WO2005020820A1

WO2005020820A1 - Method and system for determining the absolute perfusion rate of a fluid in an analysis region

Info

Publication number: WO2005020820A1
Application number: PCT/EP2004/009523
Authority: WO
Inventors: Rolf Vogel; Christian Seiler
Original assignee: University Of Bern
Priority date: 2003-08-28
Filing date: 2004-08-26
Publication date: 2005-03-10

Abstract

An ultrasonic apparatus for performing perfusion measurements of a fluid carrying an ultrasound contrast agent into an analysis region comprises an ultrasound source means capable to provide a multitude of different ultrasound signals, an ultrasound detector means capable to detect backscatter of said multitude of different ultrasound signals, a processor operatively coupled with said ultrasound source means and said ultrasound detector means for controlling the emission of said multitude of different ultrasound signals and the detection of backscattered signals. The absolute perfusion II of an analysis region with the liquid is calculated by direct measurement of backscatter signals in an analysis region and an adjacent reference region.

Description

Method and system for determining the absolute perfusion rate of a fluid in an analysis region

The invention relates to a method and a system for determining the absolute perfusion rate of a fluid in an analysis region.

Such a method can be used for any fluid in a system having a fluid flow through conduits. One of said methods is known from US 6,186,951 which can be used to assess the concentration of a contrast agent used in said system and or a particle size.

US 2003/0092991 discloses another method and apparatus to control microbubble destruction through contrast-enhanced ultrasound imaging, wherein said microbubbles constitute an ultrasound contrast agent .

Further apparatus are known from US 6,315,730, US 2001/0056236, US 6,340,348 and WO 03/052453 mainly showing systems using ultrasound imaging with microbubbles as ultrasound contrast agent.

However, these methods and systems are not quite appropriate to calculate the absolute value of the perfusion. These methods only provide relative information.

It is therefore an object of the invention to provide a method and a system giving the user of the system to calculate directly said absolute perfusion value.

The object of the invention is achieved through the features of claim 1 for a method and through the features of claim 5 for an apparatus . The absolute perfusion 77 of a region with the liquid is calculated by direct measurement of backscatter signals in an analysis region and an adjacent reference region.

One possible application of the invention - among others - is the measurement of the flow of blood within soft tissues. Until now absolute myocardial perfusion, the gold standard to assess regional blood supply, could only be studied by positron emission tomography (PET) . Based on the ultrasound contrast agent kinetics according to the invention, the measurement of absolute perfusion by myocardial contrast echocardiography is possible and has been demonstrated by its in vitro validation using a icrocirculation-mimicking phantom. Its in vivo applicability to myocardial perfusion has been shown by the comparison with PET measurements of a healthy volunteer. Furthermore, the invention generally describes the microcirculation and therefore the method and the system may be applied to quantify absolute perfusion of other organs than the heart and perfusion outside humans or animals.

The wording "absolute perfusion rate of a fluid in an analysis region" relates in other words - for the above mentioned application - to the "absolute blood volume flow in relation to the tissue weight". This value is also known under "blood volume flow rate" .

A signal intensity according to the invention is obtained by determination of a value from perfusion sequences, e.g. recorded video images, which is proportional to UCA concentration.

Further preferred embodiments of the method and apparatus according to the invention are characterised in the dependent claims .

Now, the invention will be explained exemplarily more in detail with reference to the appended drawings .

There is shown in:

Fig. 1A to ID shows an in vitro refill sequence in four video frames indicating the profile of a filter and a tube mimicking the left ventricle,

Fig. 2 shows the microphantom contrast signal distribution and plots the video intensity of a cross shaped region of interest against time,

Fig. 3A and 3B shows a representative refilling curve of filter one with a pump flow rate of 160ml/min. ,

Fig. 4 shows a summary of perfusion studies of filter one (left panels) and two (right panels) ; parameter β (A) , intravascular volume fraction μ (B) and filter flow rate assessed by MCE (C) are plotted versus calibrated pump flow rate Q_Pu_mp; solid lines indicate regression lines and 95% confidence intervals, and

Fig. 5A and 5B shows a perfusion sequence and refill curve of the mid interventricular septum from a healthy volunteer.

One possible application of the invention - among others - is the measurement of the flow of blood within soft tissues. Convection and diffusion are the responsible mechanisms for the transport of nutrients to cells and the removal of their excreta. Diffusional processes are driven by concentration gradients between microcirculation and adjacent cells. In order to sustain these gradients, microcirculatory blood volume per cell mass (ml-g^-1) has to be replaced at an appropriate rate (min^-1) . Therefore, the product of microcirculatory blood volume per cell mass and exchange rate, also referred to as absolute perfusion given in ml -min^-1 -g^-1, is the most suitable quantity to assess microcirculatory function.

To date, human studies of absolute myocardial perfusion can only be performed by PET. Since the introduction of ultrasound contrast agents (UCA) , attempts have been carried out to assess absolute myocardial perfusion using Myocardial Contrast Echocardi- ography (MCE) . UCAs consist of gas-filled microspheres or so called microbubbles that are free-flowing tracers of the in- travascular compartment and their kinetic behavior at the microcirculatory level provides information on regional blood supply. The most promising method evaluated UCA refill curves following ultrasound-induced microsphere destruction during constant venous UCA infusion. It has been shown that the video signal intensity y (t) arising from UCA refill follows an exponential curve y[t) = A-\ -e^~βt) . Parameter A and β were interpreted as myocardial blood volume and myocardial blood velocity, respectively. In vitro, ex vivo and in vivo studies revealed statistic correlation between absolute perfusion and β as well as A-β. However, due to their semiquantitative character, they only determine relative perfusion changes supposed that identical ultrasound conditions are guaranteed. The fact of which clearly limits the method's practicability and emphasizes the need for a truly quantitative measure for clinical and research purposes.

The biophysical concept of absolute perfusion enforces models of UCA kinetics that consider spatial aspects of the microcirculation and we hypothesize that absolute perfusion can be derived from such a model. Objectives of this study were (1) to develop a volumetric model of UCA kinetics that reproduces the experimental refill curve and yields absolute perfusion by parameter identification, (2) to validate this model by in vitro experi- ments using hemodialysis filters and (3) to demonstrate its in vivo applicability by assessing regional resting myocardial perfusion of a healthy volunteer by MCE and PET.

The volumetric model of UCA kinetics is derived from the equation of continuity:

₍ 1 ) l ^dC(t^^z)dV + jC(t,x,y,_Z)-v(t,x,y,z)-dS = θ ,

where C and vare the spatiotemporal distributions of micro- spheres and flow velocities, respectively. Vjy and S denote volume and surface of the intravascular volume under investigation. Since ultrasound cannot resolve microcirculation, microsphere concentration inside V_y is set to its spatial average C(t) . Hence, integration of ( 1 ) leads to:

where C;„ is the microsphere concentrations of the influx. Micro- sphere concentration of efflux is set to C(t) . Q indicates blood flow into and out of Vjy. The input function arising from ultrasound-induced microsphere destruction is approximated by a step function from Co, the microsphere concentration immediately after destruction, to C,-„. Integration of ( 2 ) gives:

According to scattering theory and experimental data, there is a linear relationship between microsphere concentration and back- scattered energy until saturation occurs for higher UCA concentrations. By selecting appropriate concentrations, saturation is prevented and equation ( 3 ) can be transformed into:

where yo and y are initial and steady state signal intensity of the refill curve, respectively. Comparison with the experimental refill curve yields the volumetric interpretation of fitting parameter β being the frequency (min^-1) of blood volume replacement within an observed quantity of tissue by its supplying blood flow:

(5) β =~ - V_IV

Absolute perfusion 77 is then obtained by considering tissue density p_τ (g'ml^"1) and total volume under investigation V_τ (ml) :

Sometimes the absolute perfusion is given as: (7) π = β-μ

and therefore expressed in ml/min for ml tissue, especially if the tissue density is not well known.

Parameter μ (ml -ml^-1) represents the intravascular volume ratio (IVR) of V_τ and is assessed by comparing video signal intensities arising from the microcirculation and an adjacent constituent of the macrocirculation such as a cardiac cavity or a large vessel. This is based on the concept that the UCA effect emerging from the microcirculation is "diluted" by the extravascular compartment. For linear operational range and steady state conditions applies : (8) y macro = k - C_in and

O) y_micro = y =k ^C-^^v v_τ

where y_mi_Cm and y_maCw are the video signal intensities of the micro- and macrocirculation and k is the constant of proportionality. IVR then calculates to:

(10) _/=2 =_^y_. V T J v macro

Within the myocardial contrast echocardiography (MCE) according to the invention a Sequoia C256 ultrasound scanner equipped with a 3VC2 transducer and Coherent Contrast Imaging (CCI) was used for all imaging (Siemens Medical Solutions, Mountain View CA, USA) . CCI provides low mechanical index (MI) , nonlinear imaging for real-time MCE. Settings were as follows: MI for microsphere detection 0.08, MI for microsphere destruction 1.3, dynamic range 60dB, linear post processing, no depth gain control, clip length 200 frames, timed trigger imaging.

MCE was performed using constant infusion of OPTISON (Amersham Health SA, Oslo, Norway) by means of an infusion pump (Perfusor f , B. Braun Melsungen AG, Melsungen, Germany) . To guarantee UCA homogenization, the infusion pump was manually agitated.

Refill sequences were generated using the manual bubble destruction (MBD) feature of the scanner. Offline image display and quantification was done with DataPro 2.11 (Siemens). Video intensity data were exported to SigmaPlot 7.0 (SPSS Science, Chicago IL, USA) for linearization and fitting. Linearization of video intensity data was performed to remove logarithmic signal compression, linearized signal intensities were expressed in acoustic units (AU) . In vitro Perfusion Studies had been conducted wherein the core of the experimental setup consisted of a hemodialysis filter and a polypropylene based tube (inner diameter 8mm, wall thickness 1.6mm, Pharmed, Ismatec, Glattbrugg, Switzerland), mimicking microcirculation and adjacent macrocirculation. Filter and tube, termed microphantom and macrophantom, were connected in series and aligned in parallel with a horizontal displacement of 50mm. Two filters with polycarbonate housings have been investigated. Filter one: Filtral 20 AN69F (Hospal, Meyzieu, France), fiber number 10' 500, inner fiber diameter 240μm, cross sectional area (CSA) 14.52cm², IVR 0.327. Filter two: Hemoflow HF80S (Fresenius Medical Care AG, Bad Homburg, Germany), fiber number 12' 288, inner fiber diameter 200μm, CSA 18.10cm², IVR 0.213. Filter cartridges were not furnished with acoustic windows, since fiber integrity is crucial for IVR determination. The phantom was mounted in upright position to prevent asymmetric gravitational effects on flow profile and buoyant icrospheres within filter inlets. Physiologic saline, held at a constant temperature of 37°C (MR 3001K, Heidolph Instruments, Schwabach, Germany), was driven downwards the macrophantom and upwards the microphantom by means of a calibrated peristaltic pump (MCP Standard PRO-280, Ismatec, Glattbrugg, Switzerland) . Downstream the pump, a polyethylene conduit (length 200cm, inner diameter 8mm, wall thickness 1mm; Semadeni AG) served as windkessel. A closed circuit was established by removing the microspheres via the transmem- branous path of a second hemodialysis filter (Hemoflow HF80S) downstream the microphantom.

For experiments, the phantom was immersed in tap water. Acoustic properties of the transmitting media, i.e. physiologic attenuation and propagation velocity, were not strived since the quantification method only exploits relative contrast signal. Scan angle was adjusted for optimal visual appearance of the ultra- sound image in the range of 80 - 85°. OPTISON was diluted with physiologic saline (1:9) and infused at a rate of 20ml-h^-1. Filter one (2 series with complete replacement of filters and tubing) and two were investigated with flow rates from 30 to 230ml -min^-1 and 10 to 200ml -min^-1, corresponding to filter flow velocities from 1.1 to 8.1mm-s^_1 and 0.4 to δ.δmm-s^-1, respectively.

For perfusion analysis, signal intensities were averaged across the selected region of interest (ROI) inside the microphantom. The first frame after MBD was used to correct for non-contrast signals arising from filter capillaries or trapped air. Fitting to equation ( 4 ) yielded β and y . Spatiotemporal averaging across the lumen of the macrophantom yielded y_ma r_oι whereat frames during MBD and macrophantom refill were discarded.

Contrast signal distribution inside the microcirculation- mimicking phantom was assessed by grey tone coded perfusion maps 2 of piecewise (15x15 pixel ) reconstructed refill curves using own developed image quantification software based on Matlab 6.5 Release 13 and Imaging Processing Toolbox 3.2 (The MathWorks, Na- tick MA, USA) .

For the comparison with the independent variable, i.e. pump volume flow rate, absolute filter perfusion TI_MCE was transformed into filter volume flow rate Q_MCE - ( 1 1 ) MCE ^" ^ MCE beam ^• Pjs tor '

where V_beam is the beam volume of CCI mode inside the filter and P_fnt_er is filter density. The rationale for this transformation was as follows. The perfusion unit is defined as the core entity that structures a tissue by parallel connection of its blood supply. In contrast to biologic tissues, perfusion unit length of the microphantom, i.e. fiber length, is larger than beam elevation. Therefore, the volume flow rate that mediates contrast refill within the beam volume equals the total volume flow rate of the filter. V_bam equals filter CSA times beam elevation of CCI mode that can be calculated from MBD beam elevation and contrast arrival delay (see below) . MBD beam elevations were measured in preliminary studies using filters with immobilized microspheres and revealed CCI beam elevations of 4.3 and 3.5 mm for filter one and two, respectively.

Within an in-vivo perfusion study absolute regional resting myocardial perfusion of a 22-year-old healthy female volunteer was assessed by MCE and ¹³N-ammonia PET (GE Advance PET-Scanner, General Electrics, Medical Systems, Milwaukee, USA) . The study protocol was approved by the ethics committee of the University of Bern and written informed consent was obtained. UCA was administered via the right cubital vein using parallel infusion of OPTISON 6ml at a rate of 20ml -h^-1 and physiologic saline at a rate of 00ml-h^-1 (VOLUMED^® μVP2001 , arcomed ag, Regensdorf, Switzerland) . Refill sequences with frame intervals of 75ms were recorded from standard and atypical views to guarantee optimal visualization of all myocardial segments.

Within the data analysis refill curves were reconstructed by selecting one frame per captured cardiac cycle at constant cycle time. Selection of cycle time was based on visual quality of myocardial enhancement. ROIs were placed and tracked manually within the myocardium and in close proximity inside the left ventricle. Frames befoxe MBD yielded y , y a_cro was calculated using all selected frames but the first after MBD. Correction of potential non-contrast signals from the myocardium was achieved by background subtraction of the first frame after MBD. The initial signal intensity yo can be obtained through use of the first signal intensity after initial discernable modification of the ultrasound contrast agent. Said first signal intensity is the signal intensity retrieved from the first frame after MBD.

Statistical data were expressed as mean (standard deviation (SD) ) . Comparisons between two groups were made with student's t test. Differences were considered significant at P<0.05. Curve fitting was performed using least-squares method. Correlations were demonstrated using linear regression analysis and accuracy of the prediction was measured by the standard error of the estimate (SEE) .

For in-vitro perfusion studies a representing sample of contrast signal distribution from filter two at a pump flow rate of 100ml -min^-1 is shown in Fig. 1A, IB, 1C and ID. As mentioned above the in vitro setup comprises microcirculation (region 1) and left vetricular cavity (region 5) modeled by a haemodialysis filter and a tube. The phantom was perfused with saline by means of a peristaltic pump.

Fig. 1 shows the in vitro refill sequence in video frames. The four frames in Fig. 1A to ID show the profile of the filter 1 and the LV-mimicking tube 5 before microsphere destruction in Fig. 1A, immediately after destruction in Fig. IB, during the course of the refill in Fig. 1C and plateau phase in Fig. ID. The pump-generated volume flow rate was set to 100ml/min. Fig. 1 shows an inverted b/w image, therefore zones 3 without the presence of microspheres are white. Filter 1 and tube 5 are filled in Fig. 1A with the fluid in presence of microspheres and are therefore gray. In Fig. IB the filter 1 has undergone a massive microsphere destruction and has become mainly white whereas the content of tube 5 did not change (or was replaced by fresh flow of microspheres) . In Fig. 1C and Fig. ID the gray level in- creases in the filter 1 because the level of microsphere concentration gradually rises.

The region of interest (ROI) within the filter 1 and the tube 5 are indicated through an ellipse 4 and a circle 2, respectively.

Fig. 2 shows the microphantom contrast signal distribution (filter two, pump flow lOOml/min) . Fig. 2A shows a first frame of a perfusion sequence representing the profile of the microphantom (left) and the macrophantom (right) . The rectangles 30 and 31 indicate ROIs for the assessment of transversal and longitudinal signal distribution, respectively. Original video signal intensities for the transversal and longitudinal signal distribution were plotted versus time in Fig. 2B and Fig. 2 C, respectively; MBD was performed between 0.9 and 1.6s. Mapping of time- intensity curves is indicated by the gray value bar representation of the ROIs 30 and 31.

Visually, the filter cross section was divided in four zones. The core zone showed homogenous contrast enhancement and included the medial portion of the proximal two thirds. The attenuation zone behind the core zone resulted from contrast shadowing. The lateral borders of the core zone were formed by transition zones with decreasing video signal intensities due to impaired acoustic coupling into the filter. The transition zones faded to echo-free zones owing to total reflectance on the housing surface. The echo-dense areas behind the phantoms emerge from mounting structures.

The ROI was divided into zones 41 to 46 and 51 to 60, respectively. The ROI shown in Fig. 2A, the signal intensity in Fig. 2B and the schematic representation of the part of the transversal rectangle 30 in Fig. 2 B have received the same reference numerals. The same principle has been applied for the ROI 51, 52, .. in Fig 2A and 2C for the longitudinal rectangle 31.

All time-intensity curves were superposed by continuous jitter, which mainly consisted of harmonics from the pump rotary frequency. Video signal intensity distribution before MBD and during plateau phase was homogenous only within the core zone. MBD, indicated by the simultaneous and rapid decline of the curves to approximately OdB, was started at 0.9s. Three curves of the longitudinal ROI showed residual enhancement that arose from remaining air bubbles due to incomplete venting, as was apparent in pre-contrast images.

Between MBD termination at 1.6s and refill curve upstroke, a silent interval was interposed representing delayed contrast arrival within the observed volume. For both filters, contrast arrival delays were inversely proportional to fiber flow velocities (data not shown) . Refill curves of the core zone showed parallel upstrokes, indicating homogenous flow velocity distribution. Flow velocity distribution within attenuation and transition zones cannot be assessed due to their poor signal to noise ratio.

Fig. 3 depicts a representative refill curve from filter one recorded at a pump flow rate of 160ml •min^-1. Fig 3A shows the video intensity in dB whereas Fig. 3B shows the linearized signal of Fig. 3A in acoustical units.

The refill curve 11 demonstrates exponential rise to plateau 12 and the signal within the tube 13 is constant as denoted by the dotted line 14 that indicates mean intensity. The dashed line 15 represents the fitting curve for signal 11. The rectangle 16 indicates the time period of microsphere destruction. MBD was performed between 0.5 and 2.1s and was followed by the contrast arrival delay. For both filters, contrast arrival delays were inversely proportional to fiber flow velocities (data not shown) . Due to the high flow velocity inside the macrophantom (53. lmm- s^-1) , its contrast signal was merely affected by MBD. The sharp peak immediately after MBD represents pressure bursts from microsphere destruction by MBD that were detected by CCI, which is very sensitive to nonlinear signals. Analysis of linearized curves revealed y_mαcr0=17.33AU, y=5.74AU and /3=95.4min^_1 (Fig 3.B, upper and lower dashed line) . IVR and flow rate assessed by MCE amounted to 0.33 and 157.8ml -min^-1 -g^-1, respectively .

Fig. 4 summarizes perfusion data of filter one (left panels A.l, B.l and C.l) and filter two (right panels A.2, B.2 and C.2). The relation of β versus Q_pUmp revealed good linear correlation and comparable precision for filter one and two (Fig. 4 A.l r²=0.98, SEE=5.69min^-1; A.2 r²=0.98, SEE=5. δgmin^""1) . Linear regression analysis revealed no correlation between flow rate and μ of filter one ( Fig. 4 B.l r²=0.07) and weak correlation for filter two (Fig. 4 B.2: r²=0.36). For low flow rates, IVR of filter one was overestimated due to insufficient fitting of incompletely recorded slow refill curves and the data (three measurements) were excluded for the further analysis. Mean (SD) estimates of IVR were 0.284 (0.028) and 0.219 (0.023) for filter one and two, respectively. Filter flow rates Q_MCE (Fig. 4 C.l and C.2) correlated linearly with Q_Pum_P for filter one (r²=0.91, SEE=16.02ml-min^_1) and two (r²=0.9β, SEE=8.26ml -min^-1) . The regression line of filter one closely aligns with the line of identity. For filter two, intercept and slope of the regression line indicate systematic underestimation of Q_pUm by about 10%. Within in-vivo perfusion studies heart rate and mean arterial blood pressure during MCE (PET) were 68 (62) min^""1 and 86 (78) mmHg, respectively. Fig. 5A shows a representing perfusion sequence of the interventricular septum.

Fig. 5A shows the sequence of endsystolic frames, recorded from the four-chamber view, representing the myocardial UCA refill in frames F3 to F16. The first two frames (top left Fl and F2) were captured before MBD that was performed from 1.65s to 3.45s. ROIs for refill curve reconstruction (line 21) and reference signal measurement (line 22) are depicted on the first frame. Fig. 5B shows linearized myocardial (•) and left ventricular (▼) time- intensity curves.

The refill curve of the mid interventricular septum (Fig. 5B) did not indicate delayed contrast arrival; y and y_maCro were 1.251AU and 9.168AU, respectively (Fig. 5B, upper and lower dotted line) . Fitting of- the refill curve to the model (Fig. 5B, solid line) yielded β = 7.289min^_1. IVR and perfusion amounted to 0.136 and 0.947ml -ml^-1 -g^"1.

It was possible to analyze perfusion of every segment by MCE (Table 1) . The refill curves of nine segments did not indicate delayed contrast arrival. The three anterior segments showed a slow initial rise and three apical segments showed delayed rise. Parameter β and μ as well as absolute perfusion demonstrated seg- mental heterogeneity. Except for the posterolateral wall, β decreased from apex to base, while μ showed the opposite behavior. Perfusion data distribution was more complex, however MCE and PET data demonstrated similar' patterns. Overall, MCE and PET perfusion data were in good agreement, main differences were encountered in inferior and lateral segments and mean segmental perfusion did not differ significantly. Table 1 Absolute regional myocardial perfusion at rest by MCE and PET

Segment^* P . IIMCE ΓIPET (min^"1) (ml-mr¹) (ml-min-g^"1) (ml-min-g^"1) (%) SEPTAL APEX 11.905 0.061 0.697 0.683 2.1 MID 7.289 0.136 0.947 0.899 5.1 MID-AS* 10.984 0.093 0.975 0.974 0.1 BASAL 8.896 0.112 0.945 0.861 8.9 BASAL-AS 7.560 0.150 1.079 0.933 13.5

ANTERIOR APEX 6.368 0.152 0.920 0.899 2.3 MID 6.371 0.147 0.894 0.861 3.7 BASAL 4.508 0.192 0.826 0.717 13.2

LATERAL APEX 10.052 0.072 0.69 0.896 -29.8 MID 5.170 0.149 0.732 0.689 5.9 BASAL 4.413 0.173 0.728 0.639 12.2

INFERIOR APEX 6.453 0.145 0.891 0.981 -10.1 MID-PL^§ 3.965 0.158 0.597 0.683 -14.3 MID 4.741 0.200 0.905 0.779 13.9 BASAL-PL 6.295 0.113 0.680 0.701 -3.1 BASAL 4.268 0.182 0.740 0.643 13.2 Mean 6.83 0.140 0.828" 0.802^" 2.3 (SD) (2.48) (0.040) (0.135) (0.122) (12.0)

Sixteen-segment model proposed by the American Society of Echocardiography.

I Δ=I oo%-(π_MCE-π_PEτ)/π_MCE.

* AS: anteroseptal. _PL: posterolateral.

^" Difference statistically not significant. The presented method was derived from a volumetric model of UCA kinetics at the microcirculatory level that was developed using principles of biophysical modeling. Microcirculation and measurement chain were understood as a sole system that can be described either by its transfer function or by a biophysical model. The transfer function can be obtained by disturbing the system with a defined input function, i.e. transient microsphere destruction, and observing the system's reaction, i.e. the refill curve. A biophysical model can be gained by defining the physical phenomenon and its biological environment. Since transfer function and biophysical model are equivalent, it is possible to estimate model parameters such as absolute perfusion from fitting parameters of the transfer function and vice versa, provided that parameter identification is successful. Therefore, measurand and biophysical process determine the design of the model. Since absolute perfusion is a volumetric quantity, a spatial model must be used to guarantee parameter identification. The physical process is transport by convection, thus the equation of continuity is a reasonable model. This volumetric model is more complex than the one-dimensional model proposed by Wei et al . , in Wei K, Jayaweera AR, Firoozan S, et al . Quantification of myocardial blood flow with ultrasound-induced destruction of microbubbles administered as a continuous venous infusion. Circulation 1998;97:473-483.

However the model according to the invention 'is able to reproduce the experimentally determined refill curve and yields absolute perfusion by parameter identification. Furthermore, the quantification method does not depend on specific contrast agents, contrast imaging techniques or scanner settings if contrast signals vary linearly with microsphere concentration. Testing the volumetric model requires tissues with well-defined intra- and extravascular compartments. While biologic tissues do not meet this condition and standardized microcirculation- mimicking phantoms are lacking, the use of hemodialysis filters is the most suitable substitute. Although their geometrical structure constitutes a gross approximation of microcirculatory complexity and capillary size, they fulfill the assumptions of the volumetric model when examined perpendicular to fiber length axis and enable to simulate physiologic flow velocities ranging from 0.3 to lOmm/s.

In agreement with published data, β demonstrated strong linear correlation with pump flow rates. In contrast to Lafitte et al, "Accuracy and reproducibility of coronary flow rate assessment by real-time contrast echocardiography, in vitro and in vivo studies" J.Am.Soc.Echocardiogr 2001, volume 14(10), pages 1010 to 1019, parameter β did not vary inside the filter core zone. Their finding of increasing plateau intensities form proximal to distal filter areas may indicate relevant microsphere destruction during contrast imaging that in turn prolonged refill time with increasing local MI and frame rate.

The estimation of myocardial blood volume by its steady state signal intensity is affected by acoustic conditions, equipment settings as well as performance of contrast administration and therefore normalization with left ventricular signal intensity has been proposed. However, there are no published data demonstrating the calibratability to absolute perfusion. Alternatively, normalization with a reference segment of the myocardium was used. While such a method may allow semiquantitative perfusion assessment, calibration to absolute perfusion is not possible. In contrast to that, IVR is a biophysical perfusion parameter and its assessment, although similar to normalization, su- persedes the need for calibration. IVR of filter one was underestimated systematically (mean -13.1%), while filter two did not demonstrate systematic differences (mean 2.8%). Due to the lack of data regarding variability of filter fiber number, inner fiber diameter and filter housing diameter, statistical comparison of μ and IVR derived from structural filter data is not feasible. Potential differences may be caused by different acoustic properties of filter housings and fibers, by insufficient transfer of microspheres via the filter inlet or nonlinear dilution function of the intravascular contrast signal by the extravascu- lar compartment .

Regression line slope of Q_pump versus filter perfusion (graph not shown) correspond to the reciprocal of average Vbeam and thus reveals an estimate of average CCI beam elevation. Average beam elevation of filter one was 4.3mm and matches with data from gel phantoms . Average beam elevation of filter two was 3.9mm or 10.3% above the value from gel phantoms. This result indicates that underestimation of β may explain the underestimation of Qpu p by QMCE (Fig. 4B) .

In vivo applicability of the method was tested by investigating resting myocardial perfusion of a healthy volunteer. In consideration of methodical differences between tomographic and volumetric assessment by MCE and PET, the perfusion data further support the correctness of our method. Substantial differences of inferior and lateral segments may be related to splenic spill over affecting PET data of the inferior wall and suboptimal echocardiographic visualization of the lateral wall. There are only few studies reporting data on parameter β from myocardial perfusion at rest, mainly from patients with heart disease. The comparison with these data is further hampered by inconsistent use of linearization and fitting methods. A reference method for the in vivo assessment of the myocardial blood volume is lacking and published ex vivo data from humans are scarce. Histological morphometry of eight adults, who died without apparent left ventricular disease, yielded a mean myocyte fractional area of 82.1 (0.9)%, indicating a mean IVR below 18%. In vivo and ex vivo quantification of IVR from various mammals revealed a range of 5 - 20%.

Since beam elevation enlarges with increasing MI, contrast arrival delay is a consequence of real-time imaging techniques using different MI for microsphere destruction and detection. While contrast arrival delay was invariably encountered in vitro, there was no evidence for a similar phenomenon in vivo. This finding can be explained by the perfusion unit length of the filter and the myocardium. The former is larger than CCI beam elevation, thus prior to refill the detection volume, micro- spheres have to perambulate the distance between destruction and detection zone. The latter is much smaller than CCI beam elevation, i.e. the destruction volume consists of many perfusion units that are, according to their definition, refilled simultaneously.

The system input from MBD rather follows some type of saturation than a step function. The impact of this approximation on the quantification of β was not investigated. The use of more sophisticated input functions may improve the model, but it remains questionable if parameter identification and stable fitting still will be feasible.

Microsphere concentration fluctuations within the macrocircula- tion, e . g. due to MBD, may propagate into the microcirculation and interfere with the quantification method. While this was prevented in vitro by adapting the microsphere containing volume between macro- and microphantom, it was not studied systematically in vivo and may play a role, most notably in segments with short supply lines.

The IVR of examined filters did not cover the range of in vivo data and the reliability of the quantification method for this range bases on theoretical considerations of acoustic properties. Appropriate in vitro studies were not performed, since hemodialysis filters with an IVR clearly below 20% were not available.

The volumetric model is a valid description of UCA kinetics at the microcirculatory level and allows the quantification of absolute perfusion of a microcirculation-mimicking phantom and, likely, the human myocardium using contrast echocardiography and ultrasound-induced microsphere destruction. Furthermore, the volumetric model generally characterizes the microcirculation and therefore the presented method may be used to assess absolute perfusion of any tissue accessible to ultrasound.

Claims

1. A method for determining the absolute perfusion rate of a fluid in an analysis region (ROI, 4, 21) , comprising

- introducing an ultrasound contrast agent (UCA) into the fluid at an upstream site from the analysis region (ROI) monitoring contrast-agent backscatter in the analysis region (ROI) to determine a steady state signal intensity y of the ultrasound contrast agent (UCA) in the analysis region (ROI) ,

- monitoring contrast-agent backscatter in a reference region (RR, 2, 22) to determine an signal intensity y_macw of the ultrasound contrast agent (UCA) in the reference region (RR) ,

- applying a sequence of pulses of ultrasonic energy sufficient each to destroy or discernibly modify the ultrasound contrast agent (UCA) such that the contrast-agent backscatter is reduced in the analysis region (ROI) ; and

- a.) monitoring the contrast-agent, backscatter in the analysis region (ROI) over time t after any of said modifying pulses with a plurality of non-destructive imaging pulses to obtain time dependant backscatter values y(t) of the analysis region (ROI) , or b.) monitoring the contrast-agent backscatter in the analysis region (ROI) over time t after each of said pulses with one non-destructive imaging pulse applied in different temporal distance from the above-mentioned modifying pulses to obtain time dependant backscatter values y(t) of the analysis region (ROI), or c.) monitoring the contrast-agent backscatter in the analysis region (ROI) over time t using a sequence of subsequent modifying pulses as imaging pulses having different temporal distances one from another to obtain time dependant backscatter values y(t) of the analysis region (ROI), - determining the absolute perfusion IT using the initial signal intensity yo after the one or the last pulse of ultrasonic energy sufficient each to destroy or discernibly modify the ultrasound contrast agent, using the steady state signal intensity 5> of the analysis region (ROI) and using the signal intensity ymacr_o of the ultrasound contrast agent (UCA) in the reference region (RR) .

2. Method according to claim 1, wherein yo is obtained by fitting of values y(t) or through use of the first signal intensity after initial discernable modification of the ultrasound contrast agent .

3. Method according to claim 1 or 2, wherein calculating the absolute perfusion JJ comprises:

- calculating β in

wherein y is obtained through averaging of the signal over time without discernable modification of the ultrasound contrast agent or by fitting of values y(t), - calculating μ from

calculating absolute perfusion IT from π = β- μ

where β and μ are the above calculated values.

4. The method according to one of claims 1 to 3, characterized in that the reference region is a region in the vicinity of the analysis region wherein said reference region is mainly filled with said fluid.

5.. The method according to claim 4, characterized in that the reference region is connected with the analysis region (ROI) through short conduits being able to transport said fluid and said ultrasound contrast agent (UCA) .

6. The method according to one of claims 1 to 5, characterized in that the ultrasound contrast agent (UCA) is a biocom- patible gas stabilized by amphiphilic liquid material.

7. An ultrasonic apparatus for per orming perfusion measurements of a fluid carrying an ultrasound contrast agent (UCA) into an analysis region (ROI) , the apparatus comprising:

- an ultrasound source means capable to provide a multitude of different ultrasound signals, an ultrasound detector means capable to detect backscatter of said multitude of different ultrasound signals,

- a processor operatively coupled with said ultrasound source means and said ultrasound detector means for controlling the emission of said multitude of different ultrasound signals and the detection of backscattered signals, wherein the different ultrasound signals can be directed in two different adjacent regions (ROI and RR) , named analysis region (ROI) and reference region (RR) ,

- wherein the processor controls said ultrasound source means and said ultrasound detector means to monitor contrast-agent backscatter in the analysis region (ROI) to determine a steady state signal intensity y of the ultrasound contrast agent (UCA) in the analysis region (ROI) ,

- to monitor contrast-agent backscatter in a reference region (RR, 2, 22) to determine an signal intensity y_macro o the ultrasound contrast agent (UCA) in the reference region (RR) , to apply a sequence of pulses of ultrasonic energy sufficient each to destroy or discernibly modify the ultrasound contrast agent (UCA) such that the contrast-agent backscatter is reduced in the analysis region (ROI) ; and a.) to monitor the contrast-agent backscatter in the analysis region (ROI) over time t after any of said modifying pulses with a plurality of non-destructive imaging pulses to obtain time dependant backscatter values y(t) of the analysis region (ROI) , or

- b.) to monitor the contrast-agent backscatter in the analysis region (ROI) over time t after each of said pulses with one non-destructive imaging pulse applied in different temporal distance from the above-mentioned modifying pulses to obtain time dependant backscatter values y(t) of the analysis region (ROI), or c.) to monitor the contrast-agent backscatter in the analysis region (ROI) over time t using a sequence of subsequent modifying pulses as imaging pulses having different temporal distances one from another to obtain time dependant backscatter values y(t) of the analysis region (ROI), the apparatus further comprising calculating means to determine the absolute perfusion IT using the initial signal intensity yo after the one or the last pulse of ultrasonic energy sufficient each to destroy or discernibly modify the ultrasound contrast agent, using the steady state signal intensity y of the analysis region (ROI) and using the signal intensity y_macr₀ of the ultrasound contrast agent (UCA) in the reference region (RR) .

8. The ultrasonic apparatus according to claim 7, wherein the calculating means determine yo by fitting of values y(t) or through use of the first signal intensity after initial discernable modification of the ultrasound contrast agent.

9. The ultrasonic apparatus according to claim 7 or wherein the calculating means - determine β in

wherein y is obtained through averaging of the signal over time without discernable modification of the ultrasound contrast agent or by fitting of values y(t), - calculate μ from y macro and determine absolute perfusion IT from U = β-μ

where β and μ are the above calculated values, in order to output said absolute perfusion IT.