Method of high spatial and temporal resolution ultrasound imaging of microvessels

Abstract

Systems and methods for high spatial and temporal resolution ultrasound imaging of microvessels in a subject are described. Ultrasound data is acquired from a region of interest in a subject to which a microbubble contrast agent has been administered. Ultrasound data is acquired as microbubbles move through or otherwise exist in a region of interest. The region of interest may include, for example, microvessels or other microvasculature in the subject. By imaging the microbubbles, a cross-correlation map between each microbubble image and a point spread function of the system may be generated. A high resolution, high contrast image of the microvasculature may then be generated using accumulation of the power-based cross-correlation maps.

Description

This application is a divisional application of the invention patent application with an international application date of 2021-06-15, an international application number of PCT / US2021 / 037464, an application number entering the Chinese national phase of 202180005313.4, and titled "Method for high spatial and temporal resolution ultrasound imaging of microvessels". Cross-references to related applications

[0001] This application claims the benefit of U.S. Provisional Patent Application Serial No. 63 / 039,549, filed on June 16, 2020, and entitled “Method for High Spatial and Temporal Resolution Ultrasound Imaging of Microvasculature,” which is incorporated herein by reference for all purposes as if fully set forth. A note about federally funded research

[0002] This invention was made with government support under NS111039 awarded by the National Institutes of Health. The government has certain rights in this invention. Background Art

[0003] Ultrasound contrast microbubbles are routinely used in the clinic to enhance ultrasound imaging. The typical size of microbubbles ranges from 1–5 μm, which is much smaller than the wavelength of ultrasound, which is typically on the order of 100–800 μm. As a result, individual microbubbles appear as fuzzy point sources with a size similar to the ultrasound wavelength. Moreover, microbubbles that are less than half the ultrasound wavelength apart cannot be individually resolved, a result known as the diffraction limit of ultrasound.

[0004] Contrast-enhanced ultrasound (CEUS) is now widely and routinely used in clinical practice to provide highly sensitive blood flow and perfusion imaging in superficial or deep tissues with extremely high temporal resolution (real time). Although CEUS can provide quantitative assessments of relative blood volume and tissue perfusion, it is still limited by its spatial resolution and has difficulty identifying microvasculature structures. Clinically, imaging of vascular structure and hemodynamics at the microvascular or capillary level (such as less than 100 microns) is ideal because many pathological changes in tissues occur at the small vessel level, such as tumors.

[0005] Super-resolution ultrasound techniques have been previously disclosed and are capable of imaging microvasculature structures beyond the ultrasound diffraction limit based on precise localization of individual microbubbles (MBs) and tracking of MB positional movement. Despite the improvement in spatial resolution, poor temporal resolution remains a major challenge, where tens of seconds of data acquisition may be required to generate microvasculature images, which seriously hinders the clinical translation of this technology. Clinically relevant acquisition times may be on the order of 1 or 2 seconds. Super-resolution techniques may not be able to provide images of fully filled vascular lumens in such time frames, and forcing the technique to do so may result in degradation of microvasculature visualization. In addition, super-resolution methods typically rely on complex localization, MB pairing, and tracking algorithms, which involve high computational costs.

[0006] Therefore, there is still a need for a high-definition ultrasound microvascular imaging technique that can achieve both high spatial and temporal resolution. It is also necessary to be able to generate images in a very short time period, which may be similar to clinical CEUS protocols. Summary of the invention

[0007] The present disclosure solves the above shortcomings by providing a method for imaging microvessels with high spatial and temporal resolution using an ultrasound system. Ultrasound data can be acquired from an image storage system or from a region of interest of a subject to which a microbubble contrast agent has been administered. Ultrasound data is acquired when microbubbles move through the region of interest or are otherwise present in the region of interest. The region of interest may include, for example, microvessels or other microvascular systems in the subject. The cross-correlation map between each microbubble image and the point spread function of the system can be used to generate a high-resolution, high-contrast microvascular system image. Velocity estimation can also be performed by determining the microbubble trajectory between frames of the cross-correlation map.

[0008] In one configuration, a method for imaging microvessels with high spatial and temporal resolution using an ultrasound system is provided. The steps of the method include accessing ultrasound data through a computer system, wherein the ultrasound data is acquired from a region of interest of a subject using the ultrasound system, and a microbubble contrast agent is already present in the region when the ultrasound data is acquired. The method also includes: generating microbubble signal data using the computer system by separating the microbubble signal in the ultrasound data from other signals in the ultrasound data. The method also includes: determining a point spread function (PSF) of the ultrasound system used to acquire the ultrasound data, and generating a cross-correlation map between the microbubble signal data and the point spread function of the ultrasound system. A microvessel image with high spatial and temporal resolution can be generated based at least in part on the generated cross-correlation map.

[0009] In one configuration, an ultrasound system for high spatial and temporal resolution imaging of microvessels is provided. The system includes a computer system configured to access ultrasound data through the computer system, wherein the ultrasound data is acquired from a region of interest of a subject using the ultrasound system, and a microbubble contrast agent is already present in the region when the ultrasound data is acquired. The computer system is also configured to generate microbubble signal data using the computer system by separating the microbubble signal in the ultrasound data from other signals in the ultrasound data. The computer system is also configured to determine a point spread function (PSF) of the ultrasound system used to acquire the ultrasound data, and to generate a cross-correlation map between the microbubble signal data and the point spread function of the ultrasound system. A microvessel image with high spatial and temporal resolution can be generated by the computer system, at least in part based on the generated cross-correlation map.

[0010] In one configuration, a method for imaging microvessels with high spatial and temporal resolution using an ultrasound system is provided. The method includes: accessing ultrasound data using a computer system, the ultrasound data having been acquired from a region of interest in a subject using an ultrasound system, wherein a microbubble contrast agent was present in the region of interest when the ultrasound data was acquired. The method also includes: generating microbubble signal data using the computer system by separating microbubble signals in the ultrasound data from other signals in the ultrasound data, and isolating individual microbubble trajectories in the microbubble signal data. The method also includes: determining trajectory parameters of the isolated microbubble trajectories, and estimating a microbubble flow velocity based on the determined trajectory parameters of the isolated microbubble trajectories. A microvessel image with high spatial and temporal resolution can be generated based at least in part on the estimated microbubble flow velocity.

[0011] The above and other aspects and advantages of the present disclosure will become apparent through the following description. In the specification, reference is made to the accompanying drawings which form a part of the specification, in which preferred embodiments are shown by way of illustration. However, such embodiments do not necessarily represent the full scope of the invention, and therefore the scope of the invention is interpreted herein with reference to the claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a flow chart setting forth the steps of an example method for generating super-resolution images of microvessels using an ultrasound system.

[0013] Figure 2 Depicted are examples of microvessels moving between consecutive time frames due to tissue motion.

[0014] Figure 3 is a flow chart setting forth the steps of an example method for generating a high-resolution microvascular image based on a power-based cross-correlation method.

[0015] Figure 4A is a non-limiting example of a power-N one-dimensional Gaussian distribution of a cross-correlation calculated between the example data and a point spread function.

[0016] Figure 4B Is from Figure 4A Non-limiting example of the relationship between the FWHM of and the power of the cross-correlation.

[0017] Figure 5 is a non-limiting example of a cross-correlation calculated using different power-N one-dimensional Gaussian distributions calculated between the example data and different power-N point spread functions.

[0018] Fig. 6A ] is an image of a non-limiting example of raw MB ultrasound image data.

[0019] Figure 6B yes Fig. 6A An image of a non-limiting example of a 1st power based cross-correlation map.

[0020] Figure 6C yes Fig. 6A An image of a non-limiting example of a 4th power based cross-correlation map.

[0021] Fig.6D yes Fig. 6A An image of a non-limiting example of a power-of-8 cross-correlation map.

[0022] Fig. 7A is a non-limiting example of a threshold-sharpened image without cross-correlation.

[0023] Figure 7B is a non-limiting example of an image with cross-correlation threshold sharpening set to 0.25.

[0024] Figure 7C is a non-limiting example of an image with cross-correlation threshold sharpening set to 0.5.

[0025] Fig.7D is a non-limiting example of an image with cross-correlation threshold sharpening set to 0.75.

[0026] Figure 8 Depicted are non-limiting examples of raw MB data and cross-correlograms across several frames, and the resulting high-resolution blood flow images through accumulation.

[0027] Fig. 9 Non-limiting examples of raw and processed microvascular images in pancreatic tumor and liver tissue are depicted.

[0028] Fig.10 Non-limiting examples of original and thresholded images are depicted.

[0029] Fig.11A A non-limiting example of MB trajectories in a three-dimensional spatiotemporal space having two spatial dimensions and one temporal dimension is depicted.

[0030] Fig. 11B Depicted Fig.11A Close-up part.

[0031] Fig. 11C A non-limiting example of an isolated MB trajectory in three-dimensional spatiotemporal space is depicted.

[0032] Fig.11D A non-limiting example of an isolated MB trajectory projected onto the axial-temporal plane is depicted.

[0033] Fig.11E A non-limiting example of an isolated MB trajectory projected onto the transverse-temporal plane is depicted.

[0034] Fig.11F is a flow chart setting forth the steps of an example method for generating high-resolution hemodynamic and microvascular images based on the MB velocity estimation method.

[0035] Fig. 12A is an original MB image of a non-limiting example spatially dependent thresholding method based on noise floor estimation.

[0036] Fig. 12B is a non-limiting example measured noise floor image based on a non-limiting example spatially dependent thresholding method for noise floor estimation.

[0037] Fig. 12C is a non-limiting example noisy floor image based on a non-limiting example spatially dependent thresholding method for noise floor estimation.

[0038] Fig.12D is another non-limiting example noisy floor image based on a non-limiting example spatially dependent thresholding method for noise floor estimation.

[0039] Fig.12E is a non-limiting example image before residual noise suppression.

[0040] Fig.12F is a non-limiting example image after residual noise suppression.

[0041] Fig.13 is a block diagram of an example ultrasound system in which the methods described in the present disclosure may be implemented.

[0042] Fig.14is a block diagram of an example computer system that may implement embodiments described in this disclosure. Specific description

[0043] Systems and methods for high spatial and temporal resolution ultrasound imaging of microvessels in a subject are described herein. High-definition microvascular imaging in a subject can be used to image the microvascular system with high resolution and contrast-to-noise ratio (CNR). Conventional ultrasound imaging and CEUS imaging typically display images as signal intensity. According to the present disclosure, high spatial and temporal resolution ultrasound imaging techniques can utilize correlation maps to generate images with significantly improved spatial resolution and CNR. Ultrasound data is acquired from a region of interest in a subject to which a microbubble contrast agent has been administered. Ultrasound data is acquired when microbubbles move through the region of interest or are otherwise present in the region of interest. The region of interest may include, for example, microvessels or other microvascular systems in the subject. By imaging microbubbles (MBs), a cross-correlation map between each MB image and the point spread function (PSF) of the system can be generated. As described in the present disclosure, the accumulation of power-based cross-correlation maps for a certain number of frames can subsequently generate a high-resolution, high-contrast image of the microvascular system.

[0044] Reference now Figure 1 , a flow chart is shown setting forth steps of an example method for using an ultrasound system to generate super-resolution images of microvessels in a subject to which a microbubble contrast agent has been administered, as disclosed in U.S. Patent Application No. 16 / 617,628, incorporated herein by reference. In general, high resolution or super-resolution refers to a resolution that is enhanced relative to a resolution achievable with an imaging system. For example, a high resolution or super-resolution ultrasound image may refer to an image having a resolution finer than the diffraction limit.

[0045] As indicated at step 102, the method includes providing ultrasound data to a computer system. At step 104, tissue motion may be removed from the ultrasound data. At step 106, microbubble signals may be isolated. At step 108, denoising of the microbubble signal data may be performed. In some configurations, microbubbles may be located at step 110 and tracked at step 112, such as as described below. Fig.11F The microvascular image may be generated at step 114 , processed at step 116 , and displayed or stored at step 118 .

[0046] In some embodiments, providing the ultrasound data to the computer system may include retrieving previously acquired ultrasound data from a memory or other data storage device, which may be part of or separate from the computer system. In some other embodiments, providing the ultrasound data may include acquiring such data using an ultrasound system and providing the acquired data to the computer system, which may be part of or separate from the ultrasound system.

[0047] The ultrasound data may be ultrasound radio frequency ("RF") data, ultrasound in-phase quadrature ("IQ") data, etc. In general, the ultrasound data includes one or more spatial dimensions, which may include a transverse dimension, an axial dimension, an elevation dimension, and combinations thereof. For example, the ultrasound data may include two spatial dimensions, such as a transverse dimension and an axial dimension. The ultrasound data may also include a time dimension, such as a slow time dimension (i.e., a time direction along which multiple ultrasound signals are collected).

[0048] As described above, ultrasound data is acquired from a subject to which a microbubble contrast agent has been administered. In some embodiments, different microbubbles (e.g., microbubbles of different sizes) having different resonant ultrasound frequencies may be used for imaging, such that by selecting a particular ultrasound frequency (e.g., transmitting or receiving at a particular frequency), only a subset of the selected microbubbles will be imaged, thereby forming ultrasound data containing an isolated microbubble source. As another example, an ultrasound pulse having energy sufficient to rupture a certain number of microbubbles may be used, wherein the ruptured microbubbles then release free bubbles from the microcapsules and generate an ultrasound signal having an amplitude different from that of the intact microbubbles. This effectively creates an isolated microbubble source that can be used for high-resolution imaging.

[0049] The microbubble signal can be obtained from both linear and nonlinear components of the ultrasound. The linear component is usually at the fundamental frequency of the applied ultrasound, while the nonlinear component can be at a harmonic frequency of the applied ultrasound, at the fundamental frequency of the applied ultrasound, or at both. For example, the nonlinearity introduced by an imaging method based on amplitude modulation can be at the fundamental frequency.

[0050] As indicated in step 104, the ultrasound data is processed to remove tissue motion. For in vivo imaging, there may be transducer movement and tissue motion caused by the cardiovascular (e.g., heartbeat and pulsating motion from arteries) and respiratory systems. The amplitude of these motions may be significantly larger than the size of the microvessels to be resolved, causing significant blurring of the microvessel images and resulting in inaccurate measurements of blood flow velocity. Therefore, the ultrasound detected microbubble signal may be processed to remove these tissue motions.

[0051] As an example, ultrasound data can be processed to realign spatially misregistered microbubble signals from the same microvascular location. Figure 2 As shown, due to physiological motion such as breathing, tissue 202 is moved to a different tissue location 204, and target microvessels 206 are moved to different microvessel locations 208. This misregistration (ie, movement) will result in errors in the final microvessel image if not taken into account.

[0052] In order to correct the misalignment caused by tissue motion, image registration and tissue motion compensation methods can be used. For example, an image registration method based on image intensity (such as affine transformation) can be used to translate, rotate, scale and shear the misaligned image back to the original position. Other image registration methods that can be used include image registration based on image features, image registration based on spectral phase, image registration based on transformation model, etc. As another example, a method based on ultrasound speckle tracking (such as two-dimensional normalized cross-correlation) can be used to estimate the motion vector between the reference signal and the moving tissue and microvascular signal. Then, the signal can be realigned by reversing the movement based on the estimated motion vector. As another example, a two-dimensional phase shift mapping between a reference image and a moving image can be estimated, and then the two-dimensional phase shift mapping is applied to the Fourier domain corresponding to the moving image, and then the movement is corrected by inverse Fourier transform. As a non-limiting example, a microbubble signal registration process is performed between each frame of the ultrasound microbubble signal and the reference frame of the microbubble signal (e.g., the first frame of the imaging sequence). Due to the high frame rate acquisition of ultrasound data, the estimated tissue motion can usually be assumed to be smooth in the time direction without sharp peaks that would otherwise indicate physiologically unrealistic rapid tissue movement. This characteristic of tissue motion can be used to suppress false peaks in tissue motion estimation and thus minimize misregistration.

[0053] Reference again Figure 1 After removing tissue motion from the ultrasound data, the microbubble signal is isolated in the ultrasound data, as indicated at step 106. Typically, isolating the microbubble signal includes isolating the microbubble signal from background signals, such as tissue signals and signals from stationary microbubbles that do not change position between acquisition frames (e.g., when the microbubbles do not move between frames). In some embodiments, the microbubble signal may be isolated using frame-to-frame signal subtraction, high-pass filtering along the time direction of the signal, singular value decomposition (“SVD”)-based filtering, and the like.

[0054] As an example, temporal high pass filtering may be used to isolate the microbubble signal. In these implementations, ultrasound data may be filtered using a cutoff frequency that is lower than the temporal frequency of the moving microbubble signal but higher than the temporal frequency of the stationary background signal to isolate the microbubble signal.

[0055] As another example, SVD-based filtering can be used, in which singular value cutoffs can be used to separate background signals (e.g., tissue signals and non-moving microbubble signals, which are typically projected to low-order singular values) from isolated moving microbubble signals (which are typically projected to medium to high-order singular values). As an example, a block-by-block adaptive SVD filter described in co-pending U.S. Patent Application Publication No. US2019 / 0053780 (incorporated herein by reference in its entirety) can be used to implement SVD-based filtering to extract microbubble signals. As another example, an accelerated SVD filter described in co-pending U.S. Patent Application Publication No. US2018 / 0220997 (incorporated herein by reference in its entirety) can be used to implement SVD-based filtering to extract microbubble signals.

[0056] Optionally, the isolated microbubble signal may then be denoised, as indicated in step 108. The microbubble microvascular imaging technique described in the present disclosure is based on a power-based cross-correlation map between the microbubble signal and the PSF. Thus, the noise in each frame of the microbubble signal is removed, which may help reduce the noise in the cross-correlation calculation and thereby improve the robustness of the microvascular imaging.

[0057] Typically, noise has similar characteristics to microbubble signals, and distinguishing the two can be a challenge as noise becomes stronger and microbubble signals become weaker in deeper areas of tissue. As a result, noise signals may be mislabeled as microbubble signals, which leads to inaccurate microbubble delineation and velocity estimation.

[0058] As an example, denoising can be achieved using intensity-based thresholding methods. Such methods are more accurate when it can be assumed that the microbubble signal is stronger than the background noise signal. For example, by suppressing pixels with intensity values ​​less than a selected value (e.g., -30 dB to the maximum intensity value in the current field of view), a large amount of background noise can be suppressed. However, these methods may not be as accurate in areas where the microbubble signal is similar to the noise (e.g., deep areas of tissue). Moreover, the threshold must be carefully selected to avoid erroneously rejecting the microbubble signal or retaining too much noise.

[0059] As another example, the microbubble signal can be denoised based at least in part on the spatiotemporal information contained in the microbubble signal. Since microbubbles move with the blood flow, the microbubble movement is a deterministic event that can be continuously tracked in multiple acquisition frames, while the noise event is random and does not show any similar track characteristics in multiple acquisition frames. These differences between microbubbles and noise can be exploited in the spatiotemporal domain for robust noise suppression. As an example, a non-local means ("NLM") denoising filter can be applied to the raw, noisy microbubble data.

[0060] Another advantage of the spatiotemporal denoising filter described above is that because denoising is performed in the spatiotemporal domain, there is little spatial blurring of the underlying microbubble signal. Other denoising methods (e.g., convolutional Gaussian smoothing, Gaussian spectral apodization, microwave thresholding, or iterative total variation ("TV") minimization) may also be used in the spatiotemporal domain to achieve similar denoising effects. In some implementations, axial-time microbubble signal data may be used for denoising, while in other implementations, transverse-time data or full axial-transverse-time 3D data may also be used for denoising.

[0061] After the microbubble signal data is denoised, a microvascular image can be generated based on the power-based cross-correlation technique detailed in the present disclosure. As indicated in step 118, the microvascular image can be displayed to the user or stored for later use, such as for later analysis. In some implementations, microvascular morphological measurements (e.g., vascular density and vascular tortuosity) can be estimated from the microvascular image. As another example, microvascular hemodynamic measurements (e.g., blood flow velocity and blood flow) can be estimated from the microvascular image. For example, the microvascular image can be superimposed or presented side by side with a B-mode image of the target tissue. Alternatively, a microvascular blood flow velocity image with color-coded flow direction can also be superimposed or presented side by side with the B-mode image of the target tissue. In such an implementation, multiple flow direction tones can be used to indicate more than two flow directions.

[0062] Conventional microbubble localization may be performed at step 110, and may include identifying the location of the microbubble in each time frame of the microbubble signal data. For example, the center position of each isolated microbubble signal is located so that the movement of the microbubble can be tracked over time. The located center position of the microbubble can also be used to construct a super-resolution microvascular image, and to track the movement of the microbubble to calculate hemodynamic measurements, such as blood flow velocity. Conventional super-resolution imaging methods rely on the location of the center position of the microbubble.

[0063] The present disclosure provides a system and method for generating an image directly using correlation mapping, by using MB signals and the Nth power of PSF to improve the resolution of the cross-correlation map (which may not use the center position of the microbubble), the image has significantly improved spatial and temporal resolution. Figure 3 , a flowchart illustrating the steps of a non-limiting example of such a cross-correlation-based method is shown. Ultrasound data may be accessed, such as through image archiving, or may be acquired from a target tissue at step 310 by administering microbubbles (MB) or other suitable ultrasound contrast agents. By processing the ultrasound data at step 320, MB signal detection may be used to separate the MB signal from the tissue signal, such as by using tissue clutter filtering techniques, etc. Preprocessing may be used to improve the signal-to-noise ratio (SNR) and prepare data for high-resolution imaging at step 330. The PSF may be determined at step 340, and an MB image may be generated. The Nth power of the MB image and the PSF may be determined and used at step 350 to improve the resolution of the cross-correlation map, where N is the order of the power. The Nth power of the MB image and the PSF may determine the clarity of the final microvascular image. At step 360, a cross-correlation map between each MB image of the system and the PSF may be generated. At step 370, the accumulation of the power-based cross-correlation map for a selected number of frames may generate a high-resolution, high-contrast image of the microvascular system. Individual MB trajectories on the spatiotemporal cross-correlation map may be utilized to provide MB movement speed and direction estimates, which may be used in step 380 to image flow hemodynamics.

[0064] Figure 3 The method can be performed in a block-by-block manner. This method provides that microvascular image processing can be performed separately on smaller data subsets, and the combination of all the subsets can generate a microvascular image with improved performance.

[0065] Still reference Figure 3In step 310, ultrasound data may be accessed or acquired from a region of interest (ROI) in the target tissue by the administered ultrasound contrast agent. Any suitable contrast agent may be used, including but not limited to MB, phase change droplets, nanodroplets, gas vesicles, etc. The ultrasound data may be in any suitable format, including but not limited to ultrasound RF data, ultrasound IQ data, ultrasound envelope data, ultrasound B-mode data, etc. The ultrasound data may be acquired by using any suitable detection sequence, including but not limited to focused beam line-by-line imaging, wide beam multi-line acquisition imaging, compound plane wave imaging, compound diverging wave imaging, synthetic aperture imaging, etc. The ultrasound data may be acquired based on linear ultrasound components (such as basic imaging), or may be acquired based on nonlinear ultrasound components (such as pulse inversion ("PI"), amplitude modulation ("AM"), and pulse inversion amplitude modulation ("PIAM") imaging, or a combination thereof). In general, the acquired ultrasound data may include one or more spatial dimensions, which may include a transverse dimension, an axial dimension, an elevation dimension, and a combination thereof. The ultrasound data may also include a time dimension, such as a slow time dimension (i.e., a time direction along which a plurality of ultrasound frames or ultrasound volumes are collected).

[0066] By extracting the signal representing blood flow information and removing the background signal representing tissue clutter and noise, Figure 3 MB signal detection and separation of MB and tissue signals in step 320. Through this signal separation, blood flow imaging can be enhanced. Generally, MB flows in blood vessels with blood cells and is different from tissue signals in terms of spatiotemporal characteristics, so suitable clutter filtering techniques for blood flow imaging can be applied to extract MB signals. In a non-limiting example, the MB signal can be acquired using frame-to-frame signal subtraction, high-pass tissue clutter filtering, SVD-based tissue clutter filtering, regression-based tissue clutter filtering, intrinsic-based filtering, and the like. For certain implementations, such as CEUS (wherein the signal from the MB can be detected by utilizing the nonlinear response of the MB and the tissue clutter can be suppressed), clutter filtering processing may be unnecessary.

[0067] A high signal-to-noise ratio (SNR) of the acquired MB signal provides for optimized microvascular image generation. A low SNR may result in false detection of MBs and, thus, artifacts in the final microvascular image. The extracted MB data can be Figure 3The step 330 in the above method is preprocessed, and the preprocessing may include denoising, enhancing, equalizing, etc. of the MB signal. In a non-limiting example, an intensity threshold may be set so that a signal below the set threshold may be identified as noise and removed from the MB data, thereby providing enhanced MB data with suppressed noise background. A fixed threshold or an adaptive threshold may be used. In another non-limiting example, the SNR of the MB data may be improved by denoising based on the spatiotemporal characteristics of the data. Since the MB usually moves consistently over multiple ultrasound frames, which is different from the behavior of random noise, any denoising filter operating on spatiotemporal data may be applied. The denoising filter may include, but is not limited to: a non-local mean (NLM) denoising filter, any form of spatiotemporal low-pass filter, a smoothing filter, a median filter, a Savitzky-Golay filter, a non-local mean filter, an amplitude threshold, etc. Due to the influence of time gain compensation (TGC), tissue attenuation, and beamforming processes, the ultrasound signal intensity may be spatially dependent, so intensity equalization may also be used to equalize the spatially dependent intensity of the MB signal for subsequent processing. Non-limiting examples of such techniques, such as using noise contours to equalize MB signals, can be found in U.S. Patent Application Publication No. US2019 / 0053780, which is incorporated herein by reference.

[0068] In another non-limiting example, the noise floor of the ultrasound image can be measured by the ultrasound system and used as a spatially varying threshold for suppressing noise in the MB image. The background noise is typically spatially varying, with deeper regions having higher noise levels due to the higher TGC used. In certain configurations, a spatially varying threshold corresponding to the spatially varying noise level can be used and provides effective noise suppression. The noise floor of the ultrasound system can be used as a spatially dependent threshold so that pixels in the MB image that are below the threshold are identified as noise and removed from the image. A scaling factor can be applied to the noise floor image to control the spatially varying threshold, such as:

[0069]

[0070] Among them, MB 去噪 (x,y) is the denoised MB image, MB 原始(x,y) is the original MB image, and N(x,y) is the spatially varying noise floor. α is a scaling factor that determines the overall level of the threshold, and α can be any positive value. If the background noise follows a normal distribution, and the noise floor is estimated as the standard deviation of the noise, then by setting α=1, 68.2% of the noisy pixels can be removed; and by setting α=2, 95.4% of the noisy pixels can be removed; and when α=3, 99.8% of the noisy pixels can be removed. In practice, a larger scaling factor can be more beneficial to suppress more noise and retain MB signals with higher confidence, but it comes at the expense of removing weak MB signals.

[0071] refer to Figures 12A-12D , shows a non-limiting example of a spatially dependent threshold based on noise floor estimation. The noise floor can be measured or estimated, such as by methods previously proposed in U.S. Patent Application Publication No. US2019 / 0053780, which is incorporated herein by reference in its entirety.

[0072] In a non-limiting example, noise data can be acquired by turning off ultrasound transmission (e.g., setting the acoustic output to zero) and only receiving the same system configuration and imaging sequence as used for actual blood flow signal acquisition (including the same transducer, TGC gain settings, receive filters, beamforming settings, etc.).

[0073] In a non-limiting example, when shutting off the acoustic output is not possible or desirable for some ultrasound systems, noise data can be collected by minimizing the acoustic output and transmitting into the air. Through clutter filtering, strong echoes reflected from the air can be removed and pure noise data can be extracted. Any suitable clutter filter can be used to remove strong echoes from the air, such as high-pass tissue clutter filtering, SVD-based tissue clutter filtering, regression-based tissue clutter filtering, eigen-based filtering, etc. The noise floor can be estimated from the collected noise data as a measure of noise variance, noise standard deviation, noise mean, or other statistical characteristics in the time direction.

[0074] refer to Fig.12E and 12F , shows a non-limiting example of residual noise suppression after noise has been suppressed by any of the above denoising methods, where residual noise may still be present in the MB image. Given that residual noise is typically randomly distributed in the MB image and has a small spatial size, while the true MB signal typically has a more consistent and larger size, a thresholding process based on the size of isolated objects in each MB image can be applied to remove objects having a size less than a given size threshold. It is shown that before object size thresholding ( Fig.12E ) and after object size thresholding ( Fig.12F) is a non-limiting example of an MB image of FIG. 1 , which reflects better suppression of residual noise. In order to optimally suppress background noise for microvascular imaging, one or more of the above-mentioned denoising methods can be applied alone or in combination. The noise suppression method can also be applied to any traditional localization and tracking-based super-resolution method to better provide denoising.

[0075] The correlation or convolution of each frame of MB data with the determined point spread function (PSF) can then be calculated to generate a cross-correlation coefficient diagram, such as Figure 3 As shown in step 360. Figure 3 The PSF determined in step 340 may depend on the imaging settings, ultrasound system, center frequency, ultrasound pulse length, etc., and may be generated in any suitable way. In some configurations, different time frames of the cross-correlogram may be used with the velocity estimation method, as described herein.

[0076] Determining the PSF may include retrieving a previously calculated or estimated PSF from a memory or other data storage device, or may include calculating or estimating the PSF on demand.

[0077] In one non-limiting example, the PSF can be obtained from a simulation. For example, the PSF can be simulated based on a multivariate Gaussian distribution or other possible theoretical models. Such a distribution is useful because it can account for ultrasound lateral resolution, which is typically coarser than axial resolution; however, other distributions can also be used to simulate the PSF.

[0078] As another non-limiting example, the PSF can be obtained from experimental measurements of very small point objects (such as objects much smaller than the ultrasound wave). For example, the signal measured from microbubbles can be used as an approximation of the PSF of the ultrasound system. The PSF can be experimentally measured from a point target using the same ultrasound system and imaging setup.

[0079] In another non-limiting example, the PSF can be obtained from isolated individual MBs derived from the same MB data set for microvascular imaging. It is also possible to select and combine multiple individual MB signals to generate a smoother and more accurate PSF. The PSF can be in any suitable data format corresponding to the MB data used, such as RF data, ultrasound IQ data, ultrasound envelope data, ultrasound B-mode data, etc. Normalized or non-normalized correlation calculations can be applied to generate a cross-correlation map. The spatially interpolated version of the MB data can be correlated or convolved to obtain a finer sample size. For each MB frame, the true MB signal may present a higher correlation coefficient, while the noise background or irrelevant signal may have a lower correlation coefficient in the cross-correlation map. Therefore, the cross-correlation map can provide information about the presence / absence and location of the MB.

[0080] To improve the resolution of the cross-correlogram, before calculating the cross-correlogram, Figure 3 Step 350 in determines the Nth power of the MB data and the Nth power of the PSF. Here, N can be any positive value; in a non-limiting example, N can be 1, 2, 3, 4, ..., and so on. When N = 1, the 1st power of the data is the same as the original data. One advantage of performing a correlation between the Nth power of the MB data and the Nth power of the corresponding PSF is that a clearer MB target can be achieved for each MB in the cross-correlation diagram. In some configurations, the power of the MB data may be the same or different from the power of the PSF. In a non-limiting example according to the present disclosure, these two powers are set to the same value in the examples given below.

[0081] refer to Figure 4A , the resolution improvement of the Nth power method can be explained by a non-limiting example 1-dimensional model, where, assuming that the original MB signal 410 is a 1-dimensional Gaussian distribution, the PSF is also a 1-dimensional Gaussian distribution. The cross-correlation or convolution of the Gaussian distribution (MB signal) with the Gaussian distribution (PSF) is also a Gaussian distribution, but it has a larger width. Directly applying the correlation calculation to the original MB data will generate a cross-correlation graph with a worse resolution than the original MB signal, such as the first-order correlation 420. In contrast, the cross-correlation of the Nth power of the Gaussian distribution can provide a steeper distribution with a smaller full width at half maximum (FWHM). As indicated by the non-limiting example figures, as the power order N increases, the cross-correlation distribution gradually becomes steeper, as shown in the 4th order correlation 430, the 6th order correlation 440, the 10th order correlation 450 and the 20th order correlation 460.

[0082] refer to Figure 4B , shows a non-limiting example FWHM graph with a quantitative relationship between the FWHM of the cross-correlation distribution and the power order N. The original line 510 indicates the FWHM of the original MB distribution, while the cross-correlation line 520 indicates the FWHM of the resulting cross-correlation distribution under different powers N. When N>2, the cross-correlation distribution starts to become steeper than the original MB distribution. For a Gaussian model, the FWHM can be improved by a factor of 1 / √N. The method of the present disclosure can be applied to two-dimensional, three-dimensional, or even higher-dimensional data.

[0083] refer to Figure 5 In some configurations, the power of MB data may be different from the power of PSF. Figure 5 A non-limiting example of resolution improvement of a power-based approach using different power orders for MB data and PSF is depicted. As indicated in the graph of this non-limiting example, as the power order increases, the cross-correlation distribution becomes gradually steeper without requiring the order of the MB data and PSF to be the same.

[0084] refer to Figures 6A-6D , showing images resulting from non-limiting examples of different degrees of the N-Power method. Fig. 6A The original two-dimensional MB signal is shown in FIG. The corresponding cross-correlation diagram obtained by using the original MB signal to the first power is shown in FIG. Figure 6B As shown in , the corresponding cross-correlation diagram obtained using the fourth power is as follows Figure 6C As shown in , and the corresponding cross-correlation diagram obtained with the 8th power is shown in Fig.6D As shown in . The higher the power order, the smaller the target distribution of each MB, and thus a steeper cross-correlation graph can be obtained. In traditional super-resolution imaging, only a single point is marked at the center of each MB, which leads to Fig. 6A The example shown in has a very sparse image with a small number of non-zero pixels. To obtain a visually smooth traditional super-resolution image, millions of MB may need to be accumulated, which may require a very long acquisition time. The increase in resolution of this traditional method comes at the expense of a very long acquisition time.

[0085] In contrast, the power correlation method according to the present disclosure can change the resolution by the selection of the power. Each sharpened MB in the power correlation map can still contain many high-resolution pixels. In the accumulation step, only fewer MBs are needed to form a visually smooth high-resolution microvascular image. The data acquisition time can be reduced, and high spatial resolution can be achieved without long acquisition time.

[0086] refer to Figures 7A-7D , shows a non-limiting example threshold-sharpened image of the cross-correlation. In some configurations, to further sharpen the image, thresholding can be applied to the generated cross-correlation map, thereby rejecting correlation coefficient values ​​below a selected threshold. Using the thresholding method, for each individual MB, a small target can be generated in the cross-correlation map. Figures 7A-7D A non-limiting example of thresholding a cross-correlation map is shown in , where a larger threshold generates smaller MB objects. The threshold can thus control the final resolution of the microvascular image.

[0087] In some configurations, a denoising process may be applied to remove suspicious targets in the cross-correlogram. True MB signals typically have a well-formed Gaussian shape, which can be exploited because noise or irrelevant signals may appear as irregular shapes in cross-correlograms with smaller target sizes. In this way, these unwanted signals may be identified and removed from the correlation map based on target shape or target size. In a non-limiting example, thresholding may be applied to remove objects with a size smaller than a given threshold, retaining objects with larger sizes in the cross-correlogram, thereby increasing the confidence and reliability of MB detection.

[0088] refer to Figure 8, shows a non-limiting example, where according to Figure 3 Step 370 in generates a high-resolution and high-contrast microvascular image by accumulating the cross-correlograms of a selected number of frames. A conventional power Doppler blood flow image 820 can be generated by directly accumulating the raw MB signals of a selected number of frames 810, and the resolution of this image may be limited by the ultrasound wavelength and imaging settings. In a non-limiting example, the accumulation can be performed on the basis of a power-based cross-correlogram 830, which can generate a microvascular image 840 with a resolution exceeding the imaging system limit. The resolution of the final microvascular system image can be controlled by the power order and thresholding of the above-mentioned cross-correlogram. In some configurations, a user interface can be provided to the user so that the user can select or adjust the final resolution to meet the needs of clinical applications.

[0089] For in vivo applications, tissue motion or ultrasound probe motion during scanning may induce blur in the final microvascular image, thereby degrading the performance of the method. In some configurations, motion registration can be introduced to remove tissue motion and thus avoid blur. Image registration can be performed based on motion estimation from the original acquired ultrasound data and / or clutter-filtered MB data. Any suitable image registration algorithm can be applied, including but not limited to global or local cross-correlation methods, methods based on global or local phase correlation, global or local optical flow methods, etc.

[0090] refer to Fig. 9 , shows two non-limiting example images of in vivo microvascular images obtained by the method of the present disclosure using a data length of about 2 seconds compared to conventional contrast-enhanced blood flow images. The figure shows the improvement of pancreatic microvascular image 920 relative to conventional pancreatic blood flow image 910, and the improvement of liver microvascular image 940 relative to conventional liver blood flow image 930.

[0091] In some configurations, additional quality control may be applied prior to accumulation to suppress false MB signals and noise. True MB targets may move steadily over multiple adjacent frames, thereby generating continuous MB tracks in a three-dimensional spatiotemporal matrix of a correlation graph, such as the matrix shown in image 830. In contrast, random noise or irrelevant signals may not have continuous motion and thus appear as isolated targets rather than long tracks in the spatiotemporal matrix. The longer the track, the higher the confidence that the signal is a valid MB. Therefore, the three-dimensional spatiotemporal matrix may be thresholded to eliminate targets whose size or length is below a selected threshold to improve the confidence and reliability of MB detection for microvascular imaging.

[0092] refer to Fig.10, a non-limiting example of a quality control adjustment image based on target size thresholding is shown. The improved noise suppression for high-definition images can be viewed as removing objects that are smaller in size than a selected number of pixels. Quality control can also be performed in the resulting accumulated two-dimensional microvascular image. In one non-limiting example, a two-dimensional accumulated image can be generated from a specific number of correlation maps (such as 10 frames of correlation maps), and thresholding can also be applied to remove small targets and maintain MB tracks in the two-dimensional microvascular image. In another non-limiting example, the same process can be repeated, such as applied to the next 10 frames of correlation maps with or without overlap. The final microvascular image can be obtained by combining or accumulating all of these individual thresholded microvascular images.

[0093] refer to Figures 11A-11E ,exist Fig.11A A non-limiting example of MB trajectory in a three-dimensional space-time space is shown in FIG. 1 , which has two spatial dimensions and one temporal dimension. Each isolated MB trajectory in the three-dimensional space-time space represents a spatial movement trace of the MB over time. In some configurations, the MB movement trajectory analysis can be used to measure the flow speed and movement direction of each MB based on the three-dimensional space-time matrix, such as Figure 3 As indicated in step 380 of , a high-resolution image of flow hemodynamics (eg, a flow hemodynamic image) such as flow velocity, flow direction, flow vector, etc. may be generated.

[0094] In some configurations, prior to velocity estimation, a quality control process may be applied to the three-dimensional spatiotemporal space to remove shorter or smaller trajectories and retain only those that are longer or larger to improve the robustness of the velocity estimate. Fig. 11B A schematic diagram showing an example of a MB trajectory in three-dimensional spatiotemporal space. Fig. 11B In FIG. 1 , the MB moves from the first frame 1115 to the Nth frame 1125, and each ellipse 1110 represents the MB in different frames of the cross-correlogram. These MB ellipses 1110 form a continuous track in the spatiotemporal space. The orientation and length of the track can be determined by the MB movement speed and direction. By measuring the orientation and length of the track, the MB flow rate can be accurately estimated. Any suitable method can be used to measure the orientation and length of the track.

[0095] In a non-limiting example, for each MB trajectory, a fit may be applied to all discrete samples / pixels in the trajectory in the three-dimensional spatiotemporal space to measure the MB movement speed. Any fit may be applied, including but not limited to linear fit, linear regression, spline fit, cubic fit, hyperbolic fit, etc. Fig. 11CSamples / pixels of an example MB trajectory in three-dimensional spatiotemporal space are shown, and a linear fit of all the trajectory samples can provide a quantitative measure of the velocity vector (including magnitude and direction). In a non-limiting example, the slope of the fitted line with respect to the time direction can give an estimate of the magnitude of the MB velocity, while the angle of the fitted line in the spatial domain gives information on the direction of MB movement.

[0096] In some configurations, the MB trajectory samples may be projected onto a two-dimensional axial-temporal plane, such as Fig.11D As indicated, it can also be projected onto a two-dimensional transverse-temporal plane, such as Fig.11E The projections can be performed separately and the fits made to the samples in these two-dimensional planes provide a measure of the velocity in the axial and transverse directions respectively. Fig.11D and Fig.11E The fitted line in represents a fitted line based on the MB trajectory samples. The slope of the fitted line gives an estimate of the MB speed in both directions. In a non-limiting example to further improve the robustness of the speed estimate, a weighted fit can be applied to the discrete samples / pixels of the MB trajectory. Any suitable weighted fit can be used, including but not limited to weighted linear fit, weighted linear regression, weighted spline fit, weighted cubic fit, hyperbolic fit, etc. Any suitable weights can be applied to the trajectory samples. In a non-limiting example, the trajectory samples can be weighted by the cross-correlation coefficient of the trajectory samples themselves, or can also be weighted by the kth power of the cross-correlation coefficient, where k can be any arbitrary value.

[0097] For each MB track, the fitting correlation coefficient (R), determination coefficient (R2), mean, standard deviation or variance of the fitting error, or other statistical features can be used as indicators of fitting robustness. Additional quality control can be applied to the velocity estimate by removing those MB tracks with low fitting robustness (indicated by low correlation coefficient (R), low determination coefficient (R2), or high fitting error, etc.). Determining the desired fitting robustness can include establishing a threshold and removing MB tracks with low fitting robustness, low correlation coefficient (R), low determination coefficient (R2), or high fitting error that do not meet the established threshold. In some configurations, the measured velocity information (such as the magnitude or direction of the velocity) for each track (corresponding to one MB) can be stored in a spatiotemporal space, and can then be used to generate a final microvascular velocity image by combining the velocity values ​​of all MB tracks passing through the same microvessel.

[0098] In some configurations, the measured velocity information of each trajectory can also be assigned to a contracted trajectory in the spatiotemporal space, which can further improve the spatial resolution of the final microvascular velocity image. In some configurations, the final microvascular velocity image can be generated by simply averaging or weighted averaging all velocity trajectories along the time direction in the spatiotemporal space. For weighted averaging, the velocity trajectories can be weighted with any possible weights, including but not limited to the cross-correlation coefficient of the trajectory samples themselves, the fitting correlation coefficient (R), the determination coefficient (R2), the mean, standard deviation or variance of the fitting error, etc. In a non-limiting example, microvascular morphological measurements (e.g., vascular density and vascular tortuosity) and microvascular hemodynamic measurements (e.g., blood flow velocity and blood flow) can be estimated from the microvascular image. The microvascular image can be superimposed, transparently superimposed, or presented side by side with the B-mode image of the target tissue. In another non-limiting example, the microvascular velocity image with or without color-coded flow direction can also be superimposed, transparently superimposed, or presented side by side with the microvascular morphological image or with the B-mode image of the target tissue. Multiple flow direction hues may be used to indicate more than two flow directions.

[0099] In another non-limiting example, each MB trajectory in the three-dimensional spatiotemporal space can be projected into the two-dimensional spatial domain, and the projected area of ​​the MB trajectory can be used to estimate the moving speed and direction of the MB. MBs with higher moving speeds may have larger projected areas, and vice versa. Calculation of the projected area of ​​each isolated MB trajectory can give an estimate of the MB speed. In some configurations, the projected area (A1) of the trajectory can be compared with the area of ​​each MB (A0, e.g. Fig. 11B The ratio of A1 / A0 can provide an estimation of the moving speed of the MB.

[0100] In a non-limiting example, a fitting line 1120 may be applied to the feature points of each MB ellipse 1110, and the length and angle of the fitting line 1120 may be obtained for velocity estimation. Any fitting may be applied, including but not limited to linear fitting, spline fitting, cubic fitting, hyperbolic fitting, etc. Any suitable feature point of each MB ellipse may be used in the present disclosure, including but not limited to the mean center of the MB ellipse, the weighted average of the MB ellipse, the maximum cross-correlation position within the MB ellipse, the focus of the MB ellipse, the edge of the MB ellipse, etc.

[0101] In some configurations, the slope of the fitted line with respect to the time direction can give an estimate of the magnitude of the MB velocity, and the angle of the fitted line in the spatial domain gives information about the direction of MB movement. In some configurations, the MB feature points can be first projected onto a two-dimensional axial-time plane and a two-dimensional transverse-time plane, respectively, and the fitting in the two-dimensional planes provides measurements of the velocity along the axial and transverse directions, respectively.

[0102] In order to further improve the robustness of the velocity estimation, a weighted fit of the feature points may be applied. Any suitable weighted fit may be used, including but not limited to weighted linear fit, weighted linear regression, weighted spline fit, weighted cubic fit, hyperbolic fit, and the like. Any suitable weights may be applied to the feature points. For each MB trajectory, the fitting correlation coefficient (R), the determination coefficient (R2), the mean, standard deviation or variance of the fitting error, or other statistical features may be used as indicators of the robustness of the fit. Additional quality control may be applied to the velocity estimation by removing those MB trajectories with low fitting robustness indicated by low correlation coefficient (R), low determination coefficient (R2), or high fitting error. In a non-limiting example, the distance and angle between the feature points of the first and last MB ellipses 1110 in the trajectory may be used to roughly estimate the length and direction of the trajectory. In this way, only the feature points of the first and last MB ellipses 1110 need to be calculated, which may reduce the computational burden.

[0103] Fig.11F Non-limiting example steps of the velocity estimation method are shown in the flowchart depicted in . In step 1150, the microbubble trajectories in the spatiotemporal space are accessed, such as from an image storage archive, or obtained from image data. Individual microbubble trajectories can be isolated in step 1155. Fit lines can be applied to the microbubble trajectories in step 1160. In some configurations, the fit line can be applied to more than one projection plane, such as the transverse or axial projection plane described above. The orientation and length of the selected trajectory can be determined in step 1165. Optional quality control can be performed in step 1170 by removing shorter trajectories or those that fit poorly to the fit line, thereby improving the robustness of the velocity estimation. The flow velocity of the microbubble can be estimated based on the determined orientation and length of the selected trajectory. Then, microvascular images and flow hemodynamic images can be generated in steps 1180 and 1185.

[0104] In some configurations, full field of view (FOV) data may be used. However, in some configurations, it may be advantageous to spatially divide the FOV data into multiple data subsets with or without spatial overlap, and apply the high-definition microvascular imaging method provided in the present disclosure to each subset separately. By combining all individual microvascular images from all data subsets together, a final high-resolution microvascular image can be obtained. Any suitable data separation method can be used, and any suitable weighting function can be applied. In a non-limiting example, the original data can be spatially divided into smaller blocks with or without spatial overlap. In a non-limiting example, the original data can be divided into smaller blocks with spatial overlap, and each data subset is spatially weighted by a weighting function, such as the data separation method disclosed in U.S. Provisional Patent Application No. 62 / 975,515, which is hereby incorporated by reference herein.

[0105] The data separation into multiple subsets can be performed on the original data before tissue clutter filtering, and the above data separation can also be performed on the MB data after tissue clutter filtering, and / or a combination of the two. When using the original data, the MB signal detection process, such as tissue clutter filtering, can be applied to each data subset separately. By dividing the original data into multiple subsets, an adaptive pre-processing unit can be applied, and the adaptive pre-processing unit can be specific to each subset. In a non-limiting example, an adaptive intensity threshold can be used to identify signals below a given threshold as noise and remove them from each subset of MB data. The threshold of each data subset can be adaptively determined based on the local statistics of the MB data. In some configurations, the threshold can be adaptively estimated based on the histogram of each data subset. In some configurations, the threshold can be adaptively determined based on the total energy of each subset of MB data.

[0106] In some configurations, the PSF may be spatially varying, such as being different from one spatial location to another due to ultrasound focusing, beamforming aperture, etc. Different PSFs may be used for different data subsets to compute spatially specific cross-correlograms. Using a specific PSF for each data subset may be beneficial for resolution improvement and optimization of each located data set. Spatially different PSFs may be estimated by model fitting, experimental measurement, direct derivation from the same MB data set, or a combination of these methods.

[0107] The present disclosure describes a method for MB imaging, but those skilled in the art will appreciate that the method can also be applied to ultrasound imaging using any other type of ultrasound contrast agent, such as phase change droplets, nanodroplets, etc., or can also be applied to non-contrast ultrasound imaging without contrast agents. The method in the present disclosure discloses that it is applicable not only to two-dimensional imaging, but also to three-dimensional (3-D) and even higher-dimensional imaging. In a non-limiting example, the method can be easily extended to three dimensions, where a three-dimensional PSF can be generated for cross-correlation calculation in three dimensions. The method proposed in the present disclosure can be combined with the MB signal separation disclosed in U.S. Provisional Patent Application No. 62 / 861,580, which is hereby incorporated by reference herein, where the original MB data can be divided into multiple subsets, each with a sparser MB concentration to allow better imaging performance. The power-based cross-correlation method proposed in the present disclosure can also be combined with a super-resolution method, such as a localization and tracking-based super-resolution method (U.S. Patent Application No. 16 / 617,628), to improve the performance (accuracy and precision) of MB localization on the cross-correlation map.

[0108] Fig.13 An example of an ultrasound system 1300 that can implement the methods described in the present disclosure is shown. The ultrasound system 1300 includes a transducer array 1302, which includes a plurality of individually driven transducer elements 1304. The transducer array 1302 may include any suitable ultrasound transducer array, including a linear array, a curved array, a phased array, etc. Similarly, the transducer array 1302 may include a 1D transducer, a 1.5D transducer, a 1.75D transducer, a 2D transducer, a 3D transducer, etc.

[0109] When excited by the transmitter 1306, a given transducer element 1304 generates a burst of ultrasonic energy. The ultrasonic energy (e.g., echoes) reflected from the object or subject under study back to the transducer array 1302 is converted into electrical signals (e.g., echo signals) by each transducer element 1304 and can be applied to a receiver 1308 separately through a set of switches 1310. The transmitter 1306, receiver 1308, and switches 1310 operate under the control of a controller 1312, which may include one or more processors. As an example, the controller 1312 may include a computer system.

[0110] The transmitter 1306 can be programmed to transmit unfocused or focused ultrasound waves. In some configurations, the transmitter 1306 can also be programmed to transmit diverging waves, spherical waves, cylindrical waves, plane waves, or a combination thereof. In addition, the transmitter 1306 can be programmed to transmit spatially or temporally encoded pulses.

[0111] Receiver 1308 may be programmed to implement an appropriate detection sequence for the imaging task at hand. In some embodiments, the detection sequence may include one or more of progressive scanning, compound plane wave imaging, synthetic aperture imaging, and compound diverging beam imaging.

[0112] In some configurations, the transmitter 1306 and the receiver 1308 may be programmed to achieve a high frame rate. For example, a frame rate associated with an acquisition pulse repetition frequency ("PRF") of at least 100 Hz may be achieved. In some configurations, the ultrasound system 1300 may sample and store at least one hundred sets of echo signals in the time direction.

[0113] Scanning may be performed by setting switches 1310 to their transmit positions, thereby directing transmitters 1306 to momentarily turn on to excite transducer elements 1304 during a single transmit event according to a selected imaging sequence. Switches 1310 may then be set to their receive positions, and subsequent echo signals generated by transducer elements 1304 in response to one or more detected echoes are measured and applied to receiver 1308. The individual separate echo signals from transducer elements 1304 may be combined in receiver 1308 to produce a single echo signal.

[0114] The echo signal is transmitted to a processing unit 1314 to process the echo signal or an image generated from the echo signal. The processing unit 1314 may be implemented by a hardware processor and a memory. As an example, the processing unit 1314 may generate an image by using the method described in the present disclosure. The image generated from the echo signal by the processing unit 1314 may be displayed on a display system 1316.

[0115] Reference now Fig.14 , a block diagram of an example of a computer system 1400 that can perform the methods described in the present disclosure is shown. The computer system 1400 generally includes an input 1402, at least one hardware processor 1404, a memory 1406, and an output 1408. Therefore, the computer system 1400 is generally implemented with a hardware processor 1404 and a memory 1406.

[0116] In some embodiments, the computer system 1400 may be a controller or processing unit of an ultrasound system. In some examples, the computer system 1400 may also be implemented by a workstation, a laptop, a tablet device, a mobile device, a multimedia device, a network server, a mainframe, one or more controllers, one or more microcontrollers, or any other general or special computing device.

[0117] Computer system 1400 may operate autonomously or semi-autonomously, or may read executable software instructions from memory 1406 or a computer-readable medium (e.g., a hard drive, a CD-ROM, a flash memory), or may receive instructions from a user or any other source logically connected to the computer or device (such as another networked computer or server) via input 1402. Thus, in some embodiments, computer system 1400 may also include any suitable device for reading a computer-readable storage medium.

[0118] Typically, the computer system 1400 is programmed or otherwise configured to implement the methods and algorithms described in the present disclosure. For example, the computer system 1400 can be programmed to generate images according to the methods described in the present disclosure.

[0119] Input 1402 may take any suitable shape or form as desired for operation of computer system 1400, including the ability to select, enter, or otherwise specify parameters consistent with performing tasks, processing data, or operating computer system 1400. In some aspects, input 1402 may be configured to receive data, such as data acquired using an ultrasound system. Such data may be processed as described above to generate an image. In addition, input 1402 may also be configured to receive any other data or information deemed useful for generating an image using the methods described above.

[0120] In processing tasks for operating computer system 1400 , one or more hardware processors 1404 may also be configured to perform any number of post-processing steps on data received via input 1402 .

[0121] The memory 1406 may include software 1410 and data 1412, such as data acquired using an ultrasound system, and may be configured to store and retrieve processed information, instructions, and data to be processed by the one or more hardware processors 1404. In some aspects, the software 1410 may include instructions that direct the generation of images according to embodiments described in the present disclosure.

[0122] Furthermore, output 1408 may take any shape or form as desired, and may be configured to display images (eg, microvascular images) and other desired information.

[0123] In some embodiments, any suitable computer-readable medium may be used to store instructions for performing the functions and / or processes described herein. For example, in some embodiments, the computer-readable medium may be transient or non-transient. For example, non-transient computer-readable media may include media such as: magnetic media (e.g., hard disk, floppy disk), optical media (e.g., compact disk, digital video disk, Blu-ray disk), semiconductor media (e.g., random access memory ("RAM"), flash memory, electrically programmable read-only memory ("EPROM"), electrically erasable programmable read-only memory ("EEPROM")), any suitable medium that is not transient or does not have any persistent appearance during transmission, and / or any suitable tangible medium. As another example, transient computer-readable media may include signals on a network, in wires, conductors, optical fibers, circuits, or any suitable medium that is transient and does not have any persistent appearance during transmission, and / or any suitable intangible medium.

[0124] This disclosure has described one or more preferred embodiments, and it should be understood that many equivalents, alternatives, variations, and modifications besides those explicitly stated are possible and within the scope of the invention.

Claims

1. An ultrasound system for imaging microvessels with high spatial and temporal resolution, the system comprising: A computer system, the computer system being configured to: (a) accessing ultrasound data, the ultrasound data having been acquired using the ultrasound system from a region of interest within a subject, wherein a microbubble contrast agent was present in the region of interest when the ultrasound data was acquired; (b) generating microbubble signal data by separating the microbubble signal in the ultrasound data from other signals in the ultrasound data; (c) determining a point spread function (PSF) of the ultrasound system used to acquire the ultrasound data; (d) determining a first Nth power of the microbubble signal data and a second Nth power of the PSF to generate a cross-correlation plot between the microbubble signal data and the point spread function of the ultrasound system; (e) generating a microvascular image with high spatial and temporal resolution based at least in part on the generated cross-correlation map.

2. The system according to claim 1, characterized in that The N is greater than 2.

3. The system according to claim 1, characterized in that The first Nth power and the second Nth power are equal.

4. The system according to claim 3, characterized in that The first Nth power of the microbubble signal data and the second Nth power of the PSF determine the sharpness of the microvessel image.

5. The system according to claim 3, characterized in that The microbubble signal data and the PSF include a Gaussian distribution, and wherein a first Nth power of the microbubble signal data and a second Nth power of the PSF provide a steeper Gaussian distribution and a smaller full width at half maximum (FWHM) than the microbubble signal data.

6. The system according to claim 5, characterized in that The FWHM is improved by a factor of 1 / √N.

7. The system according to claim 1, characterized in that The PSF is a simulated PSF.

8. The system according to claim 7, characterized in that The simulated PSF is simulated at least in part based on a multivariate Gaussian distribution.

9. The system according to claim 1, characterized in that The PSF is estimated based on measurements acquired by imaging small point objects using the ultrasound system.

10. The system according to claim 1, wherein: The PSF is determined from isolated individual microbubbles derived from the microbubble signal data.

11. The system according to claim 1, characterized in that The computer system is further configured to sharpen the cross-correlogram.

12. The system according to claim 11, characterized in that The computer system is further configured to sharpen the cross-correlogram by applying a threshold to the cross-correlogram to reject correlation coefficient values ​​below the threshold.

13. The system of claim 1, wherein: The computer system is further configured to denoise the cross-correlogram to remove targets in the cross-correlogram.

14. The system of claim 1, wherein: The computer system is further configured to generate a flow hemodynamic image based on microbubble trajectories in the cross-correlation map.

15. The system of claim 14, wherein: The computer system is further configured to estimate microbubble flow velocity from the microbubble signal data to generate the flow hemodynamic image.

16. The system of claim 15, wherein: The computer system is further configured to isolate individual microbubble trajectories between frames of the cross-correlogram to estimate the microbubble flow rate.

17. The system of claim 16, wherein: The computer system is further configured to apply a fitted line to the respective microbubble trajectories and determine at least one of an orientation or a length of the isolated microbubble trajectories to estimate the microbubble flow rate.

18. The system of claim 1, wherein: The computer system is further configured to determine a noise floor of the ultrasound system and to suppress noise in the microbubble signal data using the determined noise floor as a spatially varying threshold.

19. The system of claim 18, wherein: The computer system is further configured to control the spatially varying threshold via a scaling factor applied to the noise floor of the ultrasound system.

20. The system of claim 18, wherein: The computer system is further configured to determine the noise floor of the ultrasound system by at least one of: receiving data with ultrasound transmission of the ultrasound system turned off, or filtering the received data with the ultrasound transmission minimized.

Images