A micro blood flow imaging method based on adaptive density filtering and related device

Abstract

The application discloses a micro blood flow imaging method based on adaptive density filtering and a related device, and relates to the technical field of ultrasonic imaging. The method comprises the following steps: extracting a static frame in original ultrasonic data, screening a high-quality static frame through correlation analysis to eliminate motion interference; performing SVD clutter filtering, and cutting off high spatiotemporal coherence singular values to separate tissue clutter and microbubble signals; performing adaptive density filtering and subtracting a preliminary filtering result to realize noise suppression and overlapping microbubble decoupling; and enhancing microbubble motion signals through time-lag autocorrelation processing, and finally completing power Doppler imaging and ultrasonic localization microscope reconstruction. The scheme effectively solves the problem that the prior art cannot simultaneously consider PDI noise suppression and ULM microbubble decoupling, significantly improves image signal-to-noise ratio and super-resolution, avoids problems such as signal distortion, artifacts and loss of low-speed blood flow, and provides more reliable imaging support for clinical diagnosis such as blood vessel lesion detection and tumor blood flow evaluation.

Description

Technical Field

[0001] This application relates to the field of ultrasound imaging technology, and in particular to a micro-blood flow imaging method and related apparatus based on adaptive density filtering. Background Technology

[0002] Pulsed Doppler Imaging (PDI) is a commonly used blood flow visualization technique in ultrasound medicine. It achieves blood flow distribution imaging by detecting the power of the Doppler signal of blood flow energy, offering advantages such as high sensitivity and slow blood flow detection. However, PDI images are susceptible to speckle noise and electronic noise. These noises originate from the coherence characteristics of ultrasound waves, equipment limitations, and scattering effects from biological tissues, leading to a reduced signal-to-noise ratio and blurred details. This affects the accuracy of clinical diagnoses such as vascular lesion detection and tumor blood flow assessment, potentially causing misdiagnosis or missed diagnosis.

[0003] Ultrasound Localization Microscopy (ULM) is a super-resolution imaging technique that breaks through the traditional ultrasound diffraction limit, acquiring high-resolution blood flow images by tracking injected microbubbles. Microbubbles, acting as ultrasound contrast agents, flow with the blood into tiny blood vessels, serving as localization sources. Key steps in ULM include separating microbubbles from tissue and noise, locating sub-pixel positions of microbubbles, and performing inter-frame trajectory tracking. However, in high-concentration microbubble environments, microbubbles are prone to overlap or aggregation, leading to localization errors and image distortion. Effective microbubble decoupling techniques are needed to separate individual microbubbles to ensure imaging accuracy.

[0004] Existing processing techniques suffer from numerous drawbacks: Nonlocal mean filtering is computationally complex, sensitive to motion, and highly dependent on parameters; noise equalization only improves visualization without truly improving the signal-to-noise ratio and is complex to implement; deconvolution is computationally intensive, highly dependent on the point spread function (PSF) model, and easily amplifies noise and produces artifacts; SVD clutter filtering thresholds are difficult to determine, sensitive to motion, and easily lose low-velocity blood flow signals; Fourier domain microbubble separation methods easily lead to microbubble signal distortion and have poor adaptability. None of these techniques can simultaneously and efficiently solve the dual challenges of PDI noise suppression and ULM microbubble decoupling, and they generally suffer from computational complexity and limited applicability. Summary of the Invention

[0005] The purpose of this application is to provide a micro-blood flow imaging method and related apparatus based on adaptive density filtering, which can significantly improve the image signal-to-noise ratio and super-resolution of micro-blood flow imaging.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] In a first aspect, this application provides a micro-blood flow imaging method based on adaptive density filtering, comprising the following steps:

[0008] For the raw ultrasound data acquired by the ultrasound acquisition device, still frames are extracted from the raw ultrasound data, and a number of high-quality still frames are selected by correlation analysis between the still frames and other frames. The raw ultrasound data includes ultrasound IQ data.

[0009] SVD clutter filtering was performed on the ultrasound IQ data corresponding to each high-quality still frame, and the first K high spatiotemporal coherence singular values ​​were truncated by singular value decomposition to obtain the microblood flow signal data after preliminary filtering.

[0010] Adaptive density filtering is performed on the microblood flow signal data after preliminary filtering, and the result of adaptive density filtering is subtracted from the microblood flow signal data after preliminary filtering to obtain the microblood flow signal data after complete filtering.

[0011] The fully filtered microblood flow signal data is subjected to time lag autocorrelation processing. The lag frame data is multiplied pixel by pixel to eliminate time-related noise and enhance the microbubble motion signal, thus obtaining the microblood flow signal data sequence.

[0012] Based on the microblood flow signal data sequence, power Doppler imaging and ultrasound-guided microscopy reconstruction were performed to obtain power Doppler images and super-resolution density images of the microblood flow signal.

[0013] Optionally, adaptive density filtering is performed on the pre-filtered microblood flow signal data, and the result of adaptive density filtering is subtracted from the pre-filtered microblood flow signal data to obtain the fully filtered microblood flow signal data, specifically including:

[0014] Spatial smoothing is performed on the pre-filtered micro-blood flow signal data to obtain smoothed micro-blood flow signal data.

[0015] For the smoothed micro-blood flow signal data, the spatial neighborhood search range is determined based on the full width at half maximum (FWHM) of the point spread function, and the local maximum intensity difference of pixels is calculated.

[0016] Based on the local maximum intensity difference of pixels, a joint weight matrix of spatial and intensity values ​​for the neighborhood of all pixels in the smoothed microblood flow signal data is constructed.

[0017] The density gradient is calculated based on the spatial and intensity joint weight matrix. The fully filtered microblood flow signal data is obtained by subtracting the density gradient from the initially filtered microblood flow signal data.

[0018] Secondly, this application provides a micro-blood flow imaging system based on adaptive density filtering, including the following functional modules:

[0019] The still frame filtering module is used to extract still frames from the raw ultrasound data acquired by the ultrasound acquisition device, and to filter out a number of high-quality still frames through correlation analysis between the still frames and other frames. The raw ultrasound data includes ultrasound IQ data.

[0020] The SVD clutter filtering module is used to perform SVD clutter filtering on the ultrasound IQ data corresponding to each high-quality still frame, and to truncate the first K high spatiotemporal coherence singular values ​​through singular value decomposition to obtain the microblood flow signal data after preliminary filtering.

[0021] The adaptive density filtering module is used to perform adaptive density filtering on the microblood flow signal data after preliminary filtering, and to subtract the result of the adaptive density filtering from the microblood flow signal data after preliminary filtering to obtain the microblood flow signal data after complete filtering.

[0022] The time-lag autocorrelation processing module is used to perform time-lag autocorrelation processing on the fully filtered microblood flow signal data, multiplying the lagged frame data pixel by pixel to eliminate time-related noise and enhance the microbubble motion signal, thereby obtaining the microblood flow signal data sequence.

[0023] The microblood flow imaging module is used to perform power Doppler imaging and ultrasound-guided microscopy reconstruction based on microblood flow signal data sequences to obtain power Doppler images and super-resolution density images of microblood flow signals.

[0024] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the micro-blood flow imaging method based on adaptive density filtering described above.

[0025] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the micro-blood flow imaging method based on adaptive density filtering described above.

[0026] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the micro-blood flow imaging method based on adaptive density filtering described above.

[0027] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0028] This application provides a micro-blood flow imaging method and related apparatus based on adaptive density filtering. In this method, still frames are first extracted from the original ultrasound data, and several high-quality still frames are selected based on cross-correlation analysis. This process removes unregistered data caused by respiration or other motion, providing stable, motion-artifact-free baseline data for subsequent processing and avoiding processing distortion caused by motion interference. Subsequently, SVD clutter filtering is applied to the ultrasound IQ data corresponding to each high-quality still frame, and the first K high spatiotemporal coherence singularities are truncated through singular value decomposition to obtain pre-filtered micro-blood flow signal data. This operation effectively separates tissue clutter from microbubble signals, avoiding the shortcomings of existing SVD clutter filtering methods, such as difficulty in determining the threshold and easy loss of low-velocity blood flow signals. Then, an adaptive density filtering algorithm is used for further filtering, simultaneously suppressing Gaussian noise and impulse noise. Overlapping microbubble decoupling, after subtraction, sharpens the core signal of microbubbles and separates overlapping regions, truly improving the signal-to-noise ratio and microbubble separation effect, with low computational complexity and no need for complex parameter tuning; then, time-lag autocorrelation processing is performed on the fully filtered microblood flow signal data to further suppress static clutter and temporal noise, amplify the phase changes of microbubble motion, significantly enhancing the recognizability of microbubble motion signals compared to existing technologies, providing clearer dynamic information for imaging; finally, power Doppler imaging and ultrasound-guided microscopy reconstruction are performed, resulting in a significant improvement in the signal-to-noise ratio of power Doppler images and effective suppression of background noise; the super-resolution density image resolution of ultrasound-guided microscopy is significantly improved, and the microbubble trajectory is clearer, successfully solving the dual problem of simultaneously and efficiently solving PDI noise suppression and ULM microbubble decoupling in existing technologies, with low computational complexity and wide applicability. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart of a microblood flow imaging method based on adaptive density filtering, provided as an embodiment of this application.

[0031] Figure 2 This is a flowchart of step A3 in a microblood flow imaging method based on adaptive density filtering provided in an embodiment of this application.

[0032] Figure 3 This is a schematic diagram of subpixel localization in a microblood flow imaging method based on adaptive density filtering, provided in an embodiment of this application.

[0033] Figure 4 This is a schematic diagram illustrating the selection of a region of interest in a microblood flow imaging method based on adaptive density filtering, as provided in an embodiment of this application.

[0034] Figure 5 This is a schematic diagram showing the comparison results of power Doppler imaging of rat liver data in a microblood flow imaging method based on adaptive density filtering, provided in another embodiment of this application.

[0035] Figure 6 This is a schematic diagram of the result of ULM reconstruction after extracting only still frames from rat liver data in a microblood flow imaging method based on adaptive density filtering, which is provided in another embodiment of this application.

[0036] Figure 7 This is a schematic diagram showing the comparative results of ULM reconstruction of rat brain data in a microblood flow imaging method based on adaptive density filtering, which is provided in another embodiment of this application.

[0037] Figure 8 This is a schematic diagram comparing the ULM direction of rat brain data in a microblood flow imaging method based on adaptive density filtering, provided in another embodiment of this application.

[0038] Figure 9 This is a schematic diagram comparing the correlation curves of the ULM Fourier ring of rat brain data in a microblood flow imaging method based on adaptive density filtering, provided in another embodiment of this application.

[0039] Figure 10 This is a schematic diagram showing the comparison results of various filtering algorithms in a microblood flow imaging method based on adaptive density filtering, provided in another embodiment of this application.

[0040] Figure 11 This is a schematic diagram of the functional modules of a microblood flow imaging system based on adaptive density filtering, provided in an embodiment of this application.

[0041] Figure 12 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0042] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] Power Doppler imaging (PDI) is a commonly used technique in ultrasound medicine. It visualizes blood flow distribution by detecting the Doppler signal power of blood flow energy, offering advantages such as high sensitivity and detection of slow blood flow. On the other hand, ultrasound localization microscopy (ULM) is an emerging super-resolution imaging technique that acquires high-resolution blood flow images by tracking injected microbubbles. ULM can overcome the diffraction limit of traditional ultrasound, but in high-concentration microbubble environments, spatial coupling such as overlap or aggregation may occur between microbubbles, leading to localization errors and image distortion. Therefore, effective microbubble decoupling techniques are needed to separate individual microbubbles to ensure accurate localization and imaging.

[0044] In recent years, adaptive clustering algorithms based on density distribution estimation have been widely used in image processing, demonstrating excellent performance in color image segmentation, target tracking, and denoising. However, the application of this algorithm in power Doppler image denoising has not been fully explored, and systematic research on microbubble decoupling in high-concentration microbubble environments is also lacking. In existing technologies, mean shift is mainly used for natural or grayscale image processing, while power Doppler images and ULM microbubble data have unique characteristics, such as the need to maintain blood flow continuity and precise microbubble separation. The low contrast and different morphologies of microbubble scattering in the original contrast-enhanced ultrasound images render many processing methods for natural images ineffective. Therefore, directly applying traditional mean shift may not optimize the results.

[0045] This application aims to propose a microvascular blood flow imaging method based on adaptive density filtering, which simultaneously addresses the noise problem of power Doppler images and the microbubble decoupling challenge in ULM during ultrasound contrast-enhanced imaging of microvessels. It preserves details in noise reduction and improves accuracy in decoupling, thereby enhancing the overall quality of ultrasound imaging.

[0046] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] This application provides a micro-blood flow imaging method based on adaptive density filtering. In one exemplary embodiment, such as... Figure 1 As shown, it includes the following steps:

[0048] A1. For the raw ultrasound data acquired by the ultrasound acquisition device, extract the still frames from the raw ultrasound data, and screen out several high-quality still frames through correlation analysis between the still frames and other frames. The raw ultrasound data includes ultrasound IQ data. The raw ultrasound data acquired by the ultrasound acquisition device (such as an ultrasound transducer) is a modulated wave, which needs to be demodulated according to the probe center frequency. The commonly used method is IQ demodulation. After demodulation, a complex signal is obtained, including amplitude and phase information.

[0049] During the acquisition of raw ultrasound data, breathing or other movements may cause data misregistration. By simply using correlation analysis between still frames and other frames, a large number of high-quality still frames can be retained. Specifically, in this embodiment, the correlation between still frames and all other frames is analyzed according to the following formula:

[0050] .

[0051] in, NCC This represents the cross-correlation coefficient between the still frame and all other frames. IQ ( x , y ) is a still frame at coordinates ( x , y The pixel value at () For the reference frame in coordinates The pixel value at that location is referenced to any ultrasound frame other than a still frame; u , v ) represents the displacement of the still frame relative to the reference frame.

[0052] A2. Perform SVD clutter filtering on the ultrasound IQ data corresponding to each high-quality still frame, and truncate the first K high spatiotemporal coherence singular values ​​through singular value decomposition to obtain the microblood flow signal data after preliminary filtering.

[0053] Specifically, the SVD clutter filtering algorithm is used to filter clutter in ultrasound IQ data, decomposing the matrix into singular values ​​and singular vectors. Larger singular values ​​are retained to remove noise, while the main signal features are preserved to extract microbubble signals from the blood flow. The ultrasound data... IQ ( x , y , t Reconstructed into a Cascorati matrix S .right S The singular value decomposition formula is as follows:

[0054] .

[0055] in, U It is a left singular vector (an orthogonal basis describing the spatial dimension). VIt is a right singular vector (an orthogonal basis describing time or feature dimensions). Σ Let be a diagonal matrix representing singular values. Assume the organizational signal has high spatiotemporal coherence and occupies the top... K Each principal singular value. (By truncation) K Singular values ​​achieve clutter filtering, i.e.

[0056] .

[0057] in, To delete the previous K Columns U , To delete the previous K OK V , To delete the previous K The rows and columns corresponding to the diagonal elements Σ , For the first t Frame pixel coordinates ( x , y The micro-blood flow signal data after preliminary filtering.

[0058] In this embodiment, SVD is chosen for clutter filtering in step A2, but other clutter filters such as RPCA (Robust Principal Component Analysis) and NLM can theoretically achieve similar functions. However, they have problems such as parameter dependence and computational complexity.

[0059] A3. Perform adaptive density filtering on the microblood flow signal data after preliminary filtering, and subtract the result of adaptive density filtering from the microblood flow signal data after preliminary filtering to obtain the microblood flow signal data after complete filtering.

[0060] Adaptive density filtering can effectively remove Gaussian noise and impulse noise, and also plays a role in decoupling overlapping signals. For ultrasound super-resolution imaging applications, the point signals of each microbubble in the processed image sequence are sharpened and separated, which is more conducive to subsequent precise positioning and tracking, thereby generating high-resolution microvascular maps. For power Doppler applications, the background noise of the processed single-frame image is greatly suppressed, highlighting the blood flow signal. In this embodiment, as... Figure 2 As shown, step A3 specifically includes:

[0061] A31. Spatial smoothing is performed on the initially filtered micro-blood flow signal data to obtain smoothed micro-blood flow signal data. In some optional embodiments, various smoothing filters can be used for spatial smoothing, depending on the ease of implementation. Specifically, spatial smoothing is performed on the initially filtered micro-blood flow signal data according to the following formula:

[0062] .

[0063] in, I For smoothed microblood flow signal data, This is the micro-blood flow signal data after preliminary filtering. G ( x , y () is a Gaussian function. Using Gaussian filtering or median filtering, the signal-to-noise ratio can be easily improved.

[0064] A32. For smoothed micro-blood flow signal data, determine the spatial neighborhood search range based on the full width at half maximum (FWHM) of the point spread function (PSF). P 0. Calculate the local maximum intensity difference of a pixel. The local maximum intensity difference reflects the magnitude of intensity variation in a given area of ​​each pixel. Specifically, it is calculated using the following formula:

[0065] .

[0066] in, M ( x 0, y 0) is a pixel ( x 0, y The maximum local intensity difference of pixels (0) N ( P 0) represents the set of neighboring pixels. P 0 represents the spatial neighborhood search range determined by the full width at half maximum (FWHM) of the point spread function. I ( x 0, y 0) represents the micro-blood flow signal data after preliminary filtering. x 0, y 0) represents the center pixel coordinates of the current calculation. i , j ) represents relative to the center pixel ( x 0, y The offset of 0).

[0067] A33. Based on the local maximum intensity difference of pixels, construct a joint spatial and intensity weight matrix for the neighborhood of all pixels in the smoothed micro-blood flow signal data. Specifically, the joint spatial and intensity weight matrix is ​​determined by the spatial kernel function and the intensity kernel function, which are calculated according to the following formula:

[0068] .

[0069] in, For spatial kernel functions, For the intensity kernel function, e For natural numbers, σ Spatial neighborhood search range P The radius of 0.

[0070] A34. The density gradient is calculated based on the spatial and intensity joint weight matrix. The fully filtered microblood flow signal data is obtained by subtracting the density gradient from the initially filtered microblood flow signal data. In this embodiment, the density gradient is calculated according to the following formula:

[0071] .

[0072] in, GD This represents the density gradient.

[0073] Then, the signal distribution is adjusted using a density gradient, and the fully filtered micro-blood flow signal data is calculated according to the following formula:

[0074] .

[0075] in, IQ denoise This is the fully filtered micro-blood flow signal data.

[0076] In the overlapping region of microbubbles (local minimum of probability density), the local density gradient (GD) value is large, and the difference with the original image will produce a negative value, which is subsequently set to zero, thereby suppressing the overlap between noise and signal. At the center of microbubbles (local maximum of probability density), GD tends to zero, and the difference with the original image will retain or enhance its original positive value, thereby sharpening the core signal of the microbubbles.

[0077] In this embodiment, the Gaussian kernel is used to calculate the joint weight of spatial pre-intensity filtering. Other kernel functions that can be used to estimate kernel density, such as the Laplace kernel, Epanechnikov kernel, etc., can also be used.

[0078] Compared to traditional methods, this approach effectively separates difficult-to-process overlapping microbubble signals without prior concentration reduction, increasing the number of localization events and improving the imaging efficiency of ultrasound super-resolution. Simultaneously, it adaptively suppresses tissue background noise and clutter, significantly improving the signal-to-noise ratio of power Doppler images. As a post-processing algorithm, its non-parametric design results in low implementation and computational costs, allowing for rapid integration into different ultrasound imaging workflows.

[0079] A4. Time-lag autocorrelation processing is performed on the fully filtered microblood flow signal data. The lagged frame data is multiplied pixel-by-pixel to eliminate time-related noise and enhance the microbubble motion signal, resulting in a microblood flow signal data sequence. In this embodiment, time-lag autocorrelation processing is performed on the fully filtered microblood flow signal data according to the following formula:

[0080] .

[0081] in, This is a microblood flow signal data sequence. For the first t Fully filtered micro-blood flow signal data of the frame. IQ denoise ( t+lag () represents the fully filtered micro-blood flow signal data after lag. lag This is the lag coefficient.

[0082] The fully filtered microblood flow signal data is further eliminated by pixel-by-pixel multiplication of the hysteresis frame, which enhances the microbubble motion signal, amplifies the phase change caused by microbubble displacement, and suppresses static clutter.

[0083] A5. Based on the microblood flow signal data sequence, power Doppler imaging and ultrasound-guided microscopy reconstruction were performed to obtain power Doppler images and super-resolution density images of the microblood flow signal.

[0084] Power Doppler essentially involves performing zero-hysteresis autocorrelation on a micro-blood flow signal data sequence to obtain the Doppler power spectrum, thus acquiring the blood flow signal. As shown in the following equation:

[0085] .

[0086] in, PowDop This is a micro-blood flow signal. N Microblood flow signal data sequence Frame count, conj ( ) represents the conjugate. The conjugate of a complex number is a value with the same magnitude but opposite phase, while the conjugate of a real number is itself. This formula means that for... R ( t After squaring the intensity of each frame in the sequence, the summation of each frame is then averaged to synthesize a single power Doppler image from the sequence.

[0087] Microbubble localization was performed on the microblood flow signal data sequence after adaptive density filtering using a classic and practical radial symmetry algorithm (RS). This algorithm determines the center position of the microbubble based on the geometric characteristics of the microbubble backscatter amplitude, thus extending microbubble localization to the sub-pixel level. Sub-pixel localization is as follows: Figure 3 As shown.

[0088] For each frame, subpixel localization coordinates were used for trajectory tracking. A bipartite graph global optimal matching algorithm based on the Hungarian algorithm was employed. This algorithm sets the cost function based on the distance between microbubbles, resulting in low computational cost and efficient and accurate results. By initializing a blank image of a specified super-resolution scale (i.e., all zeros), the pixels traversed by the motion trajectory obtained by the trajectory tracking algorithm were counted and superimposed to obtain the reconstructed density image of the ultrasonic positioning microscope.

[0089] In some other alternative embodiments, the trajectory tracking algorithm in the ULM can be replaced by other methods, such as Gaussian fitting, interpolation localization algorithms, and Kalman filter-based tracking methods. This embodiment essentially improves the signal-to-noise ratio and separates sufficient localizable events for ultrasonic localization microscopy reconstruction.

[0090] Regarding the denoising effect of the filter, SNR is used as the evaluation criterion in this embodiment, as shown in the following formula:

[0091] .

[0092] in, Represents microblood flow signals PowDop Regions of interest (ROIs) were manually selected to represent blood flow and background. The selection of ROIs is as follows: Figure 4 As shown, σ background This represents the standard deviation of the background noise. The results are expressed logarithmically: 10log10( SNR ).

[0093] The ultrasound imaging acquisition equipment used was the Verasonics programmable ultrasound research system from the Vantange series. This system is widely used in the field of ultrasound research; as a multi-channel ultrasound acquisition device, it allows for customization of pulse transmission waveforms, transmission and reception modes, and different imaging methods. The rat liver data below were acquired using the parameters in Table 1, and the rat brain data were acquired using the parameters in Table 2.

[0094] Table 1. Rat liver data acquisition parameters

[0095]

[0096] Table 2. Rat brain data acquisition parameters

[0097]

[0098] For rat liver data, conventional power Doppler imaging (200 frames and 600 frames) and power Doppler imaging with adaptive density filtering (200 frames and 600 frames) were performed respectively. The comparison results are as follows: Figure 5As shown in the figure, λ on the horizontal and vertical axes represents the sampling index of the spatial dimension or the spatial scale based on the ultrasonic length, used to identify the pixel position or relative spatial distribution of the image in space; and the ULM without adaptive density filtering after still frame extraction is as follows. Figure 6 As shown in the rat liver example above, the noise reduction effect of this method is significant in a few or more frames of ultrasound data, and it highlights weak trajectories that disappear in ULM due to the reconstruction of a large number of trajectories in tens of thousands of frames, and the filtering does not cause the disappearance of visible real blood vessels.

[0099] For rat brain data, traditional ULM (10,000 frames) and (25,000 frames) reconstructions were performed, as well as ULM (10,000 frames) and (25,000 frames) reconstructions after adaptive density filtering. The results are compared as follows: Figure 7 As shown, in addition, through Figure 8 The differences between the rat brain ULM orientation maps before and after adaptive density filtering were analyzed, and further analysis was conducted through... Figure 9 The differences in Fourier loop correlation curves of the rat brain ULM (Ultraluminal Lens) before and after adaptive density filtering were analyzed. As seen in the rat brain examples above, the ULM reconstructed images after adaptive density filtering are of better quality, showing higher vascular filling and highlighting some previously invisible vessels, thus improving resolution. This is attributed to the increased number of accurately localizable events obtained after mean-shift filtering, resulting in the reconstruction of more trajectories and better ULM images, with particularly significant effects in a smaller number of frames.

[0100] In addition, through Figure 10 Comparing the various filtering algorithms with Table 3, it can be seen that the adaptive density filtering used in this embodiment improves the signal-to-noise ratio by nearly 27 dB compared to traditional filtering methods, and its visualization effect also far surpasses other filtering methods. Among these, the noise equalization method did not truly improve the signal-to-noise ratio. Post-processing methods such as vessel filtering effectively removed some noise, but noise was still retained at deeper depths, while mean-shift filtering showed good results across the entire depth range, with a dynamic range of 40.

[0101] Table 3. Comparison of signal-to-noise ratios of different filtering methods

[0102]

[0103] The method presented in this application innovatively applies the principle of adaptive density filtering to the filtering process of ultrasound images. Through density clustering, it removes a large amount of large-area Gaussian noise and impulse noise that are difficult to remove using traditional methods. Simultaneously, it separates some overlapping microbubbles in a high-concentration microbubble environment, increasing the number of microbubble localization events, thereby improving super-resolution and image quality. By employing multi-scale clustering and pre-filtering, it overcomes the limitations of previous methods, which were only applicable to natural images. Through selective multiple iterations, even better results may be obtained. This method achieves an approximate deconvolution effect without the corresponding high computational cost and artifacts. Furthermore, as a post-processing method, it has no specific requirements for experimental conditions and has low computational complexity. Finally, Fourier ring correlation (FRC) is introduced as an objective evaluation standard for ULM resolution, such as... Figure 9 As shown in the figure, in vivo super-resolution imaging results of rat livers demonstrate that the minimum resolution of the improved ULM imaging quality has been increased from 45.0 μm to 37.6 μm. Therefore, this method has significant advantages over existing technologies.

[0104] Based on the same inventive concept, this application also provides a system for implementing the micro-blood flow imaging method based on adaptive density filtering described above. The solution provided by this system is similar to the implementation described in the above method. In an exemplary embodiment, such as... Figure 11 As shown, a micro-blood flow imaging system based on adaptive density filtering is provided, including the following functional modules:

[0105] The still frame filtering module is used to extract still frames from the raw ultrasound data acquired by the ultrasound acquisition device, and to filter out a number of high-quality still frames through correlation analysis between the still frames and other frames. The raw ultrasound data includes ultrasound IQ data.

[0106] The SVD clutter filtering module is used to perform SVD clutter filtering on the ultrasound IQ data corresponding to each high-quality still frame, and to truncate the first K high spatiotemporal coherence singular values ​​through singular value decomposition to obtain the microblood flow signal data after preliminary filtering.

[0107] The adaptive density filtering module is used to perform adaptive density filtering on the microblood flow signal data after preliminary filtering, and to subtract the result of the adaptive density filtering from the microblood flow signal data after preliminary filtering to obtain the microblood flow signal data after complete filtering.

[0108] The time-lag autocorrelation processing module is used to perform time-lag autocorrelation processing on the fully filtered microblood flow signal data, multiplying the lagged frame data pixel by pixel to eliminate time-related noise and enhance the microbubble motion signal, thereby obtaining the microblood flow signal data sequence.

[0109] The microblood flow imaging module is used to perform power Doppler imaging and ultrasound-guided microscopy reconstruction based on microblood flow signal data sequences to obtain power Doppler images and super-resolution density images of microblood flow signals.

[0110] certainly, Figure 11 The architecture shown is merely exemplary; it can be omitted as needed when implementing different functionalities. Figure 11 One or at least two components of the system shown.

[0111] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 12 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it can implement the micro-blood flow imaging method based on adaptive density filtering provided in the previous embodiment.

[0112] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0113] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0114] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0115] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0116] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0117] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0118] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0119] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A micro-blood flow imaging method based on adaptive density filtering, characterized in that, include: For the raw ultrasound data acquired by the ultrasound acquisition device, still frames are extracted from the raw ultrasound data, and a number of high-quality still frames are selected by correlation analysis between the still frames and other frames. The raw ultrasound data includes ultrasound IQ data. SVD clutter filtering is performed on the ultrasound IQ data corresponding to each of the high-quality still frames, and the first K high spatiotemporal coherence singular values ​​are truncated by singular value decomposition to obtain the microblood flow signal data after preliminary filtering. Adaptive density filtering is performed on the microblood flow signal data after preliminary filtering, and the result of adaptive density filtering is subtracted from the microblood flow signal data after preliminary filtering to obtain the microblood flow signal data after complete filtering. Time-lag autocorrelation processing was performed on the fully filtered microblood flow signal data, and the lagged frame data was multiplied pixel by pixel to eliminate time-related noise and enhance microbubble motion signal, thus obtaining microblood flow signal data sequence. Based on the microblood flow signal data sequence, power Doppler imaging and ultrasound localization microscopy reconstruction were performed respectively to obtain power Doppler images and super-resolution density images of the microblood flow signal. Adaptive density filtering is performed on the pre-filtered microblood flow signal data, and the difference between the adaptive density filtering result and the pre-filtered microblood flow signal data is calculated to obtain the fully filtered microblood flow signal data, specifically including: Spatial smoothing is performed on the pre-filtered microblood flow signal data to obtain smoothed microblood flow signal data. For the smoothed micro-blood flow signal data, the spatial neighborhood search range is determined based on the full width at half maximum (FWHM) of the point spread function, and the local maximum intensity difference of the pixels is calculated. Based on the local maximum intensity difference of the pixels, a joint spatial and intensity weight matrix of the neighborhood of all pixels in the smoothed microblood flow signal data is constructed. The density gradient is calculated based on the spatial and intensity joint weight matrix. The fully filtered microblood flow signal data is obtained by subtracting the density gradient from the preliminarily filtered microblood flow signal data.

2. The micro-blood flow imaging method based on adaptive density filtering according to claim 1, characterized in that, Spatial smoothing is performed on the pre-filtered micro-blood flow signal data according to the following formula: ； in, I For smoothed microblood flow signal data, IQ svd ( x , y , t 0) represents the micro-blood flow signal data after preliminary filtering. G ( x , y () is a Gaussian function; The maximum local intensity difference of a pixel is calculated using the following formula: ； in, M ( x 0, y 0) is a pixel ( x 0, y The maximum local intensity difference of pixels (0) N ( P 0) represents the set of neighboring pixels. P 0 represents the spatial neighborhood search range determined by the full width at half maximum (FWHM) of the point spread function. I ( x 0, y 0) represents the micro-blood flow signal data after preliminary filtering. x 0, y 0) represents the center pixel coordinates of the current calculation. Represents relative to the center pixel ( x 0, y The offset of 0); The joint weight matrix of spatial and intensity kernel functions is determined by the spatial kernel function and the intensity kernel function, which are calculated according to the following formula: ； in, For spatial kernel functions, For the intensity kernel function, e It is a natural number.

3. The micro-blood flow imaging method based on adaptive density filtering according to claim 2, characterized in that, The density gradient is calculated using the following formula: ； in, GD For density gradient; The fully filtered micro-blood flow signal data is calculated using the following formula: ； in, IQ denoise This is the fully filtered micro-blood flow signal data.

4. The micro-blood flow imaging method based on adaptive density filtering according to claim 1, characterized in that, The correlation between the still frame and all other frames except the still frame is analyzed according to the following formula: ； in, NCC The cross-correlation coefficient between the still frame and all other frames except the still frame is given. IQ ( x , y ) is a still frame at coordinates ( x , y The pixel value at () For the reference frame in coordinates The pixel value at that location, with the reference frame being any ultrasound frame other than the aforementioned still frame; u , v ) represents the displacement of the still frame relative to the reference frame.

5. The micro-blood flow imaging method based on adaptive density filtering according to claim 1, characterized in that, The time-lag autocorrelation processing of the fully filtered micro-blood flow signal data is performed according to the following formula: ； in, R ( t () represents a microblood flow signal data sequence. IQ denoise ( t ) is the first t Fully filtered micro-blood flow signal data of the frame. IQ denoise ( t+lag () represents the fully filtered micro-blood flow signal data after lag. lag This is the lag coefficient.

6. A micro-blood flow imaging system based on adaptive density filtering, characterized in that, include: The still frame filtering module is used to extract still frames from the raw ultrasound data acquired by the ultrasound acquisition device, and to filter out a number of high-quality still frames through correlation analysis between the still frames and other frames; the raw ultrasound data includes ultrasound IQ data. The SVD clutter filtering module is used to perform SVD clutter filtering on the ultrasound IQ data corresponding to each of the high-quality still frames, and to truncate the first K high spatiotemporal coherence singular values ​​through singular value decomposition to obtain the microblood flow signal data after preliminary filtering. The adaptive density filtering module is used to perform adaptive density filtering on the microblood flow signal data after preliminary filtering, and to subtract the result of the adaptive density filtering from the microblood flow signal data after preliminary filtering to obtain the microblood flow signal data after complete filtering. The time lag autocorrelation processing module is used to perform time lag autocorrelation processing on the fully filtered microblood flow signal data, multiplying the lag frame data pixel by pixel to eliminate time-related noise and enhance the microbubble motion signal to obtain the microblood flow signal data sequence. The microblood flow imaging module is used to perform power Doppler imaging and ultrasound-guided microscopy reconstruction based on the microblood flow signal data sequence to obtain power Doppler images and super-resolution density images of the microblood flow signal. Adaptive density filtering is performed on the pre-filtered microblood flow signal data, and the difference between the adaptive density filtering result and the pre-filtered microblood flow signal data is calculated to obtain the fully filtered microblood flow signal data, specifically including: Spatial smoothing is performed on the pre-filtered microblood flow signal data to obtain smoothed microblood flow signal data. For the smoothed micro-blood flow signal data, the spatial neighborhood search range is determined based on the full width at half maximum (FWHM) of the point spread function, and the local maximum intensity difference of the pixels is calculated. Based on the local maximum intensity difference of the pixels, a joint spatial and intensity weight matrix of the neighborhood of all pixels in the smoothed microblood flow signal data is constructed. The density gradient is calculated based on the spatial and intensity joint weight matrix. The fully filtered microblood flow signal data is obtained by subtracting the density gradient from the preliminarily filtered microblood flow signal data.

7. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the micro-blood flow imaging method based on adaptive density filtering as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the microblood flow imaging method based on adaptive density filtering as described in any one of claims 1-5.

9. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the microblood flow imaging method based on adaptive density filtering as described in any one of claims 1-5.

Images