WO2009056857A1

WO2009056857A1 - Attenuation correction in ultrasound contrast agent imaging

Info

Publication number: WO2009056857A1
Application number: PCT/GB2008/003705
Authority: WO
Inventors: Mengxing Tang; Robert J. Eckersley; Jean-Martial Mari
Original assignee: Imperial Innovations Limited; Chancellor, Masters And Scholars Of The University Of Oxford
Priority date: 2007-10-31
Filing date: 2008-10-31
Publication date: 2009-05-07
Also published as: GB0721450D0

Abstract

A method of calibrating ultrasound contrast agent signal data from a location in a medium containing a contrast agent, the method comprising the steps of: inputting first echo signal data from a first plurality of echo signals from time series ultrasound pulses, the ultrasound pulses differing in at least one of phase, amplitude and pulse duration; determining the contrast agent signal data from the first echo signal data from one or more of the echo signals; determining an attenuation correction factor from the first echo signal data from at least one of the echo signals; calibrating the contrast agent signal data with the attenuation correction factor to obtain calibrated contrast agent data for the location; outputting the calibrated contrast agent data.

Description

Attenuation correction in ultrasound contrast agent imaging

The present invention relates to a method and system for calibrating ultrasound contrast agent signal data from a location in a medium containing a contrast agent or agents.

Tissue perfusion is a crucial indicator in clinical assessment of a wide range of clinical conditions such as heart disease and cancer. Consequently, there is a pressing need for an accurate and convenient means of quantifying perfusion, for example in imaging applications.

Ultrasound offers many advantages for taking in vivo measurements and as an imaging modality in terms of its flexibility, safety, cost and potential for realtime acquisition. The use of contrast agents which enhance the sensitivity of ultrasound to blood flow make ultrasound a useful tool for measuring perfusion. The contrast agents can be disrupted in a controlled way to facilitate imaging the replenishment of the tissues and also to facilitate drug/gene delivery if the agents are loaded with or surrounded by drugs/genes. Furthermore, these agents can also be used for molecular imaging if some target-specific ligands are attached to the agents.

Typical ultrasound contrast agents are micrometer sized bubbles consisting of a gas core stabilized by a flexible shell. When injected into the vein, they stay in the blood pool and flow where the blood flows, through both macro and micro blood vessels in the body. Therefore the concentration of bubbles can give an indication of location of blood and amount of blood at different locations within tissues. The rate of change of this concentration can also reveal further detail of blood flow into an organ or tissue region.

In conventional B-mode ultrasound imaging an image of microcirculation within tissues is produced from ultrasound echoes received from tissue containing bubbles present in blood flowing therethrough. At resolutions achievable with ultrasound, such an echo contains signals relating to both tissue and bubbles. These are shown together in the B-mode image.

Recently techniques such as Pulse Inversion (PI) have been developed to remove signals from tissue and extract signals only related to microbubbles, and thus to blood perfusion. Specifically, these techniques make multiple measurements for the imaged region and, based on these measurements, quantities related to the microbubble concentration can be derived. These quantities may be displayed as a bubble- specific image with a grey scale value or colour corresponding to concentration of the microbubbles.

Such techniques lead to the possibility of quantification of tissue perfusion.

However, without accurate correction for attenuation, such quantification is of limited accuracy. Accurate measurement and quantification of bubble concentration and of how the concentration changes over time within tissue can provide crucial information on tissue function. In particular this can be invaluable in clinical investigations of patients with heart diseases and cancer.

However there is currently a lack of an effective method for attenuation correction, which is a major problem for quantification of tissue microcirculation. When an ultrasound signal transmits though tissue containing microbubbles, it is gradually attenuated. The ultrasound echo from a target is affected by attenuation due to overlaying tissues. This must be corrected, or compensated for, to obtain an accurate measure of the echogenicity of the target. Currently, manual correction of attenuation may be performed through adjustments made by an operator in a process known as 'time gain compensation' (TGC) which uses preset values to correct for attenuation according to the distance between transducer and target. An overall gain function is then applied to all scan lines in the image. As a result the correction is crude, with significant artefacts often present in images. Consistency and reliability of any quantification is therefore compromised.

There have been some techniques developed for automatic attenuation correction for standard B -mode images which have shown some improvement over preset and manual TGC. However these are not applicable to imaging modes such as PI mentioned above which extract bubble signals. This is because the bubbles exhibit nonlinear behaviour under ultrasonic excitation, which makes attenuation correction complicated. For example the nonlinear behaviour generates higher harmonics and these are attenuated more than the fundamental frequency. Furthermore bubble clouds can introduce significant nonlinear attenuation.

The lack of an accurate automatic attenuation compensation technique is one of the major hurdles for quantification of tissue perfusion using ultrasound and contrast agents.

US6533728 discusses the use of a harmonic to fundamental ratio (HFR), derived from echo data, to quantify tissue perfusion. However, this approach has various limitations. These include firstly there being a lack of theoretical foundation for the technique which means the ability of this approach to interpret measured quantities remains limited. Secondly, the technique performs analysis of frequency spectra. Accurate analysis of the spectra requires localisation of peaks within the frequency data, which increases the time required for processing. Thirdly, analysis of echo spectra and separation of the fundamental from second harmonic components is performed by curve fitting. To effectively isolate the two components a high frequency resolution is required, which corresponds to a loss of spatial resolution. It can therefore be seen that this approach lacks accuracy and is too time consuming to implement in real time.

The present invention is set out in the claims.

The disclosed method of calibrating ultrasound contrast agent signal data uses signal data which contains attenuation information about the contrast agent signal data by virtue of originating from the same location as the contrast agent signal data but which is less related to the contrast agent than the contrast signal data itself. This gives improved accuracy of attenuation correction compared to known methods. The accurate nature of the attenuation correction enables quantitative measurement of in vivo bubble concentration. Additionally correction in real time is enabled.

Examples of the present invention will now be described with reference to the accompanying drawings, in which: Figure 1 is a diagram of an example ultrasound scanner arranged to scan a medium comprising tissue containing a microbubble contrast agent;

Figure 2 is an example flow diagram showing transmission of time-series ultrasound pulses; Figure 3 is an example flow diagram showing steps performed in calibration of contrast agent signal data;

Figure 4 is a diagram showing ultrasound echo formation;

Figure 5 is a diagram of an experimental setup for imaging a phantom using an example calibration method; Figure 6 is a diagram of an experimental setup for imaging a phantom using an example calibration method;

Figure 7 shows resultant images and intensity profiles for the phantom shown in figure 5;

Figure 8 shows resultant images and intensity profiles for the phantom shown in figure 6;

Figure 9a shows a state of the art pulse inversion image of a liver with metastases; and

Figure 9b shows the image shown in figure 9a after application of an example calibration method.

In overview, as shown in Figure 1, an example ultrasound system comprises an ultrasound scanner 1 arranged to scan a region of interest 2 in a medium which comprises tissue 3 containing contrast agent 4 present in blood flowing through the tissue. As shown in figure 1 the contrast agent 4 comprises microbubbles. These have previously been injected into the blood supply in a known manner. The scanner 1 comprises a probe 5 with a transducer 6 for producing the ultrasonic pulses. Signal processing electronics 7 are present either in the scanner 1 or remote from the scanner 1.

In operation the acquisition of echo signal data occurs as shown in figure 2. A first ultrasound pulse with phase pi and amplitude al is generated by the transducer 6 and transmitted 101 to a location in the medium. The location is determined by focussing as known by the skilled person which also determines the size of the location and hence resolution achievable. The echo signal returned from the pulse is received 102 by the transducer 6 and transformed into electrical echo signal data which is input 110 (see figure 3) to the signal processor 7 for signal processing.

After receipt of the first echo a second pulse with phase p2 and amplitude a2 is transmitted 103 to and received 104 from the same location in the medium and the associated echo signal data is also input 110 to the signal processor 7. The two pulses must differ in at least one of phase, amplitude and duration. That is pi ≠ p2 and/or al ≠ a2. This is to enable data relating to the contrast agent signal to be extracted by combining the echo signal data for the first and second echoes, through a method such as PI. This is discussed in more detail below.

As also shown in figure 2, further pulses may also be used, as discussed below and shown in steps 105, 106 of figure 2.

The raw radio frequency (RF) echo signal data from at least all the multiple echo signals to be used for determining contrast agent signal data are input 110 to the signal processor 7 as shown in figure 3. Contrast agent signal data is determined 120 from echo signal data from multiple pulses as will be described below. An attenuation correction factor is determined 130 from echo signal data from one or more echo signals as will be described below.

The contrast agent signal data is calibrated 140 by using the attenuation correction factor, for example by dividing the contrast signal data by the attenuation correction factor, to obtain calibrated contrast agent data for the location. This calibrated data gives a quantitative measure of the concentration of contrast agent in the medium and therefore of perfusion.

The calibrated contrast agent data may be used, for example, to obtain a concentration value for contrast agent at the location to be output as a quantitative measurement. The calibrated data may also be used to give an intensity value for a pixel of an image of the region of interest 2 to be displayed on a display (not shown). The data for the intensities of the remaining pixels of the image may be obtained in the same way, with each pixel representing a different location and the different locations of the image plane being located for scanning in a known manner.

In order to compensate for attenuation, where the target volume consists of a medium containing contrast agent 4, such as a tissue 3 containing bubbles present in blood flow, tissue echoes may be used to compensate for the attenuation in bubble echoes, since at ultrasound resolution the tissue 3 and microbubbles 4 are at the same location and are therefore exposed to the same forward attenuation. In general, information relating to contrast agent is calibrated by use of information which is related less or not all to contrast agent but which contains information relating to attenuation of the relevant contrast agent signal. This is achieved by calibrating with information related more to the. medium (e.g. consisting of or comprising tissue 3) than contrast agent at a particular location, with the information incorporating attenuation information which is applicable to contrast agent at the same location.

Bubble echoes may be separated from the total echoes to give contrast agent signal data by any suitable technique which combines information from multiple pulses with differing amplitudes and/or phases to extract contrast agent signals on the basis of contrast agent nonlinear behaviour. Contrast agent signal data may be divided by an attenuation correction factor to give the calibration, with the denominator being given by signal data dominated by linear data relating to the tissue 3 medium, with optional filtering and addition of a regulariser as mentioned below. The data from which the attenuation correction factor is obtained from the same set of signals from which the contrast agent signal data is obtained.

A third alternative pulse feature which may be varied between pulses instead of or in addition to pulse amplitude and/or phase is pulse duration (not shown in the example of figure 2)..

A specific example is to input echo signal data relating to two echoes received from the same location in tissue 3 containing microbubbles 4 present in blood. The echoes are from pulses of the same amplitude but inverted phase transmitted and received one after the other, i.e. they are PI signals. Contrast agent signal data is obtained using PI, that is, by adding the echo signal data values from the two pulses. This gives contrast signal data of value E_b of E_b = ECHO(H₁ ) + EChOi-H₁ )

where H₁ is the amplitude. This data value gives contrast agent signal data which is the numerator of a calibration ratio. The denominator is given by signal data value E₁

E, = ECHO(H₁ ) - Echo(-H_λ ) or E_t = 2 x EcHo(H_x ) - E_b oτ E_t = E_b - Ix EcHo(H_x)

which retains most linear information related to tissue 3 at the location and removes most of the nonlinear signal related to bubbles 4. The signal data here such as E_b E_b, Echo(-H) and EcHo[H) may be either RF data or post-processed data.

Calibrated contrast agent signal data is obtained by dividing E_b by E_t and from this result attenuation corrected ultrasound image pixel intensities may be obtained in a known manner.

A more detailed discussion of this example is given below, including discussion of optional filtering and addition of a regulariser included in calculation of the attenuation correction factor. A mathematical explanation of the components of echo signal data is firstly given as background.

As shown in Figure 4, when an ultrasound pulse of amplitude H₁ and frequency ω₀ travels through a region of tissue 3, the amplitude at a point x relative to the transducer is changed to: σ_TAH_x

where σj(x) is the beam profile due to transducer geometry and transmission beam forming

A is forward attenuation which is a product of linear attenuation Aj(x) and nonlinear attenuation A_n(x): A=A](x)-A_n(x) The linear attenuation Aj(x) is a function of spatial distribution of both linear scatterers and bubbles 4 between the transducer and x and the nonlinear attenuation A_n( x) is a function of spatial distribution of bubbles 4 alone and is both pressure and frequency dependent.

The attenuation A(x) is defined as follows:

A(X) = A. (x) - A_n Ot) = exp(- Ja₁(XH a_n (x)dx) (1) o where a.](x) includes the linear attenuation coefficient due to tissue attenuation and linear part of bubble attenuation a_n(x) is the nonlinear attenuation coefficient due to nonlinear bubble attenuation.

The scattering coefficient of the tissue 3 at frequency coo can be denoted by a function b(x). The scattering from bubbles 4 is nonlinear and can be denoted by a polynomial function of the incident pulse. Although an arbitrary order polynomial may be used, for the purpose of illustration and considering the limited transducer bandwidth a 2^nd order polynomial is used in this example.

The scattering can then be denoted as:

Scat =

+ c(x) ^■ (Jt₁H₁ ^' + Ic₂H₁ ² )

=bσ_τA_λ A_nH₁ + c(x)(k_λσ_τA_x A_nH \ + k₂σ_τ A₁ ² A²H₁ ²) (2)

where c(x) is the concentration of bubbles H] ' is the amplitude of the attenuated US pulse kj and k₂ denote the first and second order coefficients of the polynomial and are assumed constant.

While the first term in the bracket in Equation (1) is still of the fundamental frequency ω₀, the second term in the bracket corresponds to the 2^nd harmonic with frequency of 2ω₀. For the sake of brevity in the above and subsequent equations the depth dependence is no longer explicitly stated.

The scattered signals are attenuated when they travel back to the ultrasound transducer. Since the amplitude of the scattered signals is very small compared to the transmission pulse, the bubbles 4 will oscillate within the linear regime and therefore only linear attenuation A₁ of the back-scattered pulse is considered. The attenuation is frequency dependent. For the fundamental frequency the attenuation is still A₁ but for the 2^nd harmonic the attenuation is different and is denoted as A₂. Therefore the echo of the pulse can finally be written as:

EChO(H₁ ) = bGσ_τσ_RA_ι ²A_nH_ι + ck_xGσ_Tσ_RA_x ² A_nH_x + ck₂Gσ²σ _RA_X ² AIH_X ² A₂ (3)

where G denotes the gain in scanner post processing including any default TGC after the echoes are received σ_R denotes the receive sensitivity profile due to receive focusing.

It can be seen that the echo signal given in Equation (3) is affected by many confounding factors. A more detailed mathematical explanation of the example discussed above of calibrating contrast agent signals obtained using PI is given below.

The PI images are represented by the equation:

E_b = Echo(H_x ) + EChOi-H₁ ) = 2ck₂Gσ^.σ_R A₁ ² A_n ²H₁ ²A₂ (4)

which gives the bubble signal E^. In this case this is the second harmonic of the echoes. As can be seen from this equation, the nonlinear echo generated by PI is proportional to bubble concentration c, but is at the same time influenced by a number of quantities such as attenuation A₁ and A₂ and the beam profiles. The disclosed method reduces these confounding factors and extracts quantities that are proportional to bubble concentration.

The tissue signal E_t may be extracted from the reflected signal by using the opposite operation to that demonstrated in eq. (4). I.e. in this case subtraction is used instead of addition to extract the tissue signal:

E₁ = EChO(Hy) - EChOi-H₁ ) = 2bGσ_τσ_RA] A_nH₁ + 2ck_λGσ_Tσ_R AfA_nH₁ (5)

It can be seen that by performing this subtraction the nonlinear term generated by the bubbles 4 is suppressed and the quantity is effectively the echo at the fundamental frequency. Although the linear term of bubble scattering still exists, the bubble scattering is overall reduced and linear scattering by tissue 3 is doubled.

Dividing E_b by E_t, gives: E_b _ EcMHi ) + EChOJ-H₁) 2ck₂Gσlσ_RAl A_n H₁ ² A₂ ck₂σ_τH_xA_nA₂

E, ^~ Echo(H_x) - Echo(-H_λ ) ^~ 2bGσ_τσ_RA*A_nH_x + 2c)t₁Gσ₇.σ_ΛA₁ ²A_nH₁ ^~ b + ck_x

(6)

It can be seen from equation (6) that through this operation a number of confounding parameters in the numerator are cancelled out by those in the denominator and a much simplified expression is obtained.

Depending on the relative significance of the two terms in the denominator, Equation (6) can be further simplified. If the target is a region of tissue 3 containing bubbles 4 in micro vessels where a low bubble concentration can be assumed, the 2^nd term in Equation (5) and the denominator term c.kj in Equation (6) can be ignored. Equation (6) becomes:

E^ _{= c} k₂σ_τH_xA_nA₂ E, b

which presents a quantity that is proportional to bubble concentration c, but still affected by tissue scattering coefficient b, beam profile (σ), nonlinear attenuation A_n and backward attenuation for the 2^nd harmonic A₂. For σ versus

A₂ and A_n, they are actually varying in the opposite direction; σ increases with depth until the focus while A₂ and A_n decrease with depth. Therefore they oppose one another to some degree in regions before the focus. If there is no significant bubble cloud between transducer and the target, the nonlinear attenuation will be insignificant and the term A_n can just be treated as unity.

If the imaging plane consists of homogeneous tissue 3 the filtered denominator may be approximated as a constant. This may often be satisfied when scanning e.g. part of a liver. Variation of b in small spatial scale may be removed by the filter as discussed below. In order to reduce the variation of the denominator E_t due to tissue scattering, some prior information about attenuation may be used to modify the denominator. As can be seen in Equation (1), the attenuations to be corrected are effectively integrals of the attenuation coefficients along the transmission path. Therefore attenuation can be deemed as a smooth signal with gradual spatial variation. The scattering on the other hand, shows variations over relatively small spatial scales.

In order to reduce this scattering related spatial variation in E₁, a filter may be applied to E_t, in order to further increase the accuracy of the attenuation correction.

In the following, a 2D spatial median filter is used as an example. The advantage of a median filter is that rather than averaging the signal, it removes any large magnitude spatial variation in the signal with no affect on the rest of the signal. The size of the median filter kernel determines the spatial size of the variations that can be removed.

By choice of the size of the median filter to be much smaller than the variation in σ, A and G, Equation (6) can be rewritten as:

MedianFilt(E_t)

Ck₂(T₇H_xA_nA₂

MedianFiltφ + ck_x) If c.kj in the denominator is neglected, then median filtering E₁ removes the local variations in b and the remaining quantity represents a truer approximation to the bubble concentration c.

Alternatively if the target is dominated by a bubble cloud, e.g. in heart chambers or larger vessels, the term b in equation (8) can be ignored and the equation becomes:

MedianFilt(E_t )

Since the filtered concentration MedianFilt(c(x)) in the denominator acts to normalise c(x) in the numerator, the ratio is now a quantity that is not related to local bubble concentration but still affected by beam profile σ and backward attenuation for the 2^nd harmonic A₂ and nonlinear attenuation A_n.

If the target consists of neither bubbles 4 nor tissue 3, the denominator tends to zero and the quantity will tend to infinity. To overcome this instability a constant may be added to the denominator as a regulariser. The amplitude of the regulariser may be determined from the system noise level. In this case attenuation corrected contrast images may be obtained by the following formula:

F₁ EChO(H₁) + EChOj-H₁)

MedianFilt(E_t ) + regulariser MedianFilt(Echo(H \ ) — Echo(—H_{ )) + regularizer

(10)

Since in the above ratio formula the numerator is a second order term and the denominator is a first order term, an alternative ratio may be obtained by using the squared first order term as denominator. If the tissue echo is dominant at the target and the bubble term in the denominator can be ignored, then:

E_b Ck₂A₂

MedianFilt{E² )

MedianFilt(2b²Gσ_RA² )

(H)

By assuming that the attenuation is mainly caused by tissue 3 and as the tissue 3attenuation is doubled when frequency is doubled, A₂ is actually equal to A₁ squared and the above formula may be further reduced to:

F rk

" - ² (12)

MedianFilt(E² ) MedianFilt(2b²Gσ_R )

It can be seen that the quantity is still affected by the scanner TGC G and receive sensitivity profile σ_R. Again a regulariser may be added to the denominator to stabilise the formula.

If the imaging plan contains multiple types of tissues of significant sizes, b will vary accordingly and the quantification process may be adapted to take account of this.

Figures 5 and 6 show example experimental setups used to demonstrate the results obtainable with the disclosed method.

Figures 5 and 6 show phantoms consisting of tissue mimicking materials (TMM) and SonoVue™ microbubbles. The phantom shown in Figure 5 consists of blocks of TMM 7 which are submerged in a suspension of microbubbles 4. The TMM 7 is made of gelatine and does not allow bubbles 4 to perfuse into it. The suspension was gently stirred so that the bubbles 4 were well mixed.

For the phantom shown in figure 6, a piece of sponge was cut into small pieces 8(typical dimensions of 2 mm in each dimension) and put into a beaker of water. The sponge bits and water were then degassed before adding SonoVue™ bubbles 4. The suspension was again gently stirred so that the bubbles 4 were well mixed. This phantom was designed to create a model of uniformly perfused tissue. Finally one thin slice of cardboard 9 was inserted into this phantom to create some artificial attenuation in the image. The concentration of the microbubble suspension in each of the phantoms was 90μUL.

Figures 7 and 8 show resulting images from the phantoms shown in figures 5 and 6 respectively scanned using PI imaging and image intensity on a single vertical line in each image. The image intensity profiles from Figures 7c) and 8c) are shown in Figures 7d) and 8d) respectively as broken lines for comparison.

Four types of image are shown in each of figures 7 and 8: a) A standard B- mode image generated with echoes of one of the PI pulse pair; b) A PI image from both echoes of the pulse pair; c) An attenuation corrected image via Equation (10); and d) An attenuation corrected image generated using Equation (11). Both the B-mode and PI images used the default Time Gain Control of the respective scanner for attenuation correction. A median filter was used with an axial size of 8mm and a lateral size between 0.6mm and 6mm. The variation of the filter size in lateral direction is because the probe is a fan-beam phased array probe whose line density varies with depth.

Figure 7 shows that the scanner default TGC overcompensated the image distal to the probe and the intensity gradually increases towards the bottom of the image in both Figures 7a) and b). This overcompensation can also be seen in the intensity profiles for Figures 7 a) and b).

For the corrected image in Figure 7c) and d) using Equations (10) and (11) respectively, the overcompensation has been largely corrected and the image is clearer, which can also be seen in the image intensity profiles. Figure 8 shows that the cardboard in the upper middle part of the phantom created attenuation shadows in the lower part of the B-mode image (Figure 8a)) and PI image (Figure 8b)). This attenuation can also be seen in the intensity profile for Figures 8a) and b), and is seen as a gradual decay in the signals.

For the corrected image in Figure 8c) using Equation (10), the attenuation has been largely corrected and this can also be seen in the image intensity profile for Figure 8c). The images after correction better present the whole region with more homogeneous intensity. Figure 8d) and c) show the results using Equation (11), which show good correction in the middle part of the image but some artefacts on both sides.

Figures 9 a and 9b show image results obtained from liver metastases in a female patient scanned using an ultrasound scanner. The scan frequency was 2MHz and Mechanical Index was 0.1. Raw RF data were acquired approximately 40s after a bolus injection of Sonovue™ contrast agents. Pulse Inversion images with (Figure 9b) and without (Figure 9a) calibration to correct for attentuation were constructed using the RF data. Two different Regions of Interest, ROIl and ROI 2, were defined within the liver, shown in Figure 9a as dashed circles, representing two pieces of normal liver parenchyma where microbubble concentration should be similar. The image intensities of these ROIs before and after calibration were calculated and compared. The improvement by the calibration was quantified.

It can be seen from Figure 9a that the tumour boundary was missing at two places due to attenuation in the traditional Pulse Inversion image. The missing parts of the tumour boundary were successfully recovered, as shown in Figure 9b, after calibration. In Figure 9a the two ROIs, both representing normal liver parenchyma, have very different image intensity due to attenuation; the averaged intensity of ROI2 is only 53% of that for ROIl. In Figure 9b the intensity of ROI2 is 102% of ROIl after calibration, suggesting that the calibration in this case is successful.

Significant advantages of above-described method are that as well as the attenuation correction giving more accurate quantification of contrast agent and improved images, it makes use of the same data required to extract the bubble echoes so no additional acquisition is required and it only requires simple arithmetic combination and can be done in real time.

Although in the above discussion the example of PI imaging are used for separating a bubble echo from the total echo, and corresponding operations for acquiring the tissue signal for attenuation correction, any technique which uses multiple pulses with differing phases and/or amplitudes which enable extraction of the contrast agent information by combining the two differing signals in an appropriate manner may be used to obtain contrast agent signal data. For example, for AM, where, for example, H₁ and

pulses may be fired

Echo(H_x ) - 2 * may be used for extracting bubble echoes, that is,

contrast agent signal data, and the opposite operation of

TJ

EChO(H₁ ) + 2 * Echo( — ^L) may be used for extracting tissue echoes (retains most

linear signal relating to tissue) to determine an attenuation correction factor (which may also include filtering and/or a regulariser).

For Pulse Inversion Amplitude Modulation (PIAM) where, for example Hi and

— ^L pulses may be fired

Echo(H_x) + 2 * may be used for extracting bubble echoes and the

opposite operation of

TJ

EcHo(H_x ) - 2 * Echo( ^L) may be used for extracting tissue echoes.

Although these examples and the PI example discussed in detail above use echo data from two echoes to determine both the attenuation factor and the contrast signal data, it is not necessary to use data relating to the same number of pulses for each of these and it is not necessary to use multiple pulses from the same location in determining the attenuation correction factor. To determine the contrast agent signal data on the other hand, it is necessary to use multiple pulses as it is the combination of the pulse data which extracts contrast agent signal data based on identifying nonlinearities. To determine the attenuation correction factor it is necessary only to use information which is related less or not all to contrast agent compared to the medium, by being dominated by linear information relating to the medium rather than nonlinear information relating to the contrast agent at the particular location but which contains information relating to attenuation of the relevant contrast agent signal due to originating from the same location within the medium containing contrast agent.

Other examples for determining the attenuation correction factor include using echo signal data relating to a single echo signal as the attenuation factor. This is the data which would be used to produce a conventional B-mode image. Since echoes relating to contrast agents 4 within tissue 3 compared to echoes from the tissue 3 itself may be, in general, relatively small, echo signal data relating to a single echo is less related to contrast agents 4 than the contrast agent signal data from multiple differing time-series pulses itself but contains the same attenuation information.

A further example is to use data relating to a single echo as in the above example, but to filter out harmonics relating to contrast agent from the data. In this case, the attenuation factor consists of data relating to the remaining fundamental signal, which is more related to tissue 3 than contrast agent 4 and contains the same attenuation information.

The calibration of the contrast agent signal data with the attenuation correction factor may be performed by dividing a contrast agent signal data value by the attenuation correction factor or in any other suitable manner for calibration. More than two ultrasound pulses may be transmitted to and received from the same location in a time-series manner, as shown in figure 2, with each pulse differing in at least one of phase, amplitude and pulse length. The more such differing pulses that are used in each of the contrast agent data and attenuation correction factor determination, the more accurate the resulting calibrated contrast signal data that may be obtained from combining the echo signal data from the different pulses, but the longer the acquisition time. The number of pulses used is therefore a balance between the accuracy and acquisition time required.

An example of a suitable pulse sequence consisting of three pulses transmitted in a time series manner is a 'SIEMENS' CPS' pulse sequence where three pulses may be transmitted with two pulses Pl and P2 having inversed phase and half the amplitude of the third pulse P3. i..e. P1=P2= -1/2(P3). The bubble signal data may then be given by Echo(Pl) + Echo(P2) + Echo(P3) and the tissue signal given by Echo(Pl) + Echo(P2) - Echo(P3).

The number of pulses transmitted for a location, for example to obtain a image pixel intensity, may be altered 150 (see figure 3) on a real time basis, for example by an operator monitoring a resultant image on a display. In this case, calibrated contrast agent data is obtained from a first plurality of echo pulses and then further calibrated contrast agent data is obtained from a second plurality of echo pulses having a different number of pulses, with the change in the number of pulses being determined based on feedback relating to the calibrated contrast agent. For example, if image quality is not sufficient, the number of pulses from which echo pulse data is input may be increased. The echo signal data from echoes from the first, second and any further time- series pulses maybe input to the signal processor 7 after receipt of the relevant echo by the transducer or alternatively echo signal data relating to all of the echoes may be input to the signal processor in any suitable manner, for example in one transmission.

As discussed above, a filter may optionally be used in obtaining the attenuation correction factor. Any suitable filter known to the skilled person which removes or reduces variations due to scattering (for example caused by ultrasound speckles or by changes in tissue type) which could affect the resultant attenuation correction factor may be used. A 2D median spatial filter is discussed above. This is useful because attenuation should be smooth along both x and y directions. Particular window sizes filter out radical changes due to changes in tissue type whilst retaining the underlying beam and attenuation profile may be used. Other examples of suitable filters are spatial linear averaging.

The optional regulariser which may be used in a manner known to the skilled person to avoid dividing by zero may be, for example, a constant or spatially variant according to the spatial noise distribution during collection

To determine the attenuation correction factor, the echo signal data may be post-processed echo signal data rather than RF data. Post-processing of the echo signal data may include receive gain, demodulation (i.e. envelop detection), adjustment of dynamic range and log compression. After all the post-processing the echo signal data may become the B-mode image(s). As the echo signal data includes post-processed data, further examples of generating the attenuation correction factor may use either envelop detected echo signal data for a single transmission pulse, or a combination of envelop detected echo signal data for multiple pulses.

Another example of generating the attenuation correction factors may use either the B-mode image based on a single transmission pulse, or a combination of B-mode images based on multiple pulses.

As mentioned above, the disclosed method may be used in combination with TGC and if this is the case this may be in any suitable known manner.

Any suitable known ultrasound scanner 1 may be used to produce the raw data. Examples of known suitable scanners are the AN2300 digital ultrasound engine (Analogic Corporation, Peabody, MA, USA) and the UltrasonixRP system (Ultrasonix, Richmond, BC, Canada).

Any suitable known bubble contrast agent 4 and method of introducing the contrast agent 4 to the region of interest may be used.

Claims

1. A method of calibrating ultrasound contrast agent signal data from a location in a medium containing a contrast agent, the method comprising the steps of: inputting first echo signal data from a first plurality of echo signals from time series ultrasound pulses, the ultrasound pulses differing in at least one of phase, amplitude and pulse duration; determining the contrast agent signal data from the first echo signal data from one or more of the echo signals; determining an attenuation correction factor from the first echo signal data from at least one of the echo signals; calibrating the contrast agent signal data with the attenuation correction factor to obtain calibrated contrast agent data for the location; outputting the calibrated contrast agent data.

2. A method of determining concentration of an ultrasound contrast agent from contrast agent signal data from a location in a medium containing the contrast agent, the method comprising the steps of: calibrating the contrast signal data with a method according to claim 1 to obtain calibrated contrast agent data for the location; and determining the concentration of the contrast agent at the location from the calibrated contrast data.

3. A method according to any preceding claim, further comprising the steps of: transmitting a first ultrasound pulse into the medium containing the contrast agent to generate a first echo from the location in the medium; receiving first echo signal data from the first echo; transmitting a second ultrasound pulse into the medium containing the contrast agent to generate a second echo from the location in the medium; receiving second echo signal data from the second echo; wherein the first and second pulses differ in at least one of phase, amplitude and duration.

4. A method according to claim 3, further comprising the step of introducing the contrast agent to the location.

5. A method according to any preceding claim, wherein the contrast agent comprises a plurality of microbubbles.

6. A method according to any preceding claim, wherein the calibrating step comprises determining a ratio of the contrast signal data to the attenuation correction factor.

7. A method according to any preceding claim, wherein the attenuation correction factor is the first echo signal data, either post-processed or not, from only one echo signal.

8. A method according to any of claims 1 to 6 wherein the step of determining the contrast agent signal data comprises adding a multiple of first echo signal data from a first echo signal to a multiple of first echo signal data from a second echo signal of opposite phase.

9. A method according to claim 8, wherein the step of determining the attenuation correction factor comprises subtracting the multiple of first echo signal data from the first echo signal from the multiple of first echo signal data from the second echo signal of the opposite phase.

10. A method according to any of claims 1 to 6, wherein the step of determining the contrast agent signal data comprises subtracting a multiple of first echo signal data from a first echo signal from a multiple of first echo signal data from a second echo signal of the same phase.

11. A method according to claim 10, wherein the step of determining the attenuation correction factor comprises adding the multiple of first echo signal data from the first echo signal to the multiple of first echo signal data from the second echo signal of the same phase.

12. A method according to any of claims 7, 9 or 11, wherein the step of determining the attenuation correction factor further comprises applying a filter.

13. A method according to claim 12, wherein the filtering comprises median filtering, spatial averaging or other low pass filtering.

14. A method according to any of claims 9, 11, 12, and 13, wherein the step of determining the attenuation correction factor further comprises adding a regulariser.

15. A method according to any of claims 1 to 6 and 8 to 14, wherein the contrast agent data and attenuation correction factor are each determined from first echo signal data from three echo signals.

16. A method according to any preceding claim, further comprising obtaining further calibrated contrast agent data for the location from second echo signal data from a second plurality of echo signals, wherein the number of echo signals in the second plurality is different to the number of signals in the first plurality and is determined based on feedback relating to the calibrated contrast agent data.

17. A method of determining an image pixel intensity for an ultrasound contrast agent image, the method comprising the steps of: calibrating contrast signal data with a method according to any preceding claim to obtain calibrated contrast agent data for a location in a medium containing the contrast agent; determining the first image pixel intensity from the calibrated contrast agent data; and outputting the image pixel intensity.

18. A method according to claim 17, further comprising the step of generating a first image pixel on a display having the image pixel intensity.

19. An ultrasound imaging method comprising the step of generating a plurality of image pixels, wherein each image pixel is generated with a method according to claim 4 with calibrated contrast agent data for a different location in the medium.

20. A system for calibrating ultrasound contrast agent signal data from a location in a medium containing a contrast agent, the system comprising: an input arranged to input first echo signal data from a first plurality of echo signals from time series ultrasound pulses, the ultrasound pulses differing in at least one of phase, amplitude and pulse duration; contrast agent signal data determining means arranged to determine the contrast agent signal data from the first echo signal data from one or more of the echo signals; attenuation correction factor determining means arranged to determine an attenuation correction factor from the first echo signal data from at least one of the echo signals; calibrating means arranged to calibrate the contrast agent signal data with the attenuation correction factor to obtain calibrated contrast agent data for the location; an output arranged to output the calibrated contrast agent data.

21. A system for determining concentration of an ultrasound contrast agent from contrast agent signal data from a location in a medium containing the contrast agent, the system comprising: a calibrating system according to claim 20 arranged to obtain calibrated contrast agent data for the location; and contrast agent concentration determining means arranged to determine the concentration of the contrast agent at the location from the calibrated contrast data.

22. A system according to claim 20 Or 21, wherein the contrast agent comprises a plurality of microbubbles.

23. A system according to any of claims 20 to 22, wherein the calibrating means is arranged to determining a ratio of the contrast signal data to the attenuation correction factor.

24. A system according to any of claims 20 to 23, wherein the attenuation correction factor is the first echo signal data from only one echo signal.

25. A system according to any of claims 20 to 23, wherein the contrast agent signal data determining means is arranged to add a multiple of first echo signal data from a first echo signal to a multiple of first echo signal data from a second echo signal of opposite phase.

26. A system according to claim 25, wherein the attenuation correction factor determining means is arranged to subtract the multiple of first echo signal data from the first echo signal from the multiple of first echo signal data from the second echo signal of the opposite phase.

27. A system according to any of claims 20 to 23, wherein the contrast agent signal data determining means is arranged to subtract a multiple of first echo signal data from a first echo signal from a multiple of first echo signal data from a second echo signal of the same phase.

28. A system according to claim 27, wherein the attenuation correction factor determining means is arranged to add the multiple of first echo signal data from the first echo signal to the multiple of first echo signal data from the second echo signal of the same phase.

29. A system according to any of claims 24, 26 and 28, wherein the attenuation correction factor determining means is further arranged to apply a filter.

30. A system according to claim 29, wherein the filter comprises a median filter or a spatial averaging filter.

31. A system according to any of claims 26, 28, 29 and 30, wherein the attenuation correction factor determining means is further arranged to add a regulariser.

32. A system according to any of claims 20 to 23 and 25 to 31, wherein the system is arranged to determine each of the contrast agent data and attenuation correction factor from first echo signal data from three echo signals.

33. A system according to any of claims 20 to 32, further comprising a feedback means arranged to obtain feedback relating to the calibrated first contrast agent data; wherein the system is arranged to obtain further calibrated contrast agent data for the location from second echo signal data from a second plurality of echo signals, wherein the number of echo signals in the second plurality is different to the number of signals in the first plurality and is determined based on feedback relating to the calibrated contrast agent data.

34. A system for determining an image pixel intensity for an ultrasound contrast agent image, the system comprising: an ultrasound calibrating system according to any of claims 20 to 33; pixel intensity determining means arranged to determine the image pixel intensity from the calibrated contrast agent data; and an out put arranged to outputting the image pixel intensity.

35. A system according to claim 34, further comprising a pixel generator arranged to generate an image pixel on a display having the first image pixel intensity.

36. An ultrasound imaging system comprising: a pixel generator according to claim 35, wherein the pixel generator is further arranged to generate a plurality of image pixels with calibrated contrast agent data for a different location in the medium; a display arranged to display the generated image pixels.

37. A method or system substantially as described herein with reference to the accompanying drawings.