r = 0.55; range, −0.53 to 0.99).

Conclusion The FMBV method of quantifying PDU signals correlates highly with blood flow perfusion estimation using RMS in the fetal lamb adrenal gland. Copyright  2004 ISUOG. Published by John Wiley & Sons, Ltd.

INTRODUCTION Over the past 20 years, Doppler ultrasound has been used for estimation of blood flow in large vessels1 . With the advent of color Doppler imaging and analysis of the energy content of the Doppler signals (power Doppler), assessing blood movement within small vessels in a specific region of interest (ROI) has become possible2 . Blood cells acting as moving reflecting objects change the backscattered energy of the Doppler signal, and this energy shift is identified by the system and presented as various degrees of pixel intensity in a monochromatic code on the screen. In this way, power Doppler ultrasound (PDU) has been mainly used in obstetrics for qualitative assessment of increased or decreased fetal tissue perfusion or vascularity3 . There have been several attempts to quantify PDU in the fetal–placental circulation4,5 . Using the decibel (dB) power per pixel as units, and a qualitative comparison of color pixels, calculation of the average pixel density has been proposed. This approach gives an estimation of the relative amount of moving blood volume and is represented as mean pixel intensity (MPI) in a specific ROI. However, PDU signals are highly influenced by a number of factors, e.g. depth, attenuation, transducer aperture and machine settings, and the quantification of pixel intensity alone does not overcome these problems6 .

Correspondence to: Dr E. Hernandez-Andrade, Department of Obstetrics and Gynecology, Lund University Hospital, SE 221 85 Lund, Sweden (e-mail: [email protected]) Accepted: 12 December 2003

Copyright  2004 ISUOG. Published by John Wiley & Sons, Ltd.

ORIGINAL PAPER

Hernandez-Andrade et al.

364

To compensate for this dependence of PDU signals, Rubin et al.7,8 proposed a normalization procedure resulting in a measure termed fractional moving blood volume (FMBV), which gives a more reliable estimation of the amount of blood moving in a specific ROI. The basic idea is that the ratio of backscattered signal power to the power of the incident sound will always be the same if the scattering medium is the same. Thus, PDU information from blood vessels within a given ROI will appear similar, independent of depth or overlying tissue. The aim of the present study was to compare in a fetal sheep model two methods for estimation of fetal organ blood perfusion, FMBV using PDU, and blood flow measurement with radioactive microspheres (RMS) generally considered as the standard method for this purpose9 .

METHODS Blood perfusion was estimated in the adrenal gland of nine near-term fetal lambs of mixed breed. The mean gestational age calculated from the day of conception was 136 (range, 134–138) days. The study was approved by the Animal Ethics Research Committee at Lund University. A Cesarean section was performed in the pregnant ewe under isofluorane anesthesia. The fetus was exteriorized with minimal disturbance and placed in a thermostated waterbath at 39◦ C. The fetal–placental circulation was maintained. To avoid breathing, a rubber glove was placed over the fetal head. Two catheters were then inserted: one in the femoral artery with the tip located in the abdominal aorta just above the iliac bifurcation and one in the femoral vein with the tip located in the inferior vena cava. After preparation, five fetal lambs underwent total umbilical cord occlusion in order to achieve asphyxia with redistribution of blood flow and, consequently, changes in the adrenal gland blood flow (asphyxia group). Four lambs were used as sham controls (control group). As a standard method of measuring

a

adrenal gland blood perfusion a RMS technique was used9 . Hemoglobin, blood oxygen saturation (OSM 2, Radiometer, Copenhagen, Denmark) and acid–base (ABL 300, Radiometer) were measured before each RMS injection. Three RMS injections (New England Nuclear, Zaventem, Belgium) were performed through the femoral vein catheter in each lamb using 15-µm diameter RMS labeled with 141 Cesium, 46 Scandium and 51 Chromium, respectively. In the asphyxia group, RMS were injected before cord occlusion, 5 min after cord occlusion and when the mean arterial blood pressure (MABP) decreased below 25 mmHg (after mean time 9.8 min). In the control group, RMS were injected with an interval time of 5 min. A random order of the isotopes was used to minimize measurement bias. A reference sample was taken from the femoral artery catheter at a rate of 3.3 mL/min for 1.5 min using a precalibrated pump (Harvard Apparatus, Dover, MA, USA). Blood samples were collected for later analysis. The volume of fetal blood removed was replaced with fresh maternal blood. Ultrasound recordings were performed at the time of the RMS injections using ATL HDI-5000 (Philips Medical Systems, ATL Ultrasound, Bothell, WA, USA) ultrasound equipment with a linear 12–5-MHz probe. Settings were kept identical for all examinations: medium persistence, high sensitivity, normal line density, normal image display, maximum dynamic range, high frame rate, 60–70% of gain range, medium wall filter and pulse repetition frequency of 500 Hz. The thermal and mechanical indices were always kept below 1.0. The fetal adrenal gland was located at the upper pole of the kidney with a clear image of the central adrenal vein (Figure 1a). The left adrenal gland was selected for the measurements. The color power Doppler box was adjusted in order to include the complete gland with minimum of adjacent fetal structures and the HDI-Zoom option was used to magnify the image. A sequence of images starting at the time of the RMS injection and lasting for 60 s was recorded. The digital ultrasound data were transferred and stored in a

b

Figure 1 (a) Fetal lamb adrenal gland visualized with power Doppler ultrasound technique with a clear view of the central vein. (b) The region of interest (ROI) including only the adrenal gland.


Ultrasound Obstet Gynecol 2004; 23: 363–368.

Power Doppler ultrasound and fetal organ blood perfusion personal computer for offline analysis in HDI-Lab 1.81 (ATL, Philips Medical Systems, ATL Ultrasound). HDILab is a custom software application for visualization and quantification of ultrasound cineloop data, which allows analysis of the ultrasound data, prior to (or after) scan conversion rather than analysis of screen images from the ultrasound scanner. In the present method, scan-converted data presented by the program were extracted and analyzed. As a specific look-up table for conversion of color data to dB-scaled power values was used, any nonlinearity that normally is associated with this step could be avoided. The same procedure was repeated for each RMS injection. For FMBV and MPI calculations, 10 consecutive ultrasound frames towards the end of each sequence with minimal flash artifacts were selected and a customdesigned software plug-in for HDI-Lab was used for the offline analysis. The data file input and output as well as the user interface of the software module were developed in Microsoft Visual C++ 6.0 (Microsoft Corp., Redmond, WA, USA), whilst the signal processing was performed in MATLAB 5.0 (The MathWorks, Natick, MA, USA). The software module functions were used in the following way: a ROI was drawn in the grayscale image including only the adrenal gland (Figure 1b). The FMBV value was then calculated from the dB-scaled power Doppler data within the ROI10 . We employed the method described by Rubin et al.8 that is based on the analysis of the cumulative power distribution function. All power values within the ROI were used for the analysis, and the algorithm automatically found the normalization value based on a two-step procedure by finding a change in the distribution of the power values8,10 . Provided that the ROI included vessels that carried blood traveling with velocity well above the cut-off value of the wall filter, a normalization value corresponding to 100% blood was found. The ROI was the same for all frames analyzed. The individual mean and SD of FMBV, and MPI, of the selected frames were calculated. After completion of the experimental study the fetuses were sacrificed and the left adrenal glands carefully removed, weighed and fixed in formaldehyde. The nuclear activity was calculated simultaneously for all used isotopes in the adrenal gland and in the reference samples with a Ge-detector (Canberra Industries Inc., Meriden, CT, USA) using Genie 2000 analysis software (Canberra Packard, Zellik, Belgium). Activity in all calculated

365

samples represented an excess of 400 RMS. The adrenal blood flow was calculated by comparing the number of RMS in the reference samples with the number of RMS retained in the adrenal gland. As the reference sample was taken at a specific speed by the external pump, adrenal flow was then calculated and expressed in mL/min/100 g. Sequential changes in the adrenal blood flow over time of RMS, FMBV and MPI were registered in each lamb and the results statistically evaluated using linear regression and correlation analysis. To avoid the influence that different number of individual measurements might have on the total analysis, Pearson’s correlation coefficient was calculated for each lamb, and the median and range values of the individual correlation coefficients given.

RESULTS From the nine lambs, 23 successful simultaneous measurements with RMS and PDU were obtained. In one lamb from the asphyxia group, and one of the control group, the second FMBV/MPI value was not possible to estimate. In another control lamb, the second and third FMBV/MPI values were not possible to estimate. This was mainly due to the flash artifacts in the ultrasound images. The median intra-individual coefficient of variation for FMBV was 0.09 (0.03–0.18) and for MPI 0.11 (0.03–0.24). Table 1 gives the MABP, arterial blood gas and acid–base values in both groups at the time of each RMS injection. Figure 2 shows the individual measurements of adrenal blood perfusion in both groups evaluated by RMS, FMBV and MPI, respectively. In the asphyxia group, one lamb had a marked reduction in the adrenal perfusion from the first to the second RMS injections, whereas the other four lambs showed small changes. In all lambs, there was a marked reduction in the adrenal blood perfusion towards the third RMS injection. In the control group, the adrenal blood perfusion changed only slightly throughout all RMS injections. The time course of FMBV and MPI measurements was similar to that of RMS. The linear regression equation for FMBV and RMS was FMBV = 0.0194 RMS + 6.3375 (P < 0.001) (Figure 3) and between MPI and RMS, MPI = 0.0029 RMS + 0.4257 (P < 0.001) (Figure 4). The median individual correlation coefficient for FMBV and RMS was 0.90 (range, 0.43–0.99) and for MPI and RMS 0.55 (range, −0.53–0.99).

Table 1 Physiological variables in the asphyxia and control groups at the time of the microsphere injections* Asphyxia group Parameter pH pCO2 (kPa) BE (mmol/L) ArtO2 (mL/100 mL) MABP (mmHg)

Baseline

5 min

7.04 (0.11) 10.0 (1.6) −12.3 (3.9) 2.1 (0.8) 45.0 (11.3)

6.83 (0.07) 15.2 (1.8) −18.7 (2.5) 1.2 (0.8) 24.4 (7.2)

Control group MABP < 25 mmHg 6.78 (0.06) 16.4 (1.8) −19.7 (2.5) 0.9 (0.3) 13.0 (5.3)

Baseline

5 min

10 min

7.03 (0.1) 9.8 (1.0) −12.9 (5.8) 3.8 (2.2) 53.7 (11.6)

7.02 (0.09) 9.8 (0.8) −13.1 (5.5) 3.9 (2.0) 53.3 (10.3)

7.02 (0.11) 9.9 (1.6) −13.2 (5.5) 3.9 (2.7) 50.0 (9.1)

*Values are expressed as mean (SD). ArtO2 , arterial oxygen content; BE, base excess; MABP, mean arterial blood pressure.




366 Asphyxia group

Control group 2500 Microspheres (mL/min/100 g)

2000 1500 1000 500 1

2

1000 500 0

3

60

60

50

50

40 30 20 10 0

1

8 7 6 5 4 3 2 1 0

Mean pixel intensity (dB/pixel)

1500

Fractional moving blood volume (%)


0

2000

2

1

2

3

30 20 10 0

3

2

1

40

1

8 7 6 5 4 3 2 1 0


Microspheres (mL/min/100 g)

2500

3

2

1

Microsphere injections

3

2

3

Microsphere injections

60

7

50

6



Figure 2 Individual values of the fetal adrenal perfusion estimated with radioactive microspheres, fractional moving blood volume and mean pixel intensity in the asphyxia and control groups at the time of microsphere injections. Dotted lines represent cases with only two measurements.

40 30 20 10 0

0

500 1000 1500 2000 Radioactive microspheres (mL/min/100 g)

2500

5 4 3 2 1 0

0

500

1000

1500

2000

2500

Radioactive microspheres (mL/min/100 g)

Figure 3 Fractional moving blood volume in the fetal adrenal gland plotted against flow measurements with radioactive microspheres.

Figure 4 Mean pixel intensity in the fetal adrenal gland plotted against flow measurements with radioactive microspheres.

DISCUSSION

Blood perfusion can be defined as the volume of blood flowing through a volume of tissue per unit of time. Strictly only the flow through capillaries should be included when perfusion is evaluated11 . Previously, various techniques such as plethysmography, electromagnetic probes and tracer techniques with contrast agents or labeled

The results of this study showed that FMBV and MPI, measured with PDU in an animal model, correlate significantly with fetal organ blood perfusion estimated with RMS analysis.



Power Doppler ultrasound and fetal organ blood perfusion microspheres have been used to evaluate blood perfusion9 . MacLean et al.12 introduced RMS in 1960 to evaluate the cardiac blood distribution in dogs, and later Rudolph and Heymann13 used it to study blood perfusion of organs in the fetal lamb. There has been criticism of this technique and its ability to estimate true blood perfusion14 . The size and distribution of the microspheres, the creation of an artificial blood flow and the radioactivity emanated from spheres of different size are some of the factors that can alter the final result. Nevertheless, the RMS technique is still the standard for measuring perfusion in animal experiments. PDU has the advantage of being a non-invasive technique that can be safely used in clinical practice to estimate blood flow movement. Various attempts have been made to validate PDU as a method of perfusion estimation. Several groups reported a high correlation between the contrast agent enhanced PDU and RMS when measuring the brain flow in piglets15 and dogs16 . With contrast agents, PDU can be significantly enhanced. However, such enhanced signals do not necessarily represent the normal backscattered power from moving blood cells and it is difficult to assume that they reflect the true physiological values of vascular perfusion. Taylor17 recorded PDU signals without contrast agents and also found a high correlation between MPI and RMS when the cerebral blood flow of newborn lambs was measured (r = 0.84). Our results for MPI are in accordance with these previous studies. Rubin et al.7 , using a laboratory phantom, found a very high correlation between PDU and particle concentration (r = 0.97) when the normalization procedure was used. Our results showed a slightly less significant correlation than the one reported by Rubin et al.7 probably due to the differences in the experimental model used in the studies. There are some factors that might influence the relation between the temporal changes in perfusion and PDU signals. The capacity of the ultrasound equipment to record blood movements from very small vessels is limited by the signal-to-noise ratio. If the signal-tonoise ratio is low, the equipment will not be able to differentiate between noise and slow blood movements, thus underestimating the real perfusion changes. Another factor is the angle dependence. It is generally accepted that the isonation angle has a very limited effect on PDU. However, Gudmundsson et al.6 demonstrated that PDU intensity changes when the angle of isonation is above 75◦ and the signal disappears when the angle is 90◦ . This is in accordance with Bude et al.18 who showed that the shape of the accumulative distribution curve of PDU is affected by the isonation angle whereas the area under the curve is not. The same group described the volume flow to have a strong influence on PDU. We can assume that within the ROI, blood is moving in all directions, and thus it is possible that blood cells moving perpendicular to the ultrasound beam will not be recorded. Nevertheless, our results showed a very good correlation between PDU and RMS.


367

The fetal adrenal gland was selected for perfusion measurements in this study as it is one of the main organs involved in fetal adaptation to hypoxia. The adrenal gland is easy to locate and the total organ can be examined. Still, it was not possible to obtain good quality sequences in all cases. In four measurements, the flash artifacts contained in the ultrasound sequences made the estimation of flow impossible. The reasons were gasping movements of the fetus or artifacts from the abdominal aorta located very close to the gland. Despite these problems, more than 80% of our recordings were successful. The statistical analysis showed a better correlation with RMS for the FMBV than for MPI. This was probably due to the fact that when using FMBV the depth dependence of the PDU signals was compensated for, and the correlation with blood perfusion improved. The exteriorized fetal lamb model was used in order to maintain the fetal–placental circulation and to obtain high quality PDU recordings. However, the fetal lamb is extremely sensitive to manipulation. Some of the lambs started the study with a low pH, which might have produced changes in organ blood perfusion. The wide variance of our RMS and FMBV results might reflect individual responses to different pH and pCO2 values. In the asphyxia group, four lambs showed a maintained flow during the first and second RMS injections. It seems that for 5 min of total cord occlusion, the fetal lambs kept the flow to adrenal glands as an adaptation mechanism to survive. After this time the mean blood pressure dropped and the blood perfusion declined. Only one lamb from the asphyxia group showed a marked continuous reduction in adrenal perfusion from the first to the second RMS injections. In the control group the variations in the adrenal flow were small throughout the all measurements. However, for the purpose of the study, the range of adrenal blood flow values in the total material was sufficient to allow a comparison of FMBV and RMS methods. Both FMBV and MPI showed similar perfusion patterns to the RMS in all lambs. However, at the end of the experiment, there was a bigger difference in perfusion measured by RMS and FMBV. A possible explanation might be that FMBV and MPI identify blood movement in all vessels including veins. In contrast, the RMS technique does not estimate the flow in the venous system. This should be seen as an advantage of FMBV, which can give a global estimation of blood movement through small diameter arteries and veins and more complete information on the fetal adaptive hemodynamic responses. According to the results presented here, we can conclude that PDU, and in particular FMBV, might be used as a reliable non-invasive method for estimation of relative blood perfusion in fetal organs.

ACKNOWLEDGMENTS The authors acknowledge the professional assistance in the experimental laboratory provided by Ingela



368

¨ Matinsson-Sandstrom, Ulla Ganestam and Karina Liuba. Dr Hernandez-Andrade was supported by the Mexican National Council for Science and Technology (CONACyT) and the National Institute of Perinatal Medicine (INPer) in Mexico City. The study was supported by the Swedish Research Council (Grant No. 5980). The Swedish Foundation for Strategic Research (SSF, Project CORTECH) is also thanked for financial support.

9.

10.

11.

REFERENCES ´ K, Brubakk AO, Kristofferson K, Ulstein M. 1. Eik-Nes SH, Marˇsal Ultrasonic measurement of human fetal blood flow. J Biomed Eng 1982; 4: 28–36. 2. Rubin J, Bude R, Carson P, Bree R, Adler R. Power Doppler US: a potentially useful alternative to mean frequency-based color Doppler US. Radiology 1995; 190: 853–856. 3. Fortunato S. The use of power Doppler and color power angiography in fetal imaging. Am J Obstet Gynecol 1996; 174: 1828–1833. ´ K, 4. Dubiel M, Kozber H, Debniak B, Breborowicz G, Marˇsal Gudmundsson S. Fetal and placental power Doppler imaging in normal and high-risk pregnancies. Eur J Ultrasound 1999; 9: 223–230. 5. Welsh A, Bennet P, Fisk N. Quantitative digital analysis of regional perfusion using power Doppler in placental abruption. Ultrasound Obstet Gynecol 2001; 17: 203–208. 6. Gudmundsson S, Valentin L, Pirhonen J, Olofsson PA, Dubiel ´ K. Factors affecting color Doppler energy ultrasound M, Marˇsal recordings in an in-vitro model. Ultrasound Med Biol 1998; 24: 899–902. 7. Rubin J, Adler R, Fowkles J, Spratt R, Pallister J, Chen JF, Carson P. Fractional moving blood volume: estimation with power Doppler US. Radiology 1995; 197: 183–190. 8. Rubin J, Bude R, Fowkles J, Spratt R, Carson P, Adler R.


12.

13.

14.

15.

16.

17.

18.

Normalizing fractional moving blood volume estimates with power Doppler US: defining a stable intravascular point with the cumulative power distribution function. Radiology 1997; 205: 757–765. Heymann M, Payne B, Hoffman J, Rudolph A. Blood flow measurements with radionuclide-labeled microspheres. Prog Cardiovasc Dis 1977; 10: 55–79. ´ K. EstiJansson T, Hernandez-Andrade E, Lingman G, Marˇsal mation of fractional moving blood volume in fetal lung using power Doppler ultrasound. Methodological aspects. Ultrasound Med Biol 2003; 29: 1551–1559. Jansson T, Persson H, Lindstrom K. Estimation of blood perfusion using ultrasound (Review). Proc Inst Mech Eng [H] 1999; 213: 91–106. MacLean D, Kim YS, Hendenstrom P. Distribution of blood flow in canine and human hearts. Surg Forum 1961; 12: 221–223. Rudolph A, Heymann M. Measurements of flow in perfused organs using microsphere techniques. Acta Endocrinol 1972; 69(Suppl. 158): 112–127. Peeters L. Fetal blood flow at various levels of oxygen. A study in a chronic sheep preparation with radioactive microspheres. PhD thesis, University of Colorado School of Medicine, Denver, CO, USA, 1978. Taylor G, Barnewolt C, Dunnings P. Excitotoxin-induced cerebral hyperemia in newborn piglets: regional cerebral blood flow mapping with contrast enhanced power Doppler US. Radiology 1998; 208: 73–79. Greenberg R, Taylor G, Stapleton J, Hillsley C, Spinak D. Analysis of regional cerebral blood flow in dogs, with an experimental microbubble-based US contrast agent. Radiology 1996; 201: 119–123. Taylor G. Regional cerebral blood flow estimates in newborn lamb using amplitude-mode color Doppler ultrasound. Pediatr Radiol 1996; 26: 282–286. Bude R, Rubin J, Adler R. Power versus conventional color Doppler sonography: comparison in the depiction of normal intrarenal vasculature. Radiology 1994; 192: 777–780.
