Ultrasonic diagnostic imaging with blended tissue harmonic signals

0001This is a divisional application of U.S. patent application Ser. No. 08/943,546, filed Oct. 3, 1997 and entitled “ULTRASONIC DIAGNOSTIC IMAGING OF RESPONSE FREQUENCY DIFFERING FROM TRANSMIT FREQUENCY” now U.S. Pat. No. 5,879,303, which is a continuation-in-part of U.S. patent application Ser. No. 08/723,483, filed Sep. 27, 1996 and entitled “ULTRASONIC DIAGNOSTIC IMAGING WITH CONTRAST AGENTS” now U.S. Pat. No. 5,833,613, and U.S. patent application Ser. No. 08/943,546 also claims the benefit of U.S. Provisional Application No. 60/032,771, filed Nov. 26, 1996.

0002This invention relates to ultrasonic diagnosis and imaging of the body and, in particular, to new methods and apparatus for ultrasonically imaging with a response frequency which differs from the transmitted frequency.

0003Ultrasonic diagnostic imaging systems have been used to image the body with the enhancement of ultrasonic contrast agents. Contrast agents are substances which are biocompatible and exhibit uniquely chosen acoustic properties which return readily identifiable echo signals in response to insonification. Contrast agents can have several properties which enables them to enhance an ultrasonic image. One is the nonlinear characteristics of many contrast agents. Agents have been produced which, when insonified by an ultrasonic wave at one frequency, will exhibit resonance modes which return energy at other frequencies, in particular, harmonic frequencies. A harmonic contrast agent, when insonified at a fundamental frequency, will return echoes at the second, third, fourth, and higher harmonics of that frequency.

0004It has been known for some time that tissue and fluids also have inherent nonlinear properties. Tissue and fluids will, even in the absence of a contrast agent, develop and return their own non-fundamental frequency echo response signals, including signals at harmonics of the fundamental. Muir and Carstensen explored these properties of water beginning in 1980, and Starritt et al. looked at these properties in human calf muscle and excised bovine liver.

0005While these non-fundamental frequency echo components of tissue and fluids are generally not as great in amplitude as the harmonic components returned by harmonic contrast agents, they do exhibit a number of characteristics which may be advantageously used in ultrasonic imaging. One of us (M. Averkiou) has done extensive research into these properties in work described in his doctoral dissertation. In this exposition and other research, the present inventors have seen that the main lobe of a harmonic beam is narrower than that of its fundamental, which they have found has implications for clutter reduction when imaging through narrow orifices such as the ribs. They have seen that the sidelobe levels of a harmonic beam are lower than the corresponding sidelobe levels of the fundamental beam, which they have found has implications for off-axis clutter reduction. They have also seen that harmonic returns from the near field are also relatively less than returning energy at the fundamental frequency, which they have found has implications for near field clutter rejection. As will be seen, these properties may be exploited in the methods and constructed embodiments of the present invention.

0006In accordance with the principles of the present invention, an ultrasonic imaging system and method are provided for imaging tissue and fluids from response frequencies which differ from the transmitted frequency, in particular echoes returned from the tissue or fluids at a harmonic of a transmitted fundamental frequency. The imaging system comprises a means for transmitting an ultrasonic wave at a fundamental frequency, means for receiving echoes at a harmonic frequency, and an image processor for producing an ultrasonic image from the harmonic frequency echoes.

0007In a preferred embodiment of the present invention the transmitting and receiving means comprise a single ultrasonic probe. In accordance with a further aspect of the present invention, the probe utilizes a broadband ultrasonic transducer for both transmission and reception.

0008In accordance with yet another aspect of the present invention, partially decorrelated components of received harmonic echoes are produced and utilized to remove artifacts from the harmonic image, providing clearly defined images of tissue boundaries such as that of the endocardium. In a preferred embodiment the partially decorrelated components are produced by processing the harmonic echoes through different passbands.

0009The methods of the present invention include the use of harmonic echoes to reduce near-field or multipath clutter in an ultrasonic image, such as that produced when imaging through a narrow acoustic window such as the ribs. In accordance with yet a further aspect of the present invention, harmonic and fundamental echoes are blended in a common image to reduce clutter, image at appreciable depths, and overcome the effects of depth-dependent attenuation.

0010In the drawings:

0011<figref idref="DRAWINGS">FIG. 1</figref> illustrates in block diagram form an ultrasonic diagnostic imaging system constructed in accordance with the principles of the present invention;

0012<figref idref="DRAWINGS">FIGS. 2</figref>, <b>3</b>, <b>4</b>, and <b>5</b> illustrate certain properties of harmonic echoes which may be advantageously applied to ultrasonic imaging applications; and

0013<figref idref="DRAWINGS">FIGS. 6 and 7</figref> illustrate passband characteristics used to explain the performance of the embodiment of <figref idref="DRAWINGS">FIG. 1</figref>;

0014<figref idref="DRAWINGS">FIG. 8</figref> illustrates typical fundamental and harmonic frequency passbands of an embodiment of the present invention;

0015<figref idref="DRAWINGS">FIG. 9</figref> illustrates an FIR filter structure suitable for use in the embodiment of <figref idref="DRAWINGS">FIG. 1</figref>;

0016<figref idref="DRAWINGS">FIG. 10</figref> illustrates in block diagram form a portion of a preferred embodiment of the present invention;

0017<figref idref="DRAWINGS">FIG. 11</figref> illustrates the operation of the normalization stages of the embodiment of <figref idref="DRAWINGS">FIG. 10</figref>;

0018<figref idref="DRAWINGS">FIG. 12</figref> is a block diagram of one of the multiplier accumulators used in the filters of the embodiment of <figref idref="DRAWINGS">FIG. 10</figref>;

0019<figref idref="DRAWINGS">FIG. 13</figref> illustrates typical fundamental and harmonic frequency passbands of the embodiment of <figref idref="DRAWINGS">FIG. 10</figref>;

0020<figref idref="DRAWINGS">FIG. 14</figref> illustrates the blending of fundamental and harmonic signal components into one ultrasonic image; and

0021<figref idref="DRAWINGS">FIG. 15</figref> illustrates the passbands of a time varying filter used in the formation of blended images.

0022Referring first to <figref idref="DRAWINGS">FIG. 1</figref>, an ultrasonic diagnostic imaging system constructed in accordance with the principles of the present invention is shown in block diagram form. A central controller <b>120</b> commands a transmit frequency control <b>117</b> to transmit a desired transmit frequency band. The parameters of the transmit frequency band, f_tr, are coupled to the transmit frequency control <b>117</b>, which causes the transducer <b>112</b> of ultrasonic probe <b>110</b> to transmit ultrasonic waves in the fundamental frequency band. In a constructed embodiment a band of frequencies located about a central frequency of 1.67 MHz is transmitted. This is lower than conventional transmitted imaging frequencies, which generally range from 2.5 MHz and above. However, use of a typical transmit frequency of 3 or 5 MHz will produce harmonics at 6 and 10 MHz. Since higher frequencies are more greatly attenuated by passage through the body than lower frequencies, these higher frequency harmonics will experience significant attenuation as they return to the probe. This reduces the depth of penetration and image quality at greater imaging depths, although the harmonic signals, created as they are during the propagation of the transmitted wave through tissue, do not experience the attenuation of a full round trip from the transducer as the fundamental signals do. To overcome this problem, the central transmit frequency in the illustrated embodiment is below 5 MHz, and preferably below 2.5 MHz, thereby producing lower frequency harmonics that are less susceptible to depth dependent attenuation and enabling harmonic imaging at greater depths. A transmitted fundamental frequency of 1.67 MHz will produce second harmonic return signals at 3.34 MHz in the illustrated embodiment. It will be understood, of course, that any ultrasonic frequency may be used, with due consideration of the desired depth of penetration and the sensitivity of the transducer and ultrasound system.

0023The array transducer <b>112</b> of the probe <b>110</b> transmits ultrasonic energy and receives echoes returned in response to this transmission. The response characteristic of the transducer can exhibit two passbands, one around the fundamental transmit frequency and another about a harmonic frequency in the received passband. For harmonic imaging, a broadband transducer having a passband encompassing both the transmitted fundamental and received harmonic passbands is preferred. The transducer may be manufactured and tuned to exhibit a response characteristic as shown in <figref idref="DRAWINGS">FIG. 6</figref>, in which the lower hump <b>60</b> of the response characteristic is centered about the transmitted fundamental frequency f_r, and the upper hump <b>62</b> is centered about the received harmonic frequency f_rof the response passband. The transducer response characteristic of <figref idref="DRAWINGS">FIG. 7</figref> is preferred, however, as the single dominant characteristic <b>64</b> allows the probe to be suitable for both harmonic imaging and conventional broadband imaging. The characteristic <b>64</b> encompasses the transmitted fundamental frequency f_g, and also the harmonic receive passband bounded between frequencies f_Land f_cand centered about frequency f_r. As discussed above, a low fundamental transmit frequency of 1.67 MHz will result in harmonic returning echo signals at a frequency of 3.34 MHz. A response characteristic <b>64</b> of approximately 2 MHz would be suitable for these fundamental and harmonic frequencies.

0024Tissue and cells in the body alter the transmitted fundamental frequency signals during propagation and the returned echoes contain harmonic components of the originally transmitted fundamental frequency. In <figref idref="DRAWINGS">FIG. 1</figref> these echoes are received by the transducer array <b>112</b>, coupled through the T/R switch <b>114</b> and digitized by analog to digital converters <b>115</b>. The sampling frequency f_sof the A/D converters <b>115</b> is controlled by the central controller. The desired sampling rate dictated by sampling theory is at least twice the highest frequency f_cof the received passband and, for the preceding exemplary frequencies, might be on the order of at least 8 MHz. Sampling rates higher than the minimum requirement are also desirable.

0025The echo signal samples from the individual transducer elements are delayed and summed by a beamformer <b>116</b> to form coherent echo signals. The digital coherent echo signals are then filtered by a digital filter <b>118</b>. In this embodiment, the transmit frequency f_tis not tied to the receiver, and hence the receiver is free to receive a band of frequencies which is different from the transmitted band. The digital filter <b>118</b> bandpass filters the signals in the passband bounded by frequencies f_Land f_cin <figref idref="DRAWINGS">FIG. 7</figref>, and can also shift the frequency band to a lower or baseband frequency range. The digital filter could be a filter with a 1 MHz passband and a center frequency of 3.34 MHz in the above example. A preferred digital filter is a series of multipliers <b>70</b>-<b>73</b> and accumulators <b>80</b>-<b>83</b> as shown in <figref idref="DRAWINGS">FIG. 9</figref>. This arrangement is controlled by the central controller <b>120</b>, which provides multiplier weights and decimation control which control the characteristics of the digital filter. Preferably the arrangement is controlled to operate as a finite impulse response (FIR) filter, and performs both filtering and decimation. For example, only the first stage output <b>1</b> could be controlled to operate as a four tap FIR filter with a 4:1 decimation rate. Temporally discrete echo samples S are applied to the multiplier <b>70</b> of the first stage. As the samples S are applied, they are multiplied by weights provided by the central controller <b>120</b>. Each of these products is stored in the accumulator <b>80</b> until four such products have been accumulated (added). An output signal is then produced at the first stage output <b>1</b>. The output signal has been filtered by a four tap FIR filter since the accumulated total comprises four weighted samples. Since the time of four samples is required to accumulate the output signal, a 4:1 decimation rate is achieved. One output signal is produced for every four input samples. The accumulator is cleared and the process repeats. It is seen that the higher the decimation rate (the longer the interval between output signals), the greater can be the effective tap number of the filter.

0026Alternatively, temporally separate samples are delayed by delay elements τ and applied to the four multipliers <b>70</b><b>73</b>, multiplied, and accumulated in the accumulators <b>80</b>-<b>83</b>. After each accumulator has accumulated two products, the four output signals are combined as a single output signal. This means that the filter is operating as an eight tap filter with a 2:1 decimation rate. With no decimation, the arrangement can be operated as a four tap FIR filter. The filter can also be operated by applying echo signals to all multipliers simultaneously and selectively time sequencing the weighting coefficients. A whole range of filter characteristics are possible through programming of the weighting and decimation rates of the filter, under control of the central controller. The use of a digital filter provides the advantage of being quickly and easily changed to provide a different filter characteristic. A digital filter can be programmed to pass received fundamental frequencies at one moment, and harmonic frequencies at the next. The digital filter can thus be operated to alternately produce images or lines of fundamental and harmonic digital signals, or lines of different alternating harmonics in a time-interleaved sequence simply by changing the filter coefficients during signal processing.

0027Returning to <figref idref="DRAWINGS">FIG. 1</figref>, to image just a non-fundamental frequency, the digital filter <b>118</b> is controlled by the central controller <b>120</b> to pass echo signals at a harmonic frequency for processing, to the exclusion of the fundamental frequency. The harmonic echo signals from the tissue are detected and processed by either a B mode processor <b>37</b> or a contrast signal detector <b>128</b> for display as a two dimensional ultrasonic image on the display <b>50</b>.

0028The filtered echo signals from the digital filter <b>118</b> are also coupled to a Doppler processor <b>130</b> for conventional Doppler processing to produce velocity and power Doppler signals. The outputs of these processors are coupled to a 3D image rendering processor <b>162</b> for the rendering of three dimensional images, which are stored in a 3D image memory <b>164</b>. Three dimensional rendering may be performed as described in U.S. Pat. No. 5,720,291, and in U.S. Pat. Nos. 5,474,073 and 5,485,842, the latter two patents illustrating three dimensional power Doppler ultrasonic imaging techniques. The signals from the contrast signal detector <b>128</b>, the processors <b>37</b> and <b>130</b>, and the three dimensional image signals are coupled to a video processor <b>140</b> where they may be selected for two or three dimensional display on an image display <b>50</b> as dictated by user selection.

0029It has been found that harmonic imaging of tissue and blood can reduce near field clutter in the ultrasonic image. It is believed that the harmonic response effect in tissue is dependent upon the energy level of the transmitted waves. Near to an array transducer which is focused at a greater depth, transmitted wave components are unfocused and of insufficient energy to stimulate a detectable harmonic response in the near field tissue. But as the transmitted wave continues to penetrate the body, the higher intensity energy will give rise to the harmonic effect as the wave components begin to focus. While both near and far field regions will return a fundamental frequency response, clutter from these signals is eliminated by the passband of the digital filter <b>118</b>, which is set to the harmonic frequency band. The harmonic response from the tissue is then detected and displayed, while the clutter from the near field fundamental response is eliminated from the displayed image.

0030<figref idref="DRAWINGS">FIGS. 2</figref>, <b>3</b>, <b>4</b>, and <b>5</b> illustrate some of the properties of harmonic return signals which can be utilized to advantage in ultrasonic imaging. It should be appreciated that several of these properties and their interactions are not yet fully and commonly understood among the scientific community, and are still the subject of research and discussion. <figref idref="DRAWINGS">FIG. 2</figref> illustrates the spatial response, and specifically the main lobe and sidelobes, of fundamental and harmonic signals received by a transducer array <b>112</b>. In this illustration the array is directed to image an area of the body behind the ribs, such as the heart, and the main lobe is seen to extend between ribs <b>10</b> and <b>10</b>′. Overlying the ribs is a tissue interface <b>12</b>, as from a layer of fat between the skin and ribs. The FIG. shows a main lobe of the fundamental signals FL<b>1</b>, and on either side of the main lobe are sidelobes FL<b>2</b> and FL<b>3</b>. The FIG. also shows the main lobe HL<b>1</b> of a harmonic of the fundamental frequency, and sidelobes HL<b>2</b> and HL<b>3</b> of the harmonic main lobe.

0031In this example it is seen that the main lobe of the fundamental echoes is wide enough to encompass portions of the ribs <b>10</b>,<b>10</b>′. Accordingly, acoustic energy at the fundamental can be reflected back toward the transducer <b>112</b> as indicated by the arrow <b>9</b>. While some of the energy of this reflection may travel back to and be received directly by the transducer, in this example some of the reflected energy is reflected a second time by the tissue interface <b>12</b>, as indicated by arrow <b>9</b>′. This second reflection of energy reaches the other rib <b>10</b>′, where it is reflected a second time as shown by arrow <b>9</b>″ and travels back to and is received by the transducer <b>112</b>.

0032Since the intent of this imaging procedure is to image the heart behind the ribs, these echoes reflected by the ribs are unwanted artifacts which contaminate the ultrasonic image. Unwanted echoes which are reflected a number of times before reaching the transducer, such as those following the paths of arrows <b>9</b>,<b>9</b>′,<b>9</b>″, are referred to as multipath artifacts. Together, these artifacts are referred to as image “clutter”, which clouds the near field and in some cases all of the image. This near field haze or clutter can obscure structure which may be of interest near the transducer. Moreover, the multipath artifacts can be reproduced in the image at greater depths due to the lengthy multiple paths traveled by these artifacts, and can clutter and obscure regions of interest at greater depths of field.

0033But when only the harmonic return signals are used to produce the ultrasonic image, this clutter from the fundamental frequencies is filtered out and eliminated. The main lobe HL<b>1</b> of the received harmonic echoes is narrower than that of the fundamental, and in this example passes between the ribs <b>10</b>,<b>10</b>′ without intersecting them. There are no harmonic returns from the ribs, and no multipath artifacts from the ribs. Thus, the harmonic image will be distinctly less cluttered and hazy than the fundamental image, particularly in the near field in this example.

0034<figref idref="DRAWINGS">FIG. 3</figref> shows a second example in which the main lobes of both the fundamental and harmonic returns do not intersect the ribs, and the problem discussed in <figref idref="DRAWINGS">FIG. 2</figref> does not arise. But in this example the ribs <b>10</b>, <b>10</b>′ are closer to the skin surface and the transducer <b>112</b>. While the main lobes do not intersect the ribs, the sidelobes FL<b>2</b> of the fundamental do reach the ribs, allowing sidelobe energy to be reflected back to the transducer as shown by reflection path <b>9</b>. Again, this will produce clutter in the fundamental image. But the smaller and narrower sidelobes HL<b>2</b> of the received harmonic energy do not reach the ribs. Again, the harmonic image will exhibit reduced clutter as compared to the fundamental image.

0035<figref idref="DRAWINGS">FIG. 4</figref> illustrates the fundamental and harmonic beam patterns in a perspective which is across the lobes of <figref idref="DRAWINGS">FIGS. 2 and 3</figref>, that is, across the axis of the transducer. This drawing illustrates the relative amplitude responses of the fundamental and second harmonic beam patterns. Illustrated are the dynamic response DRF between the main (FL<b>1</b>) and first sidelobe (FL<b>2</b>) of the fundamental component of the sound beam, and the dynamic response DRH between the main (HL<b>1</b>) and first sidelobe (HL<b>2</b>) of the second harmonic component. If responses due to the main lobes are considered desired signal responses, and responses due to the sidelobes are considered to be clutter or noise, the signal to noise ratio of the harmonic is greater than that of the fundamental. That is, there is relatively less sidelobe clutter in a harmonic image than in the corresponding fundamental image of the same transmission, or DRH> DRF.

0036<figref idref="DRAWINGS">FIG. 5</figref> illustrates another comparison of the properties of fundamental and harmonic signals, which is the relative amount of energy (in units of acoustic pressure P) emanating from increasing depths Z in the body at the fundamental and second harmonic frequencies. The curve denoted Fund. shows the buildup of propagated acoustic energy at the fundamental frequency. While the curve is seen to peak at the focus of the array transducer, it is seen that there is nonetheless an appreciable amount of fundamental energy at the shallower depths before the focal region. In comparison, there is comparatively much less energy, and a lesser buildup of energy, at the harmonic frequency propagated at these lesser depths of field. Hence, with less energy available for multipath reverberation and other aberrations, there is less near field clutter with harmonic imaging than with imaging the fundamental echo returns from the same transmission.

0037<figref idref="DRAWINGS">FIG. 8</figref> illustrates the bands of received signals and the digital filter of a typical <figref idref="DRAWINGS">FIG. 1</figref> embodiment of the present invention for a transmitted signal of four cycles of a 1.67 MHz acoustic wave. Transmitting multiple cycles narrows the bandwidth of the transmitted signal; the greater the number of cycles, the narrower the bandwidth. In response to this transmission, the transducer <b>112</b> receives a fundamental signal in a bandwidth <b>90</b>, which is seen to peak at the transmitted frequency of 1.67 MHz. As the fundamental frequency band rolls off, the harmonic band <b>92</b> comes up, and is seen to exhibit a peak return at the harmonic frequency of 3.34 MHz. The received signals are applied to a digital filter with a passband characteristic <b>94</b>, which is seen to be centered around the harmonic frequency of 3.34 MHz. As <figref idref="DRAWINGS">FIG. 8</figref> shows, this passband will substantially suppress signals at the fundamental frequency while passing the harmonic signals on to further processing and image formation. When imaging the heart in this manner, it has been found that the harmonic response of the endocardial tissue of the heart is quite substantial, and harmonic tissue images of the heart show a clearly defined endocardial border.

0038Other signal processing techniques besides filtering may be used to separate out harmonic signals from received echo information such as cancellation of the fundamental frequencies in a broadband signal, leaving only the harmonic frequencies. For example, U.S. Pat. No. 5,706,819 discloses a two pulse technique, whereby each scanline is insonified by consecutive fundamental frequency pulses of opposite phase in rapid succession. When the resultant echoes are received from the two pulses and combined on a spatial basis, the fundamental frequencies will cancel and the nonlinear or harmonic frequencies will remain. Thus, the harmonic frequencies are separated from the broadband echo signals without the need for a filter circuit.

0039<figref idref="DRAWINGS">FIG. 10</figref> shows a portion of a preferred embodiment of the present invention in block diagram form, from the beamformer output through to the image display. This embodiment not only produces harmonic images of tissue and blood flow, but also overcomes signal dropout deficiencies of conventional imaging systems which arise when imaging patients with difficult to image pathology. Additionally, this embodiment reduces an artifact of coherent ultrasound images known as speckle. In <figref idref="DRAWINGS">FIG. 10</figref>, the signal and data lines connecting the blocks of the block diagram all represent multi-conductor digital data paths, as the processor of the illustrated embodiment is entirely digital. Scanline echo data from the beamformer <b>116</b> is applied in parallel to the two channels <b>30</b>a,<b>30</b>b of the processor illustrated in <figref idref="DRAWINGS">FIG. 10</figref>, one of which is a high frequency channel and the other of which is a low frequency channel. Each channel of the processor has a normalization stage <b>32</b>,<b>132</b> which multiplies the scanline data by a scale factor on a sample by sample basis to produce gain or attenuation that can vary with the depth of the body from which each sample returned. The scale factor for each channel is provided by normalization coefficients stored in or generated by coefficient circuits <b>32</b>,<b>132</b>, which in a preferred embodiment are digital memories. As the multiplying coefficients are changed along the sequence of scanline echoes, depth dependent gain or attenuation is produced.

0040The function of the normalization stages is two-fold. One is to compensate for a transducer aperture which expands with depth of scan. As signals from an increasing number of transducers are used with increasing depth, the magnitude of the summed beamformed signals will increase. This increase is offset by reduced gain (increased attenuation) in the normalization stage, in proportion to the rate at which channels are added to the beamforming process, so that the resultant echo sequence will be unaffected by the changing aperture.

0041The second function of the normalization stages is to equalize the nominal signal amplitudes of the two channels <b>30</b>a,<b>30</b>b. The nominal signal amplitudes of the passbands of the two channels are desirably the same, so that the original relative signal levels will be preserved after the passbands are summed to create the full harmonic passband. But ultrasound signals are subject to depth dependent attenuation which varies with frequency, higher frequencies being more greatly attenuated with depth than lower frequencies. To account for this depth dependent attenuation the coefficients for the normalization stages provide signal gain which increases with depth. Since the two channels employ different frequency passbands, the depth dependent gain of the two channels differs from one channel to the other. In particular, the rate of gain increase for the higher frequency passband channel is greater than that of the lower frequency passband channel. This is illustrated in <figref idref="DRAWINGS">FIG. 11</figref>, which, for purposes of illustration, shows the normalization gain characteristic of the higher frequency passband channel separated into two components. The depth dependent characteristic <b>200</b> offsets the effect of an increasing aperture in the channel, and the depth dependent characteristic <b>202</b> compensates for depth dependent signal attenuation. The low frequency passband channel may also have a depth dependent gain characteristic but with a different characteristic <b>202</b> for the different rate of attenuation of the lower frequencies. The high frequency passband channel has a similar but more rapidly increasing depth dependent gain characteristic to account for the more rapid rate of attenuation of the higher frequencies. Each depth dependent gain characteristic <b>202</b> is chosen to offset the effect of depth dependent gain for the particular frequency passband employed by that channel.

0042In a preferred embodiment the coefficients of the coefficient circuits apply a gain or attenuation characteristic which is a combination of the two characteristics <b>200</b>,<b>202</b>. Preferably, the coefficient memories <b>32</b>,<b>132</b> store multiple combined gain curves which are changed with memory addressing to match scanhead characteristics or the type of signals being processed (2D or Doppler). The rate of gain change may be controlled by the rate at which the coefficients are changed for the multiplier of each normalization stage <b>30</b>,<b>130</b>.

0043The normalized echo signals in each channel are coupled to quadrature bandpass filters (QBPs) in each channel. The quadrature bandpass filters provide three functions: band limiting the RF scanline data, producing in-phase and quadrature pairs of scanline data, and decimating the digital sample rate. Each QBP comprises two separate filters, one producing in-phase samples (I) and the other producing quadrature samples (Q), with each filter being formed by a plurality of multiplier-accumulators (MACs) implementing an FIR filter. One such MAC is shown in <figref idref="DRAWINGS">FIG. 12</figref>. As an echo sample of the scanline data is applied to one input of a digital multiplier <b>210</b> a coefficient is applied to the other multiplier input. The product of the echo sample and the weighting coefficient is stored in an accumulator <b>212</b> where it may be accumulated with previous products. Other MACs receive the echo samples at different phases and likewise accumulate weighted echo samples. The accumulated outputs of several MACs can be combined, and the final accumulated product comprises filtered echo data. The rate at which accumulated outputs are taken sets the decimation rate of the filter. The length of the filter is a product of the decimation rate and the number of MACs used to form the filter, which determine the number of incoming echo samples used to produce the accumulated output signal. The filter characteristic is determined by the values of the multiplying coefficients. Different sets of coefficients for different filter functions are stored in coefficient memories <b>38</b>,<b>138</b>, which are coupled to apply selected coefficients to the multipliers of the MACs. The MACs effectively convolve the received echo signals with sine and cosine representative coefficients, producing output samples which are in a quadrature relationship.

0044The coefficients for the MACs which form the I filter implement a sine function, while the coefficients for the Q filter implement a cosine function. For bandpass filtering, the coefficients of the active QBPs additionally implement a low pass filter function that is frequency shifted to form, in combination with the sine (for I) and cosine (for Q) functions, a bandpass filter for the quadrature samples. In the instant example, QBP₁in channel <b>30</b>a is producing I and Q samples of the scanline data in a first, low frequency passband, and QBP₂in channel <b>30</b>b is producing I and Q samples of the scanline data in a second, higher frequency passband. Thus, the spectrum of the original broadband echo signals is divided into a high frequency band and a low frequency band. To complete the dropout and speckle reduction process, the echo data in the passband produced by QBP₁of channel <b>30</b>a is detected by a detector <b>40</b>₁and the detected signals are coupled to one input of a summer <b>48</b>. In a preferred embodiment detection is performed digitally by implementing the algorithm (I²+Q²)^1/2. The echo data in the complementary passband produced by QBP₂of channel <b>30</b>b is detected by a detector <b>402</b> and these detected signals are coupled to a second input of the summer <b>48</b>. When the signals of the two passbands are combined by the summer <b>48</b>, the decorrelated signal dropout and speckle effects of the two passbands will at least partially cancel, reducing the signal dropout and speckle artifacts in the 2D image created from the signals.

0045Following the detector in each subchannel is a gain stage formed by multipliers <b>44</b>₁,<b>44</b>₂which receive weighting coefficients from coefficient memories <b>42</b>₁,<b>42</b>₂. The purpose of this gain stage is to partition the balance of analog and digital gains in the ultrasound system for optimal system performance. Some of the gains in the echo signal path may be automatically implemented by the ultrasound system, while others, such as manual gain control and TGC gain, may be controlled by the user. The system partitions these gains so that the analog gains preceding the ADCs (analog to digital converters) of the beamformer are adjusted optimally for the dynamic input range of the ADCs. The digital gain is adjusted to optimize the brightness of the image. The two gains together implement gain control changes effected by the user.

0046In the preferred embodiment the gain imparted to the scanline signals by the multipliers <b>44</b>₁,<b>44</b>₂is selected in concert with the gain of the preceding normalization stage <b>34</b>,<b>134</b> in the channel. The gain of each normalization stage is chosen to prevent the attainment of saturation levels in the QBPs, as may occur when strong signals from contrast agents or harmonic imaging are being received. To prevent saturation levels the maximum gain of the normalization stage is controlled, and any reduction imposed by reason of this control is restored by the gain of the succeeding multiplier <b>44</b>₁,<b>44</b>₂.

0047The gain function provided by these multipliers could be performed anywhere along the digital signal processing path. It could be implemented by changing the slope of the compression curves discussed below. It could also, for instance, be performed in conjunction with the gains applied by the normalization stages. This latter implementation, however, would eliminate the ability to effect the saturation control discussed above. The present inventors have found implementation of this gain function to be eased when provided after detection, and in the preferred embodiment by use of a multiplier after detection.

0048The signals produced by the gain stages <b>44</b>₁,<b>44</b>₂generally exhibit a greater dynamic range than may be accommodated by the display <b>50</b>. Consequently, the scanline signals of the multipliers are compressed to a suitable dynamic range by lookup tables. Generally compression is logarithmic, as indicated by log compression processors <b>46</b>₁,<b>46</b>₂. The output of each lookup table is proportional to the log of the signal input value. These lookup table are programmable so as to provide the ability to vary the compression curves, and the brightness and dynamic range of the scanline signals sent on for display.

0049The present inventors have found that the use of log compression to scale the echo signals can affect low level signals near the baseline (black) level of the signal dynamic range by exacerbating the degree and the number of echoes with components at the black level, a manifestation of the destructive interference arising from the speckle effect of the coherent ultrasonic energy. When the echo signals are displayed, many of them will be at the black level, and appear in the image to have been undetected or dropped out. The embodiment of <figref idref="DRAWINGS">FIG. 10</figref> reduces this problem by producing separate, partially decorrelated versions of the echo signals in the two channels <b>30</b>a,<b>30</b>b. This embodiment partially decorrelates the echo signal versions by separating the echo signal components into two different passbands as shown in <figref idref="DRAWINGS">FIG. 13</figref>. The two passbands can be completely separated or, as shown in this example, overlapping. In this example, the lower passband <b>300</b>a is centered about a frequency of 3.1 MHz, and the higher passband <b>300</b>b is centered about a frequency of 3.3 MHz, a center frequency separation of only 200 kHz. Even this small degree of separation has been found sufficient to decorrelate the signal components of the two passbands sufficiently such that black level signal dropout in one passband will frequently not align in frequency with its corresponding component in the other passband. Consequently, when these decorrelated replicas of the same echo signal are combined by the summer <b>48</b>, the signal dropout and speckle artifacts will be markedly reduced. This is especially significant when trying to image fine structures at deep depths in the body, such as the endocardium. A harmonic image of the endocardium is significantly improved by the artifact elimination effects of the embodiment of <figref idref="DRAWINGS">FIG. 10</figref>.

0050As discussed previously the signal gain of the two passbands <b>300</b>a,<b>300</b>b of <figref idref="DRAWINGS">FIG. 13</figref> can be matched to preserve the original signal levels after summation. However, in a preferred embodiment, the lower frequency passband is processed with less dynamic range than the higher frequency passband as shown in <figref idref="DRAWINGS">FIG. 13</figref>. This has the effect of suppressing the fundamental frequency contributions of the lower frequency passband (which contains more fundamental frequency components than the higher frequency band.) This is accomplished as a component of different compression characteristics in the log compression processors <b>46</b>₁, <b>46</b>₂, or elsewhere in the channels <b>30</b>a,<b>30</b>b subsequent to the separation of the broadband signal into separate passbands.

0051The processed echo signals at the output of the summer <b>48</b> are coupled to a lowpass filter <b>52</b>. This lowpass filter, like the QBPs, is formed by combinations of multiplier-accumulators with variable coefficients, arranged to implement an FIR filter, to control the filter characteristic. The lowpass filter provides two functions. One is to eliminate sampling frequency and other unwanted high frequency components from the processed echo signals. A second function is to match the scanline data rate to the vertical line density of the display <b>50</b>, so as to prevent aliasing in the displayed image. The FIR filter performs this function by selectively decimating or interpolating the scanline data. The filtered echo signals are then stored in an image memory <b>54</b>. If the scanlines have not yet been scan converted, that is, they have r,θ coordinates, the scanlines are scan converted to rectilinear coordinates by a scan converter and greyscale mapping processor <b>56</b>. If scan conversion has been performed earlier in the process, or is not needed for the image data, the processor <b>56</b> may simply convert the echo data to the desired greyscale map by a lookup table process. The image data may then be stored in a final image memory or sent to a video display driver (not shown) for conversion to display signals suitable for driving the display <b>50</b>.

0052It will be appreciated that, due to the advantage of the quick programmability of a digital filter, the processing described above can be performed in an embodiment which utilizes a single one of the channels <b>30</b>a, <b>30</b>b to process the echo data from a scanline twice to alternately produce a line of signals for each of the two passbands in a time-interleaved fashion. However, the use of two parallel channels affords twice the processing speed, enabling harmonic images to be produced in real time and at twice the frame rate of a time multiplexed embodiment.

0053Harmonic images produced from high frequency signals can suffer from depth dependent attenuation as the echo signals return from increasing depths in the body. Lower frequency fundamental signals may experience less attenuation, and hence in some cases may exhibit better signal to noise ratios at greater depths. The embodiment of <figref idref="DRAWINGS">FIG. 14</figref> takes advantage of this characteristic by blending fundamental and harmonic image data in one image. It is possible, for instance, to create a normal tissue image of the heart from fundamental frequencies, and overlay the fundamental frequency tissue image with a harmonic tissue image of the heart to better define the endocardial border in the composite image. The two images, one from fundamental frequency components and another from harmonic frequency components, may be formed by alternately switching the digital filter <b>118</b> between fundamental and harmonic frequencies to separately assemble fundamental and harmonic images, or by employing the two parallel filters of <figref idref="DRAWINGS">FIG. 10</figref> with two passbands, one set to pass fundamental frequencies and the other set to pass harmonic frequencies. In <figref idref="DRAWINGS">FIG. 14</figref>, the filter of channel <b>30</b>a is set to pass fundamental signal frequencies, and echo signals passed by this channel are stored in a fundamental image memory <b>182</b>. Correspondingly, harmonic signal frequencies are passed by channel <b>30</b>b and stored in a harmonic image memory. The fundamental and harmonic images are then blended together by a proportionate combiner <b>190</b>, under control of a blend control <b>192</b>. The blend control <b>192</b> may automatically implement a pre-programmed blending algorithm, or one directed by the user. For example, the proportionate combiner <b>190</b> may create a blended image which uses only echo data from the harmonic image at shallow depths, then combines echo data from both images at intermediate depths, and finally only uses echo data of the fundamental image at deep depths. This combines the reduced clutter benefit of harmonic echo data at shallow depths and the greater penetration and signal to noise ratio of fundamental echoes received from deeper depths, while affording a smooth transition from one type of data to the other at intermediate depths. Other combining algorithms are also possible, such as simply switching from one type of data to another at a predetermined depth, or outlining a region of the image to be displayed with one type of data while the remainder of the image is displayed using the other type of data.

0054It is also possible to employ the two parallel filters and blend the components together before image formation, thereby adding a controllable component of the harmonic echo signals to the fundamental frequency signals to enhance the resultant image. Such an embodiment could eliminate the need for separate fundamental and harmonic image memories and would process the signal components directly to a blended image memory.

0055A third technique for producing blended images is to receive each scanline of the image through a depth-dependent, time varying filter. Such filters are well known for improving the signal to noise ratio of received echo signals in the presence of depth dependent attenuation as shown, for instance, in U.S. Pat. No. 4,016,750. For the production of blended fundamental and harmonic images, the passband <b>210</b> of a time varying filter is initially set to pass harmonic frequencies f_h, as shown in <figref idref="DRAWINGS">FIG. 15</figref>, as echo signals begin to be received from shallow depths. When it becomes desirable to begin supplementing the image with fundamental signal components at deeper depths, the passband <b>210</b> undergoes a transition to lower frequencies, eventually moving to the fundamental frequencies ff as shown by passband <b>212</b> in <figref idref="DRAWINGS">FIG. 15</figref>. In the case of a digital filter such as that shown in <figref idref="DRAWINGS">FIG. 9</figref>, the change in passband frequencies is effected by changing the filter coefficients with time. As the filter undergoes this transition, the passband passes fewer harmonic frequencies and greater fundamental frequencies until eventually, if desired, the passband is passing only fundamental frequencies at the maximum image depth. By receiving each scanline through such a time varying filter, each line in the resultant image can comprise harmonic frequencies in the near field (shallow depths), fundamental frequencies in the far field (deepest depths), and a blend of the two in between.

0056Harmonic tissue images of moving tissue can also be formed by processing the received harmonic tissue echo signals with the processor described in U.S. Pat. No. 5,718,229, entitled MEDICAL ULTRASONIC POWER MOTION IMAGING.

0057Thus, the present invention encompasses an ultrasonic imaging system for imaging the nonlinear response of tissue and fluids of the body to ultrasound by transmitting a fundamental frequency signal, receiving an echo signal from the tissue at a non-fundamental, preferably harmonic, frequency, detecting the non-fundamental frequency echo signals, and forming an image of the tissue and fluids from the non-fundamental frequency echo signals. As used herein the term harmonic also refers to harmonic frequencies of higher order than the second harmonic and to subharmonics, as the principles described herein are equally applicable to higher order and subharmonic frequencies.