Ultrasound imaging system with contrast phase adaptive PSF thinning

Abstract

of the disclosure: An ultrasound imaging system acquires, stores, and displays a sequence of ultrasound images acquired during a cycle of contrast agent wash-in and wash-out. A time-intensity curve of contrast perfusion is produced from the sequence of images and is used by a PSF thinning processor to adaptively vary the degree of PSF thinning for different phases of contrast perfusion. Preferably the PSF thinning is varied during the processing and replay of stored images from a low degree of PSF thinning which images microbubbles with high sensitivity but low resolution at the start of contrast agent infusion, to a high degree of PSF thinning at peak perfusion for the production of high resolution, low sensitivity images of small vasculature.

Description

[0001] 2024P00266WG - 1-

[0002] ULTRASOUND IMAGING SYSTEM WITH

[0003] CONTRAST PHASE ADAPTIVE PSF THINNING

[0004] This invention relates to medical diagnostic ultrasound systems and, in particular, to ultrasound imaging systems which perform contrast-enhanced ultrasound imaging ( CEUS ) .

[0005] Harmonic contrast agents have been available for many years for enhancing blood flow and tissue perfusion in ultrasound imaging . Contrast agents are solutions of tiny microbubbles which act as strongly nonlinear echogenic reflectors in the bloodstream . Importantly, the reflective properties of microbubbles , their nonlinear behavior, and their ability to be ruptured by ultrasonic pulse energy, cause their echo returns to exhibit a broad spectrum of energy, particularly at the second harmonic band of frequencies . The echo signals can be processed to separate out a selected band or bands of frequencies which are used for imaging . The relatively high intensity of the harmonic echo signals returned from microbubble contrast agents causes the blood vessels and tissue perfused by the microbubbles in the image field to light up in the image and be easily distinguished for diagnosis of characteristics of blood flow and tissue perfusion .

[0006] The si ze of the microbubbles in a contrast agent solution, as mentioned above , is very small , often in the range of 1 to 5 micrometers . As such, a microbubble is a point source target for ultrasound imaging . But a typical ultrasound imaging system cannot clearly image point source targets of this tiny si ze . The point source imaging capability of an ultrasound imaging system, like other imaging systems , is governed by a parameter known as the point spread function ( PSF) . The PSF is determined by characteristics of the imaging system, and not of the obj ect being imaged . For an optical imaging system such as a telescope , the PSF can be a function of lens si zes and thicknesses , mirror si zes , and the spacing of the lenses and mirrors . For an ultrasound system the governing characteristics for PSF determination include the imaging frequency, the focal depth, and transducer aperture characteristics . Consequently, the image of a microbubble produced by an ultrasound system will generally appear as a blurry blob which is 0 . 5 to 1 . 0 millimeters wide .

[0007] For the same reasons , microvessels which are 20 to 50 microns in diameter can appear inaccurately as being 1 millimeter wide .

[0008] A recent advance in ultrasound imaging which addresses these problems is known as PSF thinning . PSF thinning is an image processing technique which takes the blurry image of a microbubble or other point source target and reduces its outer periphery so that it will be reproduced as a smaller circle which emphasi zes the original image ' s centermost pixels . At limit , the smallest PSF-thinned image would comprise only the center pixel of the blurry image . The resultant image will thus appear to more accurately display point source targets in a truer si ze . A number of PSF thinning algorithms for image processing are known that can reproduce point source targets such as microbubbles with di f ferent selectable si zes and appearances .

[0009] Thus , a CEUS image of microbubble perfusion of tissue and blood vessels can be processed by PSF thinning to give the resultant image the appearance of greater resolution . By selecting the amount or degree of PSF thinning, individual microbubbles in 2024P00266WC -3- the anatomy can appear as blurry blobs in one image and sharply defined dots in another, presenting the clinician with what seem to be poorly resolved and highly resolved images of the same anatomy and contrast agent . It would seem that a more highly resolved image would always be the preferred one , but there are instances when this is not the case . For instance , at the start of a CEUS exam no microbubbles are visible in the image field, either because an infusion of contrast agent into the bloodstream has j ust begun and has not yet reached the target anatomy, or a high intensity " flash" of ultrasound has disrupted all of the microbubbles in the image field . The clinician will generally be interested in viewing the first arrival of microbubbles in the image field, in order to sense where , when, and at what rate blood flow begins to perfuse the tissue and blood vessels . To do this with greater precision, it is desirable for the ultrasound imaging system to exhibit high sensitivity to the presence of small amounts of contrast agent in the anatomy . At a later time during the exam, after the flow of contrast agent has greatly perfused the tissue and blood vessels , the clinician' s interest in high sensitivity imaging can be replaced by a desire for more highly resolved images . An implementation of the present invention is directed to addressing these changing needs and desires of the clinician during a CEUS exam .

[0010] In accordance with the principles of the present invention, an ultrasound system is automatically controlled during a CEUS exam to use PSF thinning to produce images with high sensitivity and low resolution of microbubbles as the contrast agent begins to perfuse the target region, and with high resolution but reduced sensitivity to microbubbles during later phases of perfusion during the exam . The time-varying control is implemented by determining di f ferent phases of microbubble perfusion from a time-intensity curve of perfusion produced from the ultrasound image data . Typically, the timeintensity curve is used to delineate the early, mid- and peak phases of perfusion of the target anatomy .

[0011] A variable parameter of the PSF thinning technique is changed in correspondence with the changing perfusion phases so that sensitivity to microbubbles is emphasi zed during the initial phase of contrast agent perfusion, and resolution of microbubbles is emphasi zed during later phases of contrast agent perfusion .

[0012] A preferred ultrasound system of the present invention comprises an ultrasound probe adapted to acquire ultrasonic echo signals from an image field experiencing a wash-in and wash-out of a microbubble contrast agent ; a beamformer coupled to the ultrasound probe and adapted to produce beamformed echo signals ; a contrast image processor coupled to the beamformer, and adapted to produce images of anatomy perfused with the microbubble contrast agent ; an image display coupled to the contrast image processor and adapted to display the images of anatomy perfused with the microbubble contrast agent ; a time-intensity curve processor responsive to echo signals returned from the microbubble contrast agent and adapted to produce a time-intensity curve of microbubble contrast agent wash-in and wash-out ; a Cineloop memory coupled to the contrast image processor and adapted to store a sequence of the images of anatomy perfused with the microbubble contrast agent acquired during wash-in and wash-out of the microbubble contrast agent ; a PSF thinning processor coupled to the Cineloop memory and to the image display, and responsive to the stored sequence of images of anatomy perfused with the microbubble contrast agent and to the time-intensity curve of microbubble contrast agent wash-in and wash-out , and adapted to identi fy di f ferent phases of perfusion from the time-intensity curve and to perform di f ferent degrees of PSF thinning of microbubble images in the sequence of images for di f ferent phases of perfusion which vary from high sensitivity to microbubbles in an early perfusion phase to high resolution of microbubbles in a peak perfusion phase .

[0013] A preferred method of the present invention comprises the steps of using an ultrasound probe to acquire ultrasonic echo signals from an image field experiencing a wash-in and wash-out of a microbubble contrast agent ; using a beamformer to produce beamformed echo signals ; using a contrast image processor to produce images of anatomy perfused with the microbubble contrast agent ; using an image display to display the images of anatomy perfused with the microbubble contrast agent ; using a timeintensity curve processor to produce a time-intensity curve of microbubble contrast agent wash-in and washout ; using a Cineloop memory to store a sequence of the images of anatomy perfused with the microbubble contrast agent acquired during wash-in and wash-out of the microbubble contrast agent ; and using a PSF thinning processor to identi fy di f ferent phases of perfusion from the time-intensity curve and to perform di f ferent degrees of PSF thinning of microbubble images in the sequence of images for di f ferent phases of perfusion which vary from high sensitivity to microbubbles in an early perfusion 2024P00266WG -6- phase to high resolution of microbubbles in a peak perfusion phase; and displaying a sequence of the stored images which has been processed by the PSF thinning processor.

[0014] IN THE DRAWINGS:

[0015] FIGURES 1 (a) , 1 (b) , and 1 (c) illustrate an ultrasound image of a point source target processed with different degrees of PSF thinning.

[0016] FIGURES 2 (a) , 2 (b) , and 2 (c) illustrate Gaussian plots of the intensity distribution of the image data of FIGURES 1 (a) to 1 (c) .

[0017] FIGURE 3 illustrates ultrasound images of a tumor and its feeder vessels during successive phases of a contrast agent wash-in, wash-out cycle when processed by high sensitivity, low resolution PSF thinning .

[0018] FIGURE 4 illustrates ultrasound images of a tumor and its feeder vessels during successive phases of a contrast agent wash-in, wash-out cycle when processed by high resolution, low sensitivity PSF thinning .

[0019] FIGURE 5 illustrates in block diagram form an ultrasound system constructed in accordance with the principles of the present invention.

[0020] FIGURES 6(a) , 6(b) , and 6(c) illustrate timeintensity curves of contrast agent wash-in, wash-out with different phase delineations.

[0021] FIGURE 7 is a flowchart illustrating the major steps of processing a sequence of CEUS images in accordance with the principles of the present invention .

[0022] FIGURE 8 is a flowchart illustrating the computation of a time-intensity curve and its use to delineate phases for dynamic PSF thinning of a sequence of CEUS images. 2024P00266WG -7-

[0023] FIGURE 9 is a CEUS image where a high degree of PSF thinning is used to visuali ze areas of the imaging containing the vasculature of a tumor and a low degree of PSF thinning is used to visuali ze areas of normal tissue .

[0024] FIGURE 10 is a flowchart illustrating three methods to employ PSF thinning to adaptively vary the balance of sensitivity to resolution during successive perfusion phases of a sequence of CEUS images .

[0025] FIGURE 1 ( a ) illustrates a contrast agent microbubble when imaged by an ultrasonic imaging system exhibiting a typical point spread function . This particular image was acquired by repetitively acquiring images of the same microbubble , then compounding the images , to more clearly illustrate the consequences of the point spread function when imaging a point source target such as a microbubble . As the image illustrates , the point source image is not a single dot , but a blurry disc which is brightest in the center and progressively less distinct at the periphery . Its large si ze is not representative of the actual si ze of a microbubble in the image field . The intensity distribution of the blurred image is illustrated by the Gaussian plot of image intensity of FIGURE 2 ( a ) shown to the right of FIGURE 1 ( a ) .

[0026] FIGURE 1 (b ) illustrates the microbubble image of FIGURE 1 ( a ) when processed by a low degree of PSF thinning . As the image shows , the brightness in the center of the image is stronger and more concentrated, and the peripheral outer ring of lower intensities has been diminished or eliminated . The outer boundary of the microbubble image is more clearly defined . As a consequence , the microbubble 2024P00266WG -8- image appears to be more highly resolved and distinct, and is closer in size to its true size as compared with FIGURE 1 (a) . FIGURE 2 (b) to the right illustrates the intensity distribution of FIGURE 1 (b) in a Gaussian plot.

[0027] FIGURE 1 (c) illustrates the microbubble image of FIGURE 1 (a) when processed by a high degree of PSF thinning. This has resulted in an even greater concentration of intensity at the center of the microbubble and an even greater reduction in the size of the microbubble image. The outer boundary of the microbubble is even more clearly defined. Image resolution appears to be far better than either of the preceding microbubble images. FIGURE 2 (c) to the right illustrates the intensity distribution of FIGURE 1 (c) in a Gaussian plot.

[0028] What the microbubble images of FIGURES 1 (a) -1 (c) illustrate is that the image processing of PSF thinning can be used to greatly enhance the apparent resolution of an ultrasound image, particularly one with point sources or specular reflectors such as microbubbles. However, it should be kept in mind that the images of FIGURES 1 (a) -1 (c) are of a single microbubble and are thus at a scale which is calibrated in millimeters or fractions of a millimeter. Diagnostic ultrasound is generally not used to image a single microbubble, it is used to image whole organs or sections of organs with extensive vasculature containing dozens or hundreds or more microbubbles when well perfused with a contrast agent. The scale of such images is calibrated in centimeters. Consequently, when the microbubble image of FIGURE 1 (c) is scaled up to the scale of a typical CEUS diagnostic image, the dot representing the single microbubble will become so 2024P00266WG - 9- tiny as to virtually disappear . In imaging terms , the imaging sensitivity to individual or small numbers of microbubbles is quite low when a high degree of PSF thinning is employed . This is important at the beginning of a CEUS imaging procedure , when the clinician is striving to view the initial arrival of the first few microbubbles in vasculature of interest and imaging sensitivity is of considerable importance .

[0029] This aspect of PSF thinning is more readily apparent in the sequence of images of FIGURES 3 and 4 . Each of these illustrations shows a sequence of the same three images of the same anatomy, a tumor and its feeding vessels , as they appear during three phases of contrast infusion . The first image of each drawing is at an early phase ( 1 . 18 sec . ) of contrast infusion, the second image is at a mid-phase ( 2 . 94 sec . ) and the third image ( 4 . 71 sec . ) is at a peak phase of perfusion . The arrows 10 and 12 are pointing to the larger feeding vessels for the tumor at the left edge of the images . To the right of the feeding vessels is the tumor and its microvasculature , which is supplied with blood from the feeding vessels . The images of FIGURES 3 and 4 di f fer only in that the FIGURE 3 images are processed with a low degree of PSF thinning and thus appear more sensitive to the initial arrival of contrast agent microbubbles but with low resolution, whereas the FIGURE 4 images are processed with a high degree of PSF thinning and thus appear more highly resolved but with less sensitivity . It is thus seen that the feeding vessels indicated by arrow 10 are clearly illuminated by the recently arrived contrast agent in the first phase image of FIGURE 3 , but in the first phase image of FIGURE 4 , the low number of recently 2024P00266WG - 10- arrived microbubbles are di f ficult to discern in the image due to the low sensitivity created by the higher degree of PSF thinning . Of the two images , the 1 . 18 sec . image of FIGURE 3 would be of greater interest to the clinician as it is more sensitive to the initial arrival of contrast agent .

[0030] But as more contrast agent arrives and perfuses the anatomy, the image preference changes . At peak perfusion ( 4 . 71 sec . ) , the feeding vessels and the tumor microvasculature become well perfused with microbubbles . The large amount of microbubble perfusion causes the feeding vessels and tumor microvasculature become so bright as to bloom in the image , as seen in the peak perfusion image of FIGURE 3 , when little or no PSF thinning is applied . But when a large degree of PSF thinning is performed during peak perfusion, the feeding vessels and tumor microvasculature are both brightly illuminated and well defined, as seen in the peak perfusion image at the right side ( 4 . 71 sec . ) of FIGURE 4 . A clinician may therefore prefer the first image of FIGURE 3 during the initial phase of contrast agent perfusion due to its greater sensitivity, and the third image of FIGURE 4 at peak perfusion due to its better resolution and with good sensitivity due to the high concentration of contrast agent . An implementation of the present invention will provide the clinician with the balance of sensitivity and resolution which is optimal for the current phase of contrast agent perfusion being observed during the progress of a CEUS exam .

[0031] Referring now to FIGURE 5 , an ultrasound system constructed in accordance with the principles of the present invention is shown in block diagram form . An ultrasound probe 100 includes an array 102 of ultrasonic transducer elements that transmits ultrasonic pulses and receives ultrasonic echo signals . The array may be a one-dimensional linear or curved array for two-dimensional imaging, or may be a two-dimensional matrix of transducer elements for electronic beam steering and focusing in two or three dimensions . The ultrasonic transducer elements in the array 102 transmit beams of ultrasonic energy by their timed actuation under control of a transmit controller 28 , and receive echoes returned in response to each transmission . Echoes from the transmitted ultrasonic energy are received by the transducer elements of the array 102 , which generate echo signals that are coupled through a transmit / receive ( T / R) switch 22 and digiti zed by analog to digital converters when the system uses a digital beamformer 30 . Analog beamformers may alternatively be used . Control of the ultrasound system and of various control settings for imaging such as probe selection and ROI ( region of interest ) delineation is ef fected by user manipulation of the controls of a user control panel 20 , such as keys , pushbuttons , and a trackball or computer mouse , which are coupled to various circuitry and processors of the ultrasound system . In the illustrated system the user controls are coupled to provide user input to the transmit controller 28 and a PSF thinning processor 46 , as well as other controllers and processors of the ultrasound system .

[0032] The echo signals received by the transducer elements of the array 102 are delayed and summed by the beamformer 30 to form coherent echo signals along scanline (beam) directions for an image . The digital coherent echo signals are then filtered by a signal processor 24 , which may also perform noise reduction as by spatial or frequency compounding or persistence processing. The signal processor can also shift the frequency band of the coherent echo signals to a lower or baseband frequency range. The signal processor can be configured as shown in U.S. Patent No. 5,833, 613 (Averkiou et al.) , for example. When phase information is needed as is the case for Doppler processing, quadrature (I and Q) demodulation may also be performed on the echo signals. In the illustrated system, the transmit band centered around frequency foand the receiver frequency band are individually controlled so that the beamformer 30 is free to receive a band of frequencies which is different from that of the transmitted band such as one including a harmonic frequency band around frequency 2fo, the second harmonic.

[0033] The beamformed and processed coherent echo signals are coupled to a nonlinear signal separator 32. The nonlinear signal separator can separate second harmonic echo signals with a high pass filter, but preferably it separates harmonic frequencies of echoes returned from contrast agent microbubbles by the pulse inversion technique, in which echo signals resulting from the transmission of multiple, differently phased (inverted) pulses to an image location are additively combined to cancel fundamental signal components and enhance harmonic components, thus producing echo signals in a second harmonic band 2fo. The harmonic signals can alternatively be separated by amplitude-modulated pulse inversion as described in US Pat. 5,577,505 (Brock-Fisher et al.) The same echo signals are subtractively combined to produce echo signals in a fundamental frequency band fo. A preferred phase (polarity) pulse inversion technique is described in U.S. patent 6,186,950 (Averkiou et al.) and in U.S. patent 5,706,819 (Hwang et al.) for instance.

[0034] Harmonic echo signals from a contrast agent, such as microbubbles, are coupled to a contrast image processor 38. Contrast agents are often used to more clearly delineate blood vessels, or to perform perfusion studies of the microvasculature of tissue as described in US Pat. 6, 692,438 (Skyba et al.) for example, or in an implementation of the present invention. In the system shown in FIGURE 5, echoes from a contrast agent are used to produce both contrast images by the contrast image processor, and time-intensity curves (TICs) of perfusion from selected regions of interest (individual pixel locations or groups of pixels) in an image field. For the parametric contrast images described in the Skyba et al. patent, a 3 by 3 group of pixels is preferred for a pixel area from which to form a timeintensity curve. The contrast image processor produces an anatomical contrast image by amplitude (or envelope) detection of the harmonic frequency echoes from each point in the image field. One way to do this when the echoes are quadrature demodulated is to calculate the signal amplitude at each pixel location in the form of (I2+Q2)1'5. These contrast intensity signals are mapped to the desired display format by scan conversion which converts samples from R-0 coordinates (used in radial scanning) to Cartesian (x,y) coordinates for display of a spatially defined image.

[0035] The fundamental frequency echo signals are coupled to a B mode processor 36 which produces a standard B mode tissue image. The B mode processor performs in the same manner as the contrast image processor, but operates on fundamental frequency 2024P00266WC - 14- echoes . The echo signals are amplitude ( envelope ) detected and scan converted to produce a spatially delineated image of tissue in the image field . The contrast and B mode images are coupled to a display processor 40 which performs the processing needed to display the images on an image display 42 . This may include displaying two images at the same time , side- by-side . It may also comprise overlaying perfusion parameter colors over the B mode images so that perfusion parameters are shown in relation to the tissue structure in which the contrast agent is located .

[0036] The harmonic frequency signals returned from contrast agent microbubbles may also be used to delineate phases of contrast wash-in and wash-out by forming time-intensity curves of the contrast wash-in and wash-out . Time-intensity curves are formed by a time-intensity curve ( TIC ) processor 34 for each point (pixel ) in a contrast image , or by selected groups of pixels or the pixels of an entire image . Using the harmonic signal amplitudes acquired during wash-in and wash-out of the contrast agent , curves of contrast intensity at each pixel location or group of pixel locations are calculated by the TIC processor as described in US pat . pub . no . 2011 / 0208061 ( Chang) . A time-intensity curve can be graphically displayed by a graphics processor 26 , which is coupled to a display processor 40 and the display 42 . Images produced by the contrast image processor and the B mode processor are coupled to the display processor 40 , which processes the images for display on the image display 42 , either alone or side-by-side with a contrast image or a B mode image from the B mode processor . The system may also include a Doppler processor for processing and display of 2024P00266WC - 15- spectral and colorflow Doppler images .

[0037] In accordance with the principles of the present invention, the ultrasound system of FIGURE 5 also includes a Cineloop® memory 44 and a PSF thinning processor 46 . As real-time images are acquired by the system and displayed on the image display 42 , they can simultaneously be stored as an image sequence ( loop ) in the Cineloop memory for later recall and review . For example , the live images of an entire CEUS exam can be saved in the Cineloop memory starting from prior to the arrival of contrast agent in the image field, and continuing through wash-in and the completion of wash-out of the perfusion cycle . The clinician can then replay the entire CEUS exam from memory at a later time or date such as when diagnosis is performed .

[0038] In accordance with a further aspect of the present invention, the images of a CEUS exam can be recalled from memory such as the Cineloop memory 44 , selectively processed in sequence by the PSF thinning processor 46 , and the modi fied images replayed in real time on the image display and, i f desired, the modi fied exam images can be stored as a sequence ( loop ) in the Cineloop memory . The PSF thinning processor, using the time-intensity curve produced by the TIC processor 34 during the CEUS exam and stored with the image sequence in the Cineloop memory, can perform di f ferent degrees of PSF thinning on the microbubbles in the image sequence and can vary the degree of PSF thinning in correspondence with the changing phases of contrast agent perfusion .

[0039] FIGURES 6 ( a ) - 6 ( c ) illustrate examples of timeintensity curves which may be produced by the TIC processor 34 . In FIGURE 6 ( a ) the perfusion curve 60 is seen to begin to build at starting time to and rise 2024P00266WG - 16- during contrast agent wash-in to a peak perfusion point A. Thereafter, as the passage of a bolus of contrast agent declines , the time-intensity curve declines steadily during the wash-out phase . In FIGURE 6 (b ) the time-intensity curve 60 is divided into phases delineated by the attainment of di f ferent levels of perfusion by the contrast agent . Following time to the time-intensity curve reaches a noise threshold level for the imaging system which starts a perfusion phase that continues until a contrast threshold is attained . This phase is often referred to as the early arterial perfusion phase . Another phase commences at that point , the mid-arterial perfusion phase , and continues until the peak arterial perfusion phase is attained around the peak A of the perfusion curve 60 . Thereafter a first wash-out phase occurs until the contrast threshold is reached, after which the time-intensity curve ends with a final wash-out phase .

[0040] In FIGURE 6 ( c ) a time-intensity curve 60 is divided into phases by time markers occurring during perfusion wash-in and wash-out . An early wash-in phase occurs from the commencement of the curve to time ti as indicated by point 63 on the time-intensity curve 60 . A mid- wash-in phase extends from time ti to time t2 as indicated by point 65 on the curve . Thereafter a peak perfusion phase extends from time t2 to time ta as indicated by point 67 . A first wash-out phase extends from point 67 to point 69 on the curve at time ta , and the final wash-out phase extends for the balance of wash-out following time ta . Phases of perfusion which are delineated in this manner are used by the PSF thinning processor 46 to vary the degree of PSF thinning during the contrast agent wash-in, wash=out cycle . The PSF thinning processor 46 can execute any known or later developed PSF thinning process or algorithm. For example, a morphological operation using image erosion may be used. Image erosion is performed on each frame of the contrast image sequence stored in the Cineloop memory to erode away the boundaries of microbubbles in an image and leave shrunken areas of contrast agent signals. The degree of PSF thinning is controlled by the size of a structuring element (e.g., the desired degree of PSF thinning defines a size of the structuring element) . The structuring element may be a shape used to apply functions to the image frame. The shape of the structuring element can be a simple square or rectangle in some instances, but may be more complex shapes in other instances. In this technique, a high degree of PSF thinning corresponds to a large size of structuring element. The larger the structuring element, the more boundaries of a microbubble image are eroded away, leaving smaller sizes of remaining contrast agent signals. A low degree of PSF thinning corresponds to a smaller size of the structuring element and fewer microbubble boundaries are eroded away, leaving larger sizes of remaining contrast agent signal areas. The size of the structuring element may be adapted spatially and / or temporally.

[0041] In some implementations, the output of the image erosion operation may be referred to as a mask (Mask) . The Mask may be normalized with values between 0 and 1. A power of an exponent is applied to the Mask. The final output is the product of the original input (Input) and the normalized Mask with the exponent as shown in the equations below: Equation (1) Equation (2) The exponent (Exponent) is used to control the degree of PSF thinning in addition to the structuring element. The degree of PSF thinning increases as the exponent increases (e.g., greater than 1) .

[0042] In a second PSF thinning technique, a morphological operation using image dilation may be used. In this technique, a structuring element such as the one described in the first example is also used. Image dilation is provided by the formula:

[0043] Input (dilation (Inputs Equation (3)

[0044] The Output is the output (i.e., thinned) image and the Input is the input contrast image. The output image is equal to the input image divided by input image after a dilation operation is applied. The degree of PSF thinning in this example is determined by the size of a structuring element of image dilation. For a high degree of PSF thinning (large size of structuring element) , more boundaries are enlarged, leaving smaller regions of remaining contrast agent signals. For a low degree of PSF thinning (smaller size of structuring element) , less boundaries are enlarged, leaving larger regions of remaining contrast agent signals. The size of the structuring element may be adapted temporally in correspondence with the current phase of the wash-in, wash-out cycle. The dilation technique may be applied to each image frame of the contrast loop.

[0045] In one implementation, the image dilation is a grayscale dilation algorithm. In such an algorithm, the dilation of an image pixel is the maximum of the image pixel in its neighborhood, with that neighborhood defined by the structuring element. After the dilation operation, the microbubble image expands to a larger size depending on the degree of PSF thinning (size of the structuring element) . The output of this step (for each image frame) is referred as dilation ( Input ) in Equation 3. The input image may then be scaled based on the dilation output. According to Equation 3, the input image is scaled with the output of the dilation step. Specifically, each pixel of the output image, output (x, y) , is the product of an image pixel, input (x, y) , and the corresponding scaling factor, 1 / [dilation (input) (x, y) ] . Due to the scaling factor, the output microbubble image shrinks to a smaller size depending on the degree of PSF thinning. Optionally, a normalization step may be applied to the output to remove the outliers (e.g. infinite elements) , and normalize the output dynamic range to certain limits.

[0046] Similar to the first example describing image erosion, the Output of Equation 3 may be referred to as a mask which may be normalized and raised to an exponent that may be used to control the degree of PSF thinning as described in Equation 2. The final output may then be the original image multiplied by the normalized mask with an exponent as described in Equation 2.

[0047] In a third example, a spatial smoothing thinning technique is used. The spatial smoothing may be described by the equation: fnput smoothing ( / nput) Equation (4) s the output (i.e., thinned) image and the Input is the input contrast image. The output image is equal to the input image divided by the input image after a spatial smoothing operation. The degree of PSF thinning is determined by the size of the smoothing kernel for spatial smoothing. A high degree of PSF thinning corresponds to a large smoothing kernel size (e.g., 8x8) and more boundaries are smoothed, leaving smaller sizes of the remaining microbubble regions. A low degree of PSF thinning corresponds to a smaller smoothing kernel size (e.g., 3x3) and fewer boundaries are smoothed, leaving larger sizes of remaining microbubble regions. Generating a smoothing kernel is similar to generating the structuring element of the previous examples. The shape of the smoothing kernel may be a simple square or rectangle. The size of the smoothing kernel may be adapted temporally in correspondence with the current phase of the wash-in, wash-out cycle.

[0048] Similar to the examples utilizing morphological operations, the Output of Equation 4 may be referred to as a mask which may be normalized and raised to an exponent that may be used to control the degree of PSF thinning as described in Equation 2. The final output may then be the original image multiplied by the normalized mask with an exponent as described in Equation 2.

[0049] In a fourth example, a low-pass filter (LPF) thinning technique is used. The LPF technique may be described by the equation: Equation (5)

[0050] The Output is the output (i.e., thinned) image and the Input is the input contrast image. The output image is equal to the input image divided by input image after a spatial smoothing operation. The degree of PSF thinning is the cut-off spatial frequency of the spatial LPF. A high degree of PSF thinning corresponds to a lower cut-off frequency and more boundaries are filtered, leaving smaller sizes of remaining microbubble regions. A low degree of PSF thinning corresponds to a higher cut-off frequency and fewer boundaries are filtered, leaving larger sizes of remaining microbubble regions. The cut-off frequency may be adapted spatially and / or temporally in correspondence with the current phase of the wash-in, wash-out cycle.

[0051] In the low-pass filter technique, a 2D spatial Fast Fourier Transform (FFT) is performed on the input image. The LPF is then applied to the output of the FFT. This step removes and / or suppresses high spatial frequency components (i.e., the spatial frequency components above the cut-off frequency) of the input image. An inverse FFT is then performed on the filtered image which brings the image from the frequency domain back to the image (spatial) domain. Because the high frequency components have been removed and / or suppressed, the microbubble image expands to a larger size depending on the degree of PSF thinning. The output of this step is referred to as LPF (Input) in Equation 5. The input image is then scaled based on the LPF output. According to Equation 5, the input image is scaled with the output of the LPF step. Specifically, each pixel of the output image, output (x, y) , is the product of the image pixel, input (x, y) , and the corresponding scaling factor, 1 / [LPF (input) (x, y) ] . Due to the scaling factor, the output microbubble image shrinks to a smaller size depending on the degree of PSF thinning. Optionally, a normalization step may be applied to the output to remove the outliers (e.g. infinite elements) , and normalize the output dynamic range to certain limits.

[0052] Similar to the examples describing morphological operations, the Output of Equation 5 may be referred 2024P00266WC -22- to as a mask which may be normali zed and raised to an exponent that may be used to control the degree of PSF thinning as described in Equation 2 . The final output is then the original image multiplied by the normali zed mask with the exponent as described in Equation 2 .

[0053] FIGURE 7 illustrates at a high level a method for performing PSF thinning of the images of a CEUS exam in correspondence with the phases of the contrast agent wash-in, wash-out cycle . In step 70 , a Cineloop of CEUS images is acquired . In step 72 measurements of image data are used to identi fy di f ferent phases of contrast perfusion . In a preferred implementation a time-intensity curve processor acquires measures of the buildup of contrast agent in some or all of the image field which are used to produce a time-intensity curve . In step 74 the Cineloop of images is replayed and the images processed by PSF thinning that varies dynamically in correspondence with di f ferent phases of contrast agent perfusion . In a preferred implementation a low degree of PSF thinning is employed initially to maintain a high sensitivity (but relatively low resolution) for the arrival of contrast agent microbubbles at the earliest phase of the wash-in, wash-out cycle , and changing to a high degree of PSF thinning to produce images with good resolution (but reduced sensitivity) at the peak perfusion phase . The changing balance between sensitivity and resolution provided by varying the PSF thinning thus provides the clinician with the desired balance of these characteristics throughout a CEUS exam diagnosis . This enables the clinician to clearly visuali ze the feeding vessels of a tumor at an early phase of contrast perfusion while being able to visuali ze the smallest vessels of the tumor at peak perfusion .

[0054] FIGURE 8 illustrates a method for computing and using a time-intensity curve in an implementation of the present invention . In step 80 a time-intensity curve is computed using contrast image data . There are various ways to do this . One is to compute the average intensity of all of the pixels of each image . As the images progressively light up with the arrival of more contrast agent in the image field, the perfused tissue and vessels of the image will light up with increasing intensity and each image as a whole will increase in overall intensity . The pixel intensity over the entire image is averaged, and these averaged values are used to compute a point of the time-intensity curve . Another way to develop values for computation of a time-intensity curve is to only use image intensity values for certain areas of each image . For example , the clinician could define two regions of interest (ROI s ) in the image field, one in a region of diagnostic interest and another in a region which is not of diagnostic interest . For diagnosis of a tumor, one ROI could be in the tumor and a second ROI could be in nearby normal tissue , for instance . The di f ferences in average intensity between these two ROI s are then used to create the time-intensity curve , and the time-intensity curve is then used to separate the image frames of the Cineloop into early- , mid- , and peak-perfusion phases as indicated in step 82 . These phases could then be followed by wash-out phases as desired . The delineated perfusion phases could then be included in the Cineloop images such as by indicating its perfusion phase in the corner of each image . Another alternative is to color-code the phases and add a border to each image which is in the color of its perfusion phase .

[0055] With the perfusion phases thus defined for each image of the Cineloop sequence , the image sequence is replayed with perfusion-phase-corresponding PSF thinning of each image before it is displayed . The sensitivity / resolution balance of the images is thereby dynamically adj usted for di f ferent phases of perfusion as stated in step 84 . There are various ways to adj ust the degree of PSF thinning during the image sequence . One is to incrementally vary the degree of PSF thinning during replay of the image sequence , with the increments of change varying with the slope of the perfusion curve during each phase . For instance , the "mid-" section of the wash-in curve generally exhibits a much steeper slope that latter phases of the wash-out curve . This would dictate the use of larger increments of increased PSF thinning during the mid- portion of the wash-in curve and smaller increments of decreased PSF thinning during the latter phase of wash-out . Another possibility is to use the same degree of PSF thinning continuously during each phase of perfusion, with step-wise increases or decreases at each change of phase .

[0056] Another way to control the degree of PSF thinning is to process each image two ways , one with a high degree of PSF thinning and another with a low degree of PSF thinning . During Cineloop replay, a blend of the two modes of PSF thinning is produced and displayed for each image . For example the weighting of the high sensitivity / low resolution image would predominate in the early phases of contrast agent perfusion, and would gradually or step-wise change in successive perfusion phases so that the weighting of the high resolution / low 2024P00266WG -25- sensitivity image would predominate at peak perfusion. A further variation of this approach is to determine the relative weighting of the two images using local (e.g., ROI) image information. For example, for an image region where feeding vessels for a tumor are present, the weighting for blending would favor the high sensitivity / low resolution image. For an image region containing many microvessels, the weighting would favor the high resolution / low sensitivity image. An ultrasound image produced using this technique is shown in FIGURE 9, which may be compared with the images of FIGURES 3 and 4. Arrow 100 indicates two localized windows in the FIGURE 9 image where feeding vessels and microvessels of a tumor are located. The tumor- related anatomy in these windows is processed with a high degree of PSF thinning so that the vessels in the windows are displayed with high resolution but reduced sensitivity. The rest of the image is processed with a low degree of PSF thinning that favors high sensitivity over resolution. Thus, the majority of the image, where only normal anatomy is present, and which is of less diagnostic interest, is sensitive to visualizing the extensive vasculature in the normal regions of the image.

[0057] FIGURE 10, as shown and labelled in block 90, summarizes three different ways that the PSF thinning processor or equivalent may adjust PSF thinning for replay of a Cineloop of CEUS images. The first approach, in block 92, is to employ different degrees of PSF thinning so as to increase resolution and decrease sensitivity during successive phases of perfusion during replay of the Cineloop. The second approach, in block 94, is to progressively blend images produced in a high sensitivity, low resolution 2024P00266WC -26- imaging mode with images produced in a low sensitivity, high resolution imaging mode during Cineloop replay . Preferably, the blending favors the high sensitivity, low resolution mode images at the beginning of contrast agent perfusion, and varies the blending to progressively more greatly favor low sensitivity, high resolution mode images as the peak perfusion phase of contrast perfusion is approached and attained . The third approach described in block 96 is to adaptively blend images produced by a high sensitivity, low resolution imaging mode with images produced by a low sensitivity, high resolution imaging mode in di f ferent ROI s of an image during Cineloop replay . One ROI may favor blending that favors high sensitivity imaging while another ROI may favor blending that favors high resolution, with the blending varying with di f ferent phases of a timeintensity curve of contrast agent perfusion .

[0058] Other features of an ultrasound system of the present invention can provide additional user benefit or convenience . During the initial live acquisition of a loop of CEUS images as a bolus of contrast agent washes into the region of interest and thereafter washes out , the time-intensity curve or curves produced by the TIC processor 34 cannot be fully formed until contrast agent wash-in, wash-out is completed . It would therefore be useful for the ultrasound system to create a first frame of the Cineloop sequence which contains the time-intensity curve ( s ) so that a completed curve can be immediately acquired by the PSF thinning processor 46 at the outset of Cineloop replay and used by the processor to identi fy the perfusion phases used for dynamic variation of PSF thinning as replay of the loop proceeds . Another alternative is to embed the time- intensity curve in each image of the Cineloop . Such a display of both the ultrasound image and the timeintensity curve in each image frame would enable the display of a marker on each curve of each image in the Cineloop, showing the precise perfusion phase point of each image . As the image loop is replayed, the marker would appear to travel along the timeintensity curve , continuously showing the clinician precisely where the current image is in the wash-in, wash-out cycle of contrast agent perfusion phases .

[0059] It should be noted that an ultrasound system suitable for use in an implementation of the present invention, and in particular the component structure of the ultrasound system of FIGURE 5 , may be implemented in hardware , software , or a combination thereof . The various embodiments and / or components of an ultrasound system, for example , the transmit controller, the TIC processor, the contrast image processor, the graphics processor, and the PSF thinning processor and the components and controllers therein, also may be implemented as part of one or more computers or microprocessors . The computer or processor may include a computing device , an input device , a display unit and an interface , for example , for accessing the Internet . The computer or processor may include a microprocessor . The microprocessor may be connected to a communication bus , for example , to access a PACS system or the data network for importing training images . The computer or processor may also include a memory . The memory devices such as the Cineloop memory 44 may include Random Access Memory (RAM) and Read Only Memory (ROM) . The computer or processor further may include a storage device , which may be a hard disk drive or a removable storage drive such as a floppy disk drive , optical disk drive , solid-state thumb drive , and the like . The storage device may also be other similar means for loading computer programs or other instructions into the computer or processor .

[0060] As used herein, the term " computer" or "module" or "processor" or "workstation" may include any processor-based or microprocessor-based system including systems using microcontrollers , reduced instruction set computers (RISC ) , AS ICs , logic circuits , and any other circuit or processor capable of executing the functions described herein . The above examples are exemplary only, and are thus not intended to limit in any way the definition and / or meaning of these terms .

[0061] The computer or processor executes a set of instructions that are stored in one or more storage elements , in order to process input data . The storage elements may also store data or other information as desired or needed . The storage element may be in the form of an information source or a physical memory element within a processing machine .

[0062] The set of instructions of an ultrasound system including those controlling the acquisition, processing, and display of ultrasound images as described above may include various commands that instruct a computer or processor as a processing machine to perform speci fic operations such as the methods and processes of the various embodiments of the invention . The set of instructions may be in the form of a software program . The software may be in various forms such as system software or application software and which may be embodied as a tangible and non-transitory computer readable medium . Further, the software may be in the form of a collection of separate programs or modules such as a neural network model module , a program module within a larger program or a portion of a program module . The software also may include modular programming in the form of obj ect-oriented programming . The processing of input data by the processing machine may be in response to operator commands , or in response to results of previous processing, or in response to a request made by another processing machine .

[0063] Furthermore , the limitations of the following claims are not written in means-plus- function format and are not intended to be interpreted based on 35 U . S . C . 112 , sixth paragraph, unless and until such claim limitations expressly use the phrase "means for" followed by a statement of function devoid of further structure .

Claims

WHAT IS CLAIMED IS :1 . An ultrasonic diagnostic imaging system for conducting contrast-enhanced ultrasound imaging comprising : an ultrasound probe ( 100 ) adapted to acquire ultrasonic echo signals from an image field experiencing a wash-in and wash-out of a microbubble contrast agent ; a beamformer ( 30 ) coupled to the ultrasound probe and adapted to produce beamformed echo signals ; a contrast image processor ( 38 ) coupled to the beamformer, and adapted to produce images of anatomy perfused with the microbubble contrast agent ; an image display ( 42 ) coupled to the contrast image processor and adapted to display the images of anatomy perfused with the microbubble contrast agent ; a time-intensity curve processor ( 34 ) , responsive to echo signals returned from the microbubble contrast agent , and adapted to produce a time-intensity curve of microbubble contrast agent wash-in and wash-out ; a Cineloop memory ( 44 ) , coupled to the contrast image processor, and adapted to store a sequence of the images of anatomy perfused with the microbubble contrast agent acquired during wash-in and wash-out of the microbubble contrast agent ; a point spread function ( PSF) thinning processor ( 46 ) , coupled to the Cineloop memory and to the image display, and responsive to the stored sequence of images of anatomy perfused with the microbubble contrast agent and to the time-intensity curve of microbubble contrast agent wash-in and wash-out , and adapted to identi fy di f ferent phases of perfusion from the time-intensity curve and to perform2024P00266WC -31- di f ferent degrees of PSF thinning of microbubble images in the sequence of images for di f ferent phases of perfusion which vary from high sensitivity to microbubbles in an early perfusion phase to high resolution of microbubbles in a peak perfusion phase .2 . The ultrasonic diagnostic imaging system of Claim 1 , wherein the PSF thinning processor is further adapted to perform a low degree of PSF thinning for images produced during the early perfusion phase and a high degree of PSF thinning for images produced during the peak perfusion phase .3 . The ultrasonic diagnostic imaging system of Claim 2 , wherein the PSF thinning processor is further adapted to identi fy an early perfusion phase , a mid-perfusion phase , and a peak perfusion phase during the contrast agent wash-in portion of the time-intensity curve , and to perform a di f ferent degree of PSF thinning during each of those three phases .4 . The ultrasonic diagnostic imaging system of Claim 3 , wherein the PSF thinning processor is further adapted to continuously vary the degree of PSF thinning during each of the three phases .5 . The ultrasonic diagnostic imaging system of Claim 1 , wherein the PSF thinning processor is further adapted to identi fy di f ferent phases of perfusion delineated by amplitude levels of the timeintensity curve .6 . The ultrasonic diagnostic imaging system of Claim 1 , wherein the PSF thinning processor is2024P00266WC -32- further adapted to identify different phases of perfusion delineated by points of time of the timeintensity curve during a wash-in, wash-out cycle of contrast agent.

7. The ultrasonic diagnostic imaging system of Claim 1, wherein the PSF thinning processor is further adapted to perform different degrees of PSF thinning of microbubble images in the sequence of images by: processing each image in the sequence of images with a low degree of PSF thinning so as to produce a high sensitivity, low resolution image; processing each image in the sequence of images with a high degree of PSF thinning so as to produce a low sensitivity, high resolution image; and blending the two processed images with different weighting to produce a final image, the blending weighting varying for different phases of perfusion.

8. The ultrasonic diagnostic imaging system of Claim 7, wherein the blended weighting varies from greater weighting for high sensitivity, low resolution images during an early phase of contrast agent perfusion to greater weighting for low sensitivity, high resolution images during a peak phase of contrast agent perfusion.

9. The ultrasonic diagnostic imaging system of Claim 7, wherein the blended weighting is different for different regions of interest in the same image.

10. A method for conducting a contrast-enhanced ultrasound imaging exam comprising: using an ultrasound probe (100) to acquire2024P00266WC -33- ultrasonic echo signals from an image field experiencing a wash-in and wash-out of a microbubble contrast agent ; using a beamformer ( 30 ) to produce beamformed echo signals ; using a contrast image processor ( 38 ) to produce images of anatomy perfused with the microbubble contrast agent ; using an image display ( 42 ) to display the images of anatomy perfused with the microbubble contrast agent ; using a time-intensity curve processor ( 34 ) to produce a time-intensity curve of microbubble contrast agent wash-in and wash-out ; using a Cineloop memory ( 44 ) to store a sequence of the images of anatomy perfused with the microbubble contrast agent acquired during wash-in and wash-out of the microbubble contrast agent ; and using a point spread function ( PSF) thinning processor ( 46 ) to identi fy di f ferent phases of perfusion from the time-intensity curve and to perform di f ferent degrees of PSF thinning of microbubble images in the sequence of images for di f ferent phases of perfusion which vary from high sensitivity to microbubbles in an early perfusion phase to high resolution of microbubbles in a peak perfusion phase ; and displaying a sequence of the stored images which has been processed by the PSF thinning processor .11 . The method of Claim 10 , wherein using a PSF thinning processor further comprises : increasing the resolution and decreasing the sensitivity of the images of the sequence of images during replay of a Cineloop of the images for2024P00266WC -34- di f ferent phases of contrast agent perfusion .12 . The method of Claim 10 , wherein using a PSF thinning processor further comprises : progressively blending high sensitivity, low resolution images with low sensitivity, high resolution images during replay of the images stored in Cineloop .13 . The method of Claim 10 , wherein using a PSF thinning processor further comprises : adaptively blending high sensitivity, low resolution images with low sensitivity, high resolution images in di f ferent regions of interest of images during Cineloop replay .14 . The method of Claim 10 , further comprising : visually identi fying the perfusion phase of a displayed image in the display of each image15 . The method of Claim 14 , further comprising : displaying the time-intensity curve concurrently with the display of each image .16 . A computer program product implemented in a tangible and non-transitory computer readable medium, comprising instructions for executing in a computer processor the following steps : obtaining a sequence of ultrasound images of anatomy perfused with a microbubble contrast agent ; producing a time-intensity curve ( TIC ) of microbubble contrast agent wash-in and wash-out ; storing the sequence of images acquired during wash-in and wash-out of the microbubble contrast2024P00266WC -35- agent ; identifying different phases of perfusion from the TIC; performing different degrees of a point spread function (PSF) thinning of microbubbles in the sequence of images for the different phases of perfusion, the PSF thinning varying from high sensitivity to microbubbles in an early perfusion phase to high resolution of microbubbles in a peak perfusion phase; and storing a sequence of the stored images which has been processed by the PSF thinning processor.

17. The computer program product of Claim 16, further comprising instructions to display the sequence of stored images which has been processed by the PSF thinning processor.

18. The computer program product of Claim 17, wherein the PSF thinning is different for different regions of interest in the same image, and wherein the display displays the different regions of interest .

Images