| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Experimental Studies |
1 From the Parametric Imaging Laboratory, UMR 7623 CNRS and Paris University VI, 15 rue de l’École de Médecine, 75006 Paris, France (O.L., S.F.A., J.M.C., S.L.B., E.K., G.B.); Department of Radiology, Necker Hospital, AP-HP, Paris, France (O.L., S.F.A., J.M.C.); and Department of Radiology, Pitié-Salpêtrière Hospital, AP-HP, Paris, France (O.L.). Received June 13, 2002; revision requested August 8; revision received October 17; accepted December 10. Address correspondence to S.L.B. (e-mail: bridal@lip.bhdc.jussieu.fr).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
MATERIALS AND METHODS: During continuous injections of an ultrasonographic contrast agent, nonlinear gray-scale images were obtained in vitro in the longitudinal plane of a renal dialysis cartridge flow phantom (flow rates of 100, 200, and 400 mL/min) and in vivo in the coronal plane of the left kidneys of two rabbits (two kidneys). Destruction-replenishment curves were obtained for the dialysis cartridge in ROIs located immediately after the entrance of the microbubbles into the image plane and further from the entrance, after microbubbles had traveled across the complete length of the imaging plane. Replenishment curves were also obtained from ROIs in the rabbit kidneys at the level of segmental arteries, distal interlobar arteries, and the cortex.
RESULTS: The ROIs immediately after the entrance of the microbubbles in the image plane of the dialysis cartridge and in the segmental artery of the kidney followed a typical exponential function, A(1 - e-
t). Early portions of curves obtained in ROIs filled with microbubbles that had already passed through the image plane of the dialysis cartridge or in the renal cortex were not well described by such a function. The shape of the curve and the variations as a function of flow rate can be explained by means of a mathematical model based on indicator-dilution theory.
CONCLUSION: When the feeding vessels of an ROI travel across the ultrasound field before they reach the measurement region, the typical shape of the replenishment curve is modified (reduced velocity parameter and plateau).
© RSNA, 2003
Index terms: Blood, flow dynamics, 81.12988 • Kidney, US, 81.12984, 81.12988 • Microbubbles, 81.12988 • Phantoms, 81.12984, 81.12988 • Test objects, 81.12984, 81.12988 • Ultrasound (US), contrast media 81.12984, 81.12988
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
The purpose of this study was to investigate changes in destruction-replenishment curves (in vitro and in vivo) that result from microbubble destruction in feeding vessels that pass through the imaging plane before microbubbles enter the ROI.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
In Vitro Experiment
All in vitro measurements were performed with a nonrecirculating flow phantom consisting of a renal dialysis cartridge (Hémoflow model F60S; Fresenius Medical Care, Bad Homburg, Germany) with an internal diameter of 47 mm, which contained approximately 9,000 parallel polysulfone hollow fibers with a diameter of 200 µm, length of 25 cm, and total fiber volume of 70 mL (Fig 1). A dialysis cartridge was chosen because of the unidirectional alignment of the hollow fibers and because flow rate and the fractional circulating volume are constant throughout. The dialysis cartridge was connected to a 10-L reservoir positioned 2 m above the dialysis cartridge. The distal tubing was connected to a peristaltic pump (Masterflex, model 7523-27; Cole-Parmer Instruments, Vernon Hills, Ill). An electric syringe (Pulsar Ultrasound Injection System; Medrad, Indianola, Pa) was connected to the proximal side of the tubing system to inject the US contrast agent in the circuit. The direction of the flow at measurement sites within the circuit was oriented upward through vertical segments to minimize microbubble distribution gradients across the diameter of the filter and tubing that could be caused by upward movement of microbubbles that results from buoyant forces.
|
Contrast agent injection.—The US contrast agent used in this study (SonoVue; Bracco, Milan, Italy) consists of sulphur hexafluoride microbubbles obtained by means of reconstitution of a lyophilized powder with 5 mL of 0.9% saline solution. One milliliter of solution contains approximately 2 x 108 microbubbles stabilized by a layer of phospholipids (11). The mean microbubble diameter is 2.5 µm with a narrowed distribution (90% with a diameter of less than 8 µm) (11). Concentration of the microbubbles was kept constant at 5 mL of the US contrast agent per liter of fluid in the phantom; thus, the infusion rates of the US contrast agent were 0.5, 1.0, and 2.0 mL/min, respectively, for peristaltic pump flow rates of 100, 200, and 400 mL/min. No attempt was made to agitate the contrast agent in the syringe during the infusion. Duration of each infusion was less than 120 seconds.
Image acquisition.—Once the microbubble concentration in the circulation had stabilized, the dialysis cartridge was imaged (O.L.) with a clinical US unit (HDI 5000; ATL-Philips, Bothell, Wash) with a curvilinear transducer (C5-2 probe; ATL-Philips). The image plane was chosen parallel to the hollow fibers (sagittal plane). Nonlinear gray-scale mode was obtained with a pulse-inversion technique. To eliminate all microbubbles from the region, we acquired a set of 20 frames at maximum power (mechanical index of 1.1) for 0.8 second (24 frames per second). This destruction set was followed immediately by acquisition of 408 observation frames at low output power (mechanical index of 0.08) for 17 seconds (24 frames per second). Persistence of the US machine image display was set to zero. Focal zone was positioned in the first third (1.5 cm deep) of the dialysis cartridge. The destruction-replenishment experiments were repeated three times—30, 60, and 90 seconds after the beginning of the infusion—to allow averaging of the data.
In Vivo Experiment
Animal model.—After the institutional animal care committee at Necker Hospital approved the experimental procedure, two 1.5-month-old New Zealand male rabbits, which weighed 1.7 and 1.8 kg, were studied with general anesthesia (atropine; Aguettant, Lyon, France), 20 µg; ketamine (Ketalar; Parke Davis, Courbevoie, France), 35 mg per kilogram of body weight; and xylazine (Paxman; Vibrac Production Animale, Carros, France), 5 mg/kg. No paralytic agents were used, and oxygen (4 L/min) was delivered during the procedure. The animals were placed in the supine position, and the left kidney was imaged. The study was focused on the kidney because it presents a well-organized arterial tree with the feeding vessel from the hilum to the cortex included in a coronal section. The US contrast agent was administered in the right femoral vein at a continuous infusion rate of 0.5 mL/min (total dose of 3 mL). As in the flow phantom experiment, Doppler intensitometry was performed to detect the steady state of microbubble blood concentration. The 8-MHz pencil probe was stabilized with a mechanical device (S.F.A.) to measure the Doppler intensity in the left femoral vein.
The imaging sequence was performed as soon as the blood concentration of microbubbles reached equilibrium. Nonlinear gray-scale imaging with the pulse-inversion technique was performed with a low mechanical index (0.08) by two observers (S.F.A., J.M.C.) by using a linear transducer (model L7-4; ATL-Philips). The left kidney of each rabbit was imaged in the coronal plane (two kidneys). After four destructive frames were acquired at maximum power, 225 frames were acquired at a rate of 15 frames per second for 15 seconds. Persistence of the pulse sequence was set to zero. The focal zone was positioned in the middle of the kidney.
Imaging Analysis
Digital cine loops were transferred to a personal computer. Gray-scale enhancement was measured with the manufacturer software (HDI Lab, version 1.90a; ATL-Philips), which allows quantification of pixel intensity before logarithmic compression for video display.
For the in vitro experiment, two ROIs of 5 x 5 mm were positioned (O.L.) at the proximal and distal ends of the field of view. The proximal ROI included microbubbles that were just entering the ultrasound field. The distal ROI was filled with microbubbles that had traveled across the complete length of the ultrasound field. The proximal and distal ROIs were both located 0.5 cm deep within the phantom to minimize effects of shadowing. They were positioned symmetrically with respect to the central axis of the probe to ensure comparable beam characteristics (which influence backscatter amplitude and regularity) in the two zones.
For the in vivo experiment, four contiguous or slightly overlapping ROIs (1 x 1 mm) were positioned (O.L.) on the segmental arteries near the hilum (segmental artery ROI), on the distal part of the corresponding interlobar artery (interlobar artery ROI), and on the renal microcirculation of the cortex fed by this artery (cortex ROI) (Fig 2). For each location, the signal intensities in the four ROIs were averaged to reduce the signal variations and noise within each ROI. The segmental artery ROI was the first ROI to intersect with flowing contrast agent (the renal artery was not included in the imaging plane). In the interlobar and cortex ROIs, the signal intensity was measured in ROIs filled with microbubbles that had traversed the ultrasound field via the segmental artery or other portions of the microvasculature. In addition, we measured the enhancement of the cortex of each left kidney in a large cortex ROI, which surrounded the cortical area (Fig 2), to obtain maximum smoothing of the replenishment curve of the cortex.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
t) based on a single-tank model has generally been proposed to fit destruction-replenishment curves obtained in vivo in myocardium, brain, and kidney (5,6,8,12,13) during continuous perfusion of US contrast agents. According to this equation, microbubble velocity in the microcirculation is related directly to , and the plateau echo enhancement A correlates with the fractional blood volume when the microbubble destruction induced by the ultrasound field is null. The most critical condition for this type of evaluation is to know the function that relates image intensity to microbubble concentration. Since such a function remains difficult to characterize, often only a relative flow assessment can be achieved (same subject and same machine settings). This approach can still be very useful, for example, to study the effectiveness of angiogenic drugs or reperfusion therapy (5).
This exponential model, however, is valid only if the concentration of microbubbles that enter the ROI immediately after the destruction pulse is constant. This supposes that two hypotheses must be fulfilled: First, the injection rate of the contrast media remains constant; and second, the blood vessels that enter the ROI have not already been submitted to the ultrasound beam. Findings in our in vitro study show that when vessels that feed the ROI have previously been emptied of microbubbles, the replenishment curve is no longer an exponential that approaches a maximum but is a sigmoid; we describe this theoretically with an indicator-dilution model (Appendix). In this case, the initial slope of the replenishment curve is theoretically equal to zero. For example, the slope measured at the inflexion point, even with a constant flow rate and a constant microbubble concentration in the circulation, decreases when the length of the path followed by the microbubbles in the ultrasound field increases before they reach the ROI (Fig 3). Consequently, microbubble velocity in the ROI estimated from the initial slope () of the refill curve, in the classic model described by an exponential function that approaches a maximum (3), is not correct if microbubble destruction occurs in the feeding vessels.
In the same way, the plateau of enhancement, which classically gives information about the fractional blood volume, is related to the value C0/(1 + 
)n given by our model (Appendix). When the velocity of the microbubbles is high (ie, when is very small) compared with the destruction rate or when the destruction rate is close to zero, is negligible, and the measured signal intensity reflects the fractional blood volume. However, findings in our in vitro experiment (Fig 3) showed a decrease in the signal intensity measured at the plateau when the flow rates decreased (when increased). This effect becomes more visible at the distal extremity of the field of view (distal ROI). This can be explained by the effect of the number of subvolumes n, which reflects the length of the path followed by the microbubbles in the ultrasound field before they reach the ROI. The higher the value of n, the more the microbubble destruction coefficient becomes perceptible. In our study, even with the mechanical index set at 0.08, the effect of destruction induced by the continuous observation on the level of the plateau was clearly perceptible at a flow rate of 100 mL/min, particularly in the distal part of the ultrasound field (Fig 3).
Consequently, the plateau of enhancement depends on the fractional blood volume in the ROI, the length of the feeding vessels subjected to the ultrasound beam before they reach this region, and the flow rate. Thus, when possible, to relate the plateau level to the fractional blood volume, the signal intensity in tissue should be normalized by that in a blood vessel within the ROI (mean intensity in the lumen of a vessel filled by the same feeding vessel and at the same location and depth in the imaging plane). In addition to these factors, the total time needed to reach the plateau increases as the length of feeding vessels exposed to the initial ultrasound destruction pulse increases. This in turn increases the necessary observation duration. The parameter in our model describes the fraction of microbubbles destroyed by ultrasound per unit time. The more stable the microbubbles are, the smaller this parameter becomes. Thus, the use of microbubbles that are less sensitive to ultrasound should reduce the sigmoid distortion of the curve and increase the rapidity of arrival time at the plateau.
In our in vivo study, the cortical territories fed by the segmental arteries already included in the imaging and destruction plane exhibited replenishment curves that were comparable to those obtained in the distal ROI of the dialysis cartridge (Figs 4, 5). Since the two rabbits were healthy and were of the same age and weight, the fractional blood volume in the renal cortex of both animals should be comparable. The distances between each group of ROIs were similar in the two rabbits. The initial concentrations of the arriving contrast material were similar as assessed on the basis of the signal intensity at the plateau in the lumen of the segmental artery ROI. Consequently, the higher plateau in the interlobar arterial ROI and the cortex ROI curves for rabbit 1 may indicate a higher flow rate than that for rabbit 2. This would reduce the destruction that occurred in feeding vessels in rabbit 1 and, according to the model and results observed in vitro, should also lead to shortening of the low slope portion of the sigmoid curve in the interlobar artery ROI. Such shortening was observed (Fig 4), but no attempt was made in this study to correlate results with a physical measurement of renal blood flow or with pathologic findings in the kidneys.
Recently, Schlosser et al showed in an in vitro kidney perfusion model that different forms (comparable to our results) are observed for the replenishment curves from hilum vessels and the renal cortex (8). The hilum replenishment exhibited an exponential that asymptotically approached a maximum, but the cortical replenishment showed a sigmoid curve. We consider, as do these authors, that the important difference between the slope they calculated in the macrocirculation (hilum) and the microcirculation (cortex) is mainly a result of physiologic differences of local flow velocities. However, we also believe that the phenomenon of "entrance in the section" described in our study plays an important role and explains the delay and the shape of the curves observed by the authors and by ourselves for cortical replenishment. Furthermore, a voxel in the cortex may be supplied by several in-flow circuits that may bring microbubbles to the voxel at different velocities. Variations in microbubble arrival time due to effects such as these (unrelated to microbubble destruction) could also contribute to the sigmoid-shaped destruction-reperfusion curve. Similar results were obtained in an in vivo brain perfusion study in dogs (5,6). The replenishment curves the authors obtained in the brain cortex showed a sigmoid shape rather than an exponential shape that asymptotically approached a maximum. We believe that feeding vessels may have been included in the imaging plane.
Some limitations of the study should be mentioned. One concern is that if the microbubbles are layered in the electric syringe during injection, distribution of the contrast agent microbubbles may vary from one injection to another. To limit potential consequences of layering, in vitro injection times did not exceed 120 seconds. Furthermore, during each injection, three sets of data were obtained at fixed times (30, 60, and 90 seconds) after the beginning of injection and averaged. In vivo, the replenishment curves were acquired for both rabbits immediately after Doppler intensity equilibrium was reached, which occurred approximately 80 seconds after the beginning of the injection. Thus, because replenishment curves were measured at comparable times after the beginning of injection, the distribution of contrast agent in the syringe should have been comparable. Another consideration is that the measurements in selected ROIs in the dialysis cartridge were not compensated for beam characteristics or attenuation. For estimation of sets of compared curves, ROIs were positioned as similarly as possible to minimize potential effects of attenuation or field differences.
Additional in vivo studies, particularly in organs with well-organized terminal vascularization, such as the myocardium and the kidney, or on the contrary in a disorganized vascular lesion, such as a tumor, must be performed to better appreciate the consequences of the entrance in the section effect on blood flow and fractional blood volume quantification.
In conclusion, we highlighted and modeled an important and previously unexplained aspect of destruction-reperfusion curves. Results of our in vitro experiments show the effect on the initial slope and the plateau of the replenishment curve when the vessels that feed the ROI travel across the ultrasound field. This effect is predicted by our model, which was developed on the basis of indicator-dilution behavior. In vivo, this effect can explain the aspect of replenishment curves observed in the renal cortex. It also adds another complicating factor that must be taken into account for contrast material–enhanced functional US of an organ, especially one with a well-organized vasculature.
Practical application: Findings in this study suggest that it is best to minimize insonification of feeding arteries during destruction-reperfusion measurements in parenchyma and to limit the zone of microbubble destruction to only the studied zone. Such precautions should help determine reliable and reproducible estimates of both the flow rate and the fractional blood volume of a selected organ.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
|
represents the fraction of microbubbles destroyed by the ultrasound beam per unit time (per second), which is assumed to be constant.
C0 is the concentration of microbubbles in blood vessels that enter the ROI (number of microbubbles per liter), which is assumed to be constant with time. This supposes that the blood vessels that enter the ROI were not themselves submitted to the destruction pulse. Consider now several subvolumes in the insonified zone that are oriented serially with respect to the direction of flow. When the microbubbles that come into the subvolume n have already traveled through the ultrasound field in the subvolume n - 1, the number of microbubbles that enter the subvolume n is itself a function of the output function of the subvolume n - 1 (Fig A1a). Evolution of the microbubble concentration in the subvolume n, Cn(t), is given by the following equation:
|
= F/Vb. Cn-1 is the microbubble concentration in subvolume n - 1.
|
|
)/.
Finally, for the first subvolume, Equation (A3) gives the following equation:
|
= 0, case of intermittent imaging) or considered negligible (case of continuous imaging with low transmit power), Equation (A4) corresponds to the classic exponential function, which asymptotically approaches a maximum, that results from a single-tank model; that model was initially proposed by Wei et al (3).
For subvolumes other than the first (when n > 1),
|
|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviation: ROI = region of interest
Author contributions: Guarantors of integrity of entire study, O.L., S.F.A.; study concepts, O.L., S.F.A., J.M.C., G.B.; study design, O.L., S.F.A., J.M.C.; literature research, O.L., S.F.A.; experimental studies, O.L., S.F.A., J.M.C.; data acquisition, O.L., S.F.A.; data analysis/interpretation, O.L., S.F.A., E.K., S.L.B.; manuscript preparation, O.L., S.L.B.; manuscript definition of intellectual content, O.L., S.L.B., E.K.; manuscript editing, O.L.; manuscript revision/review, O.L., J.M.C., E.K., G.B.; manuscript final version approval, S.F.A., J.M.C., S.L.B., E.K., G.B.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. J. Paltiel, L. A. Kalish, R. A. Susaeta, F. Frauscher, P. L. O'Kane, and L. G. Freitas-Filho Pulse-Inversion US Imaging of Testicular Ischemia: Quantitative and Qualitative Analyses in a Rabbit Model Radiology, June 1, 2006; 239(3): 718 - 729. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Krix, M.-A. Weber, H. Krakowski-Roosen, H. B. Huttner, S. Delorme, H.-U. Kauczor, and W. Hildebrandt Assessment of Skeletal Muscle Perfusion Using Contrast-Enhanced Ultrasonography J. Ultrasound Med., April 1, 2005; 24(4): 431 - 441. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Kruskal Science to Practice: Blood Flow Quantification with Contrast-enhanced US: "Entrance in the Section" Phenomenon--Phantom and Rabbit Study Radiology, August 1, 2003; 228(2): 298 - 299. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) and rabbit 2 (
). Averaging of the data from all pixels in these ROIs leads to smoothing of the curves, which emphasizes the sigmoid aspect of the initial portion of the replenishment curves obtained in the parenchyma when the imaging plane intersects the feeding arteries.
