Matrix array full-matrix capture three-dimensional ultrasonic imaging method and system based on frequency-wavenumber domain

Abstract

The invention provides a frequency-wavenumber domain-based matrix array full-matrix capture three-dimensional ultrasonic imaging method and system, and belongs to the technical field of ultrasonic imaging. Five-dimensional time-space domain radio frequency data are collected through a two-dimensional full-sampling matrix array and are converted into a frequency-wave number domain through five-dimensional fast Fourier transform, angular frequency is mapped into axial wave number through Stolt interpolation, data resampling is achieved, then three-dimensional image frequency spectrum data are synthesized by defining spatial wave number coordinates of an image and conducting coherence stacking, and the image frequency spectrum data are obtained. And finally, reconstructing a high-quality three-dimensional ultrasonic image through three-dimensional inverse fast Fourier transform. Compared with a traditional delay summation beam forming method, the method has the advantages that the calculation efficiency is remarkably improved, the speed is increased by several times to dozens of times, the spatial resolution, the contrast noise ratio and the spot signal-to-noise ratio are remarkably improved, and the method is suitable for high-resolution real-time three-dimensional and four-dimensional ultrasonic imaging application.

Description

Technical Field

[0001] This invention belongs to the field of ultrasound imaging technology, specifically relating to a three-dimensional ultrasound imaging method and system based on a matrix array full-matrix capture in the frequency-wavenumber domain. Background Technology

[0002] Three-dimensional (3D) ultrasound imaging is an advanced medical imaging modality that provides volumetric visualization of internal structures. [1][2] Typically, 3D ultrasound imaging technology uses mechanical scanning... [3] Freehand scanning [4][5] and two-dimensional (2D) array transducers [6][7] Mechanical scanning is achieved by mechanically controlling a one-dimensional (1D) array to move along different scanning trajectories (such as linear, tilt, and rotational scans). Multiple cross-sectional images are acquired sequentially at different spatial locations and then reconstructed into a 3D volumetric dataset using image processing techniques. In contrast, handheld scanning is achieved by manually moving the 1D array while using position sensors to track the probe position. Handheld scanning provides flexible control over the scanning trajectory to meet different clinical needs. Additionally, 2D array transducers use electronic scanning and pulse delay control to manipulate and focus the beam, enabling the acquisition of high frame rate 3D ultrasound images.

[0003] Full-sampling matrix arrays are a typical example of 2D arrays, where transducer elements are arranged in a checkerboard-like grid pattern. Each transducer element in a matrix array can be independently driven to manipulate and focus the ultrasonic beam in both the transverse and longitudinal directions. Due to their full-sampling capability and precise beam control, matrix arrays typically offer superior performance compared to other 2D array configurations (such as sparse arrays). [8][9] and row-column addressing array

[10]

[11] Image quality.

[0004] 3D ultrasound imaging using matrix arrays has been extensively explored in various studies. (Gonçalves et al.)

[12] Fetal structures, including the fetal heart, were examined using a 2D matrix array. (Provost et al.)

[13] 3D ultrafast ultrasound imaging was achieved using divergent or plane waves emitted by a sparse virtual array located behind a 32×32 matrix array. Gennison et al.

[14] Four-dimensional (4D) shear wave ultrasound imaging was performed using a 32×32 matrix phased array driven by a 1024-channel ultrasound system. (Rabut et al.)

[15] Four-dimensional (4D) functional ultrasound imaging of the whole brain was achieved using multiple plane waves emitted by a 2D matrix array transducer. (Heiles et al.)

[16] A volumetric ultrasound localization microscopy technique was achieved using a 32×32 element, 9MHz matrix probe. (Bureau et al.)

[17] Volumetric imaging of tissue-simulated phantoms was performed using ex vivo tissue, subsequently demonstrating the potential of 3D matrix imaging in transcranial applications. Full-sampling matrix arrays play a crucial role in ultrasound volumetric imaging.

[0005] The Total Focusing Method (TFM) is an ultrasound imaging technique that uses a full matrix capture (FMC) dataset to reconstruct a high-resolution image by dynamically focusing at each point in the region of interest. However, in the case of matrix arrays, conventional Delay Summation (DAS) beamformers become computationally inefficient as the number of channels and pixels increases.

[0006] Beamformers based on the frequency-wavenumber (fk) domain have proven to be an effective method for reducing the computational complexity of image reconstruction. This algorithm is adapted from the Fourier domain Stolt migration algorithm originally developed for seismic imaging.

[18] This algorithm has also been applied to the fields of synthetic aperture radar (SAR) and sonar.

[19]

[20] Subsequently, this method was extended to the field of ultrasound imaging, promoting further development in this field. Stepinski et al.

[21] Synthetic Aperture Focusing Technique (SAFT) for synthetic aperture radar and sonar was implemented in the frequency domain. (Lu et al.)

[22] A pulse-echo imaging method was developed to achieve high frame rate imaging with a finite diffraction beam. Subsequently, Cheng et al.

[23] This further extends the theory of high frame rate imaging for rapid 3D ultrasound imaging. Garcia et al.

[24] Using an improved exploded mirror model, fk-domain ultrasound imaging was achieved on plane wave echo data. Hunter et al.

[25] A wavenumber algorithm-based Fourier domain full-matrix imaging method was introduced for ultrasonic nondestructive testing and evaluation. Bernard et al.

[26] A novel ultrasound acquisition scheme based on Fourier slice imaging is proposed to achieve finer lateral resolution compared to other Fourier-based methods. (Vol et al.)

[27] A two-stage beamforming method is introduced, which combines line-based DAS with subsequent FFT-based beamforming to reduce the computational load and data transmission rate of linear array systems while maintaining resolution comparable to traditional methods. Zhang et al.

[28] By proposing an explicit transform, Fourier-based techniques are extended to reconstruct fan-shaped images from divergent wave (DW) data. Moghimirad et al.

[29] A highly efficient Fourier beamformer for multi-base synthetic aperture ultrasound imaging using a virtual source is proposed. Results demonstrate that the processing time for Fourier beamforming using a virtual source is reduced by 20 times compared to DAS. (Albulayli et al.)

[30] By modifying two classic algorithms used for geophysical data processing (namely, Stolt migration and tilt stacking depth migration under the assumption of zero migration constant velocity), plane wave ultrasound imaging at ultrafast image acquisition rates was achieved. Merabet et al.

[31] Computationally efficient 2D and 3D reconstruction algorithms were developed in the fk domain to achieve rapid imaging of defects in solids. (Guo et al.)

[32] Integrating a wavenumber beamformer with an under-Nyquist sampling phase reduces the computational complexity of conventional focused ultrasound imaging.36 (Lin et al.)

[33] A fast 3D ultrasound imaging method based on Fourier transform using a row and column addressing 2D array is proposed, which reduces computational complexity.

[0007] Previous studies have demonstrated the effectiveness of fk beamformers in achieving efficient image reconstruction. However, fk beamformers for 3D TFM imaging of matrix arrays remain to be investigated.

[0008] [1]QH Huang and ZZ Zeng, "A Review on Real-Time 3D UltrasoundImaging Technology," Biomed Research International, vol. 2017, 2017, Art no.6027029. 40

[0009] [2]A. Fenster and DB Downey, "3-D ultrasound imaging: A review," IEEE Eng. Med. Biol. Mag., vol. 15, no. 6, pp. 41-51, 1996.

[0010] [3]C. C. Chang, P. Y. Chen, H. Huang, and C. C. Huang, "In VivoVisualization of Vasculature in Adult Zebrafish by Using High-FrequencyUltrafast Ultrasound Imaging," IEEE Transactions on Biomedical Engineering,vol. 66, no. 6, pp. 1742-1751, 2019, doi: 10.1109 / TBME.2018.2878887.

[0011] [4]S.-W. Chung, C.-C. Shih, and C.-C. Huang, "Freehand three-dimensional ultrasound imaging of carotid artery using motion trackingtechnology," Ultrasonics, vol. 74, pp. 11-20, 2017 / 02 / 01 / 2017, doi: https: / / doi.org / 10.1016 / j.ultras.2016.09.020.

[0012] [5]M. H. Mozaffari and W.-S. Lee, "Freehand 3-D Ultrasound Imaging: ASystematic Review," Ultrasound Med. Biol., vol. 43, no. 10, pp. 2099-2124,2017 / 10 / 01 / 2017, doi: https: / / doi.org / 10.1016 / j.ultrasmedbio.2017.06.009.

[0013] [6]A. Fenster and D. B. Downey, "Three-dimensional ultrasoundimaging," Annu. Rev. Biomed. Eng., vol. 2, no. 1, pp. 457-475, 2000.

[0014] [7]B. Heiles et al. , "Ultrafast 3D Ultrasound LocalizationMicroscopy Using a 32 $\times$ 32 Matrix Array," IEEE Trans. Med. Imaging,vol. 38, no. 9, pp. 2005-2015, 2019, doi: 10.1109 / TMI.2018.2890358.

[0015] [8]A. Ramalli, E. Boni, E. Roux, H. Liebgott, and P. Tortoli, "Design, implementation, and medical applications of 2-D ultrasound sparsearrays," IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 69, no. 10,pp. 2739-2755, 2022.

[0016] [9]M. H. Masoumi, T. Kaddoura, and R. J. Zemp, "Costas sparse 2-Darrays for high-resolution ultrasound imaging," IEEE Trans. Ultrason.Ferroelectr. Freq. Control, vol. 70, no. 5, pp. 460-472, 2023.

[0017]

[10] Z. Dong, U. W. Lok, M. R. Lowerison, C. Huang, S. Chen, and P.Song, "Three-Dimensional Shear Wave Elastography Using Acoustic RadiationForce and a 2-D Row-Column Addressing (RCA) Array," IEEE Trans. Ultrason.Ferroelectr. Freq. Control, vol. 71, no. 4, pp. 448-458, 2024, doi: 10.1109 / TUFFC.2024.3366540.

[0018]

[11] M. Flesch et al. , "4D in vivo ultrafast ultrasound imaging usinga row-column addressed matrix and coherently-compounded orthogonal planewaves," Phys. Med. Biol., vol. 62, no. 11, p. 4571, 2017.

[0019]

[12] L. F. Gonçalves et al. , "Applications of 2‐dimensional matrixarray for 3‐and 4‐dimensional examination of the fetus: a pictorial essay,"J. Ultrasound Med., vol. 25, no. 6, pp. 745-755, 2006.

[0020]

[13] J. Provost et al. , "3D ultrafast ultrasound imaging in vivo,"Phys. Med. Biol., vol. 59, no. 19, p. L1, 2014.

[0021]

[14] J.-L. Gennisson et al. , "4-D ultrafast shear-wave imaging," IEEETrans. Ultrason. Ferroelectr. Freq. Control, vol. 62, no. 6, pp. 1059-1065,2015.

[0022]

[15] C. Rabut et al. , "4D functional ultrasound imaging of whole-brain activity in rodents," Nature methods, vol. 16, no. 10, pp. 994-997,2019.

[0023]

[16] B. Heiles et al. , "Ultrafast 3D ultrasound localizationmicroscopy using a 32×32 matrix array," IEEE Trans. Med. Imaging, vol. 38,no. 9, pp. 2005-2015, 2019.

[0024]

[17] F. Bureau, J. Robin, A. Le Ber, W. Lambert, M. Fink, and A.Aubry, "Three-dimensional ultrasound matrix imaging," Nat. Commun., vol. 14,no. 1, p. 6793, 2023.

[0025]

[18] R. H. Stolt, "Migration by Fourier transform," Geophysics, vol.43, no. 1, pp. 23-48, 1978.

[0026]

[19] C. Cafforio, C. Prati, and F. Rocca, "SAR data focusing usingseismic migration techniques," IEEE transactions on aerospace and electronicsystems, vol. 27, no. 2, pp. 194-207, 1991.

[0027]

[20] M. Soumekh, "Reconnaissance with ultra wideband UHF syntheticaperture radar," IEEE Signal processing magazine, vol. 12, no. 4, pp. 21-40,1995.

[0028]

[21] T. Stepinski, "An implementation of synthetic aperture focusingtechnique in frequency domain," IEEE Trans. Ultrason. Ferroelectr. Freq.Control, vol. 54, no. 7, pp. 1399-1408, 2007.

[0029]

[22] J.-y. Lu, "2D and 3D high frame rate imaging with limiteddiffraction beams," IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol.44, no. 4, pp. 839-856, 1997.

[0030]

[23] J. Cheng and J.-y. Lu, "Extended high-frame rate imaging methodwith limited-diffraction beams," IEEE Trans. Ultrason. Ferroelectr. Freq.Control, vol. 53, no. 5, pp. 880-899, 2006.

[0031]

[24] D. Garcia, L. L. Tarnec, S. Muth, E. Montagnon, J. Porée, and G.Cloutier, "Stolt's f-k migration for plane wave ultrasound imaging," IEEETrans. Ultrason. Ferroelectr. Freq. Control, vol. 60, no. 9, pp. 1853-1867,2013.

[0032]

[25] A. J. Hunter, B. W. Drinkwater, and P. D. Wilcox, "The wavenumberalgorithm for full-matrix imaging using an ultrasonic array," IEEE Trans.Ultrason. Ferroelectr. Freq. Control, vol. 55, no. 11, pp. 2450-2462, 2008.

[0033]

[26] O. Bernard et al. , "Ultrasound Fourier slice imaging: a novelapproach for ultrafast imaging technique," in 2014 IEEE InternationalUltrasonics Symposium , 3-6 Sept. 2014 2014, pp. 129-132, doi: 10.1109 / ULTSYM.2014.0033.

[0034]

[27] H. J. Vos, P. L. van Neer, M. M. Mota, M. D. Verweij, A. F. vander Steen, and A. W. Volker, "F–k domain imaging for synthetic aperturesequential beamforming," IEEE Trans. Ultrason. Ferroelectr. Freq. Control,vol. 63, no. 1, pp. 60-71, 2015.

[0035]

[28] M. Zhang et al., "Extension of Fourier-Based Techniques forUltrafast Imaging in Ultrasound With Diverging Waves," IEEE Trans. Ultrason.Ferroelectr. Freq. Control, vol. 63, no. 12, pp. 2125-2137, 2016, doi:10.1109 / TUFFC.2016.2616300.

[0036]

[29] E. Moghimirad, C. A. V. Hoyos, A. Mahloojifar, B. M. Asl, and J.A. Jensen, "Synthetic aperture ultrasound Fourier beamformation using virtualsources," IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 63, no. 12,pp. 2018-2030, 2016.

[0037]

[30] M. Albulayli and D. Rakhmatov, "Fourier domain depth migrationfor plane-wave ultrasound imaging," IEEE Trans. Ultrason. Ferroelectr. Freq.Control, vol. 65, no. 8, pp. 1321-1333, 2018.

[0038]

[31] L. Merabet, S. Robert, and C. Prada, "2-D and 3-D reconstructionalgorithms in the Fourier domain for plane-wave imaging in nondestructivetesting," IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 66, no. 4,pp. 772-788, 2019.

[0039]

[32] H. Guo, S. Freear, and GQ Zhou, "Wavenumber beamforming with sub-Nyquist sampling for focus-beam ultrasound imaging," IEEE Trans.Ultrason. Ferroelectr. Freq. Control, vol. 71, no. 8, pp. 972-984, 2024.

[0040]

[33] S.-C. Lin and P.-C. Li, "Fourier-based fast 3-D ultrasound imaging using row-column-addressed 2-D arrays," IEEE Trans. Ultrason.Ferroelectr. Freq. Control, vol. 71, no. 1, pp. 85-101, 2024. Summary of the Invention

[0041] This invention is made to solve the above-mentioned problems, and aims to provide a three-dimensional ultrasound imaging method and system based on frequency-wavenumber domain matrix array full matrix capture.

[0042] This invention provides a three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture, characterized by the following steps: S10, acquiring five-dimensional spatiotemporal domain radio frequency data P(u) by performing full matrix capture of the medium through a two-dimensional full sampling matrix array. x u y v x v y After t), a five-dimensional fast Fourier transform is performed on it to convert it to the frequency-wavenumber domain, obtaining the spectral data P(k). ux k uy k vx k vy ,ω), where, (u x u y (v) represents the coordinates of the transmitting element. x v y ) represents the coordinates of the receiving array element, t represents time, and k represents the time. ux k uy k represents the transverse wave number emitted. vx k vyThis indicates the received transverse wavenumber, and ω represents the angular frequency; S20, based on the angular frequency ω and the corresponding axial wavenumber k Z Relationship For spectral data P(k ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis, where c represents the speed of sound in the medium, k uz k represents the transmitted axial wave number. vz S30 indicates the received axial wavenumber; S30, defines the spatial wavenumber coordinates of the image as: k X =k ux +k vx k Y =k uy +k vy The spectral data after Stolt interpolation mapping in step S20 are coherently superimposed to synthesize three-dimensional image spectral data F(k). X k Y k Z S40, for the three-dimensional image spectral data F(k) X k Y k Z After performing a three-dimensional inverse fast Fourier transform, a three-dimensional ultrasound image f(x, y, z) in the spatial domain is reconstructed.

[0043] The three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture provided by the present invention may also have the following features: wherein, in step S10, the two-dimensional full sampling matrix array includes multiple array elements, the multiple array elements are arranged in a rectangular grid on the x-axis and y-axis, the multiple array elements are used as transmitting array elements to transmit ultrasonic waves in sequence, and all array elements are used as receiving array elements to receive echo signals.

[0044] The three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture provided by this invention may also have the following feature: In step S10, before performing the five-dimensional fast Fourier transform, the five-dimensional spatiotemporal domain radio frequency data P(u x u y v x v y ,t) is zero-padded in its five dimensions.

[0045] The three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture provided by this invention may also have the following feature: wherein step S20 includes the following sub-step: S21, the three-dimensional Green's function from the origin to the spatial point (x, y, z) is expressed as follows according to the Weyl identity: , where k x k represents the wave number in the x-axis direction. y k represents the wave number in the y-axis direction. z This represents the wave number along the z-axis, where i represents the imaginary unit. S22, based on the Green's function, the ultrasonic wave is scattered from the transmitting array element to the theoretical scattering point f. theory The complete pulse echo process of (x, y, z) and returning the received array element can be represented by the following spatial integral: S23, After converting the spatial integral to the frequency-wavenumber domain, the spectral data P(k) of the received signal is obtained. ux k uy k vx k vy ,ω) and the theoretical scattering point f theory The theoretical value F of the three-dimensional image spectrum data corresponding to (x, y, z) theory (k X k Y k Z The relationship between ) S24, Establish the image spectral wavenumber (k) based on the relationship in step S23. X k Y k Z ) and data spectrum wavenumber (k ux k uy k vx k vy The relationship between k and ω: X =k ux +k vx k Y =k uy +k vy , S25, Based on the relationship in step S24, process the spectrum data P(k) ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis.

[0046] This invention also provides a frequency-wavenumber domain-based matrix array full-matrix capture three-dimensional ultrasound imaging system, characterized by using any of the aforementioned frequency-wavenumber domain-based matrix array full-matrix capture three-dimensional ultrasound imaging methods, including: a two-dimensional full-sampling matrix array for full-matrix capture of the medium to acquire five-dimensional spatiotemporal domain radio frequency data P(u x u y v x v y ,t); Frequency domain conversion module, used for converting five-dimensional spatiotemporal radio frequency data P(u x u y v x v y Perform a five-dimensional fast Fourier transform on t) to convert it to the frequency-wavenumber domain to obtain the spectral data P(k). ux k uy k vx k vy , ω); Wavenumber mapping module, used to map angular frequency ω and corresponding axial wavenumber k Z Relationship For the spectral data P(k ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On-axis; Overlay and synthesis module, used to define the spatial wavenumber coordinates of the image as k X =k ux +k vx And k Y =k uy +k vy Then, the spectral data after Stolt interpolation mapping by the wavenumber mapping module are coherently superimposed to synthesize the three-dimensional image spectral data F(k). X k Y k Z ); and an inverse transform module and a three-dimensional imaging module, used for processing the three-dimensional image spectral data F(k X k Y k Z After performing a three-dimensional inverse fast Fourier transform, a three-dimensional ultrasound image f(x, y, z) in the spatial domain is reconstructed.

[0047] The role and effect of invention

[0048] (1) Improve computational efficiency: This invention reduces the computational complexity of the traditional DAS algorithm from O(N) to O(N) computational efficiency. 4 M 2 The delayed superposition process of H) operations is converted into an O(N)-time FFT-based O(N) time superposition process. 4 Klog(N 4 The method involves processing K)) operations and Stolt interpolation. Test results show that the method of this invention achieves a reduction in computation time by several times or even more than ten times on the CPU platform.

[0049] (2) Improved image quality: Compared with the DAS method, this invention is based on a more accurate wavefield inverse scattering mathematical model, which not only corrects phase deviation but also introduces a precise amplitude compensation mechanism. Test results show that this invention has significant improvements in spatial resolution (FWHM average improvement of 19.18% to 25.00%), contrast-to-noise ratio (CNR average improvement of 11.73% to 18.18%), and speckle signal-to-noise ratio (sSNR average improvement of 11.16% to 19.74%).

[0050] (3) Broad application prospects: This invention provides efficient algorithm support for achieving high-resolution, high-frame-rate 3D / 4D ultrasound imaging (such as real-time hemodynamic monitoring, functional ultrasound imaging, etc.). Attached Figure Description

[0051] Figure 1 This is a flowchart of a three-dimensional ultrasound imaging method based on a matrix array full-matrix capture in the frequency-wavenumber domain, according to an embodiment of the present invention.

[0052] Figure 2 This is a schematic diagram of a two-dimensional full sampling matrix array and its full matrix capture in an embodiment of the present invention.

[0053] Figure 3 This is a schematic diagram of the architecture of a matrix array full-matrix capture three-dimensional ultrasound imaging system based on the frequency-wavenumber domain in an embodiment of the present invention.

[0054] Figure 4 This is a schematic diagram illustrating the principle of TFM-DAS beamforming under existing technology.

[0055] Figure 5 These are the reconstructed images and quantitative comparison results of simulated point targets in the test examples of this invention.

[0056] Figure 6 These are the reconstructed images and quantitative comparison results of the phantom line target in the test examples of this invention.

[0057] Figure 7 This is a reconstructed image of the phantom cyst target in the test example of the present invention.

[0058] Figure 8 This is a reconstructed image of the in vivo carotid artery in a test example of the present invention.

[0059] Figure 9 This is a graph comparing the computation time of the embodiment method in the test example of the present invention with that of the TFM-DAS method in the prior art. Detailed Implementation

[0060] To make the technical means, creative features, objectives and effects of the present invention easy to understand, the following embodiments, in conjunction with the accompanying drawings, specifically illustrate a matrix array full-matrix capture three-dimensional ultrasound imaging method and system based on the frequency-wavenumber domain.

[0061] Example

[0062] Figure 1 This is a flowchart of a three-dimensional ultrasound imaging method based on a matrix array full-matrix capture in the frequency-wavenumber domain, according to an embodiment of the present invention.

[0063] like Figure 1 As shown, this embodiment provides a three-dimensional ultrasound imaging method based on full matrix capture of a matrix array in the frequency-wavenumber domain, including the following steps:

[0064] S10, after acquiring five-dimensional spatiotemporal domain radio frequency data, convert it to the frequency-wavenumber domain, specifically including the following sub-steps S11~S13:

[0065] S11, via such Figure 2 The two-dimensional full sampling matrix array shown performs full matrix capture on the medium to acquire five-dimensional spatiotemporal radio frequency data P(u). x u y v x v y ,t). Among them, (u x u y (v) represents the coordinates of the transmitting element. x v y ) represents the coordinates of the receiving array element, and t represents time.

[0066] In this step, the two-dimensional full sampling matrix array comprises multiple array elements arranged in a rectangular grid along the x and y axes. Each array element is sequentially activated as a transmitting element to emit an ultrasonic pulse. After each element emits, all elements in the two-dimensional full sampling matrix array (including the transmitting element itself) simultaneously act as receiving elements, recording the echo signal. This process is repeated until all elements have emitted at least one pulse.

[0067] S12, for five-dimensional spatiotemporal domain radio frequency data P(u x u y v xv y ,t) performs zero-padding in its five dimensions to improve spectral resolution.

[0068] S13, perform zero-filling on the five-dimensional spatiotemporal domain radio frequency data P(u) after step S12. x u y v x v y ,t) performs a five-dimensional fast Fourier transform (5D-FFT) to transform it from the spatiotemporal domain to the frequency-wavenumber domain to obtain spectral data:

[0069] .

[0070] In the above formula, P(k ux k uy k vx k vy ω) represents the frequency-wavenumber domain spectral data, k ux k uy k represents the transverse wave number emitted. vx k vy This indicates the received transverse wavenumber, and ω represents the angular frequency. This represents the five-dimensional Fourier transform operator.

[0071] S20, Stolt interpolation mapping, specifically includes the following sub-steps S21~S25:

[0072] S21, according to the Weyl identity, the three-dimensional Green's function from the origin to the spatial point (x, y, z) is expressed as:

[0073] .

[0074] In the above formula, k x k represents the wave number in the x-axis direction. y k represents the wave number in the y-axis direction. z This represents the wave number along the z-axis, where i represents the imaginary unit. .

[0075] S22, based on the Green's function, transmits ultrasonic waves from the transmitting array element (u x u y ) to the theoretical scattering point f theory (x, y, z) and return the receiving array element (v) x v y The complete pulse echo process can be represented by the following spatial integral:

[0076] .

[0077] S23, after converting the spatial integral to the frequency-wavenumber domain, the spectral data P(k) of the received signal is obtained. ux k uy k vx k vy ,ω) and the theoretical scattering point f theory The theoretical value F of the three-dimensional image spectrum data corresponding to (x, y, z) theory (k X k Y k Z The relationship between )

[0078] .

[0079] S24, Establish the image spectral wavenumber (k) based on the relationship in step S23. X k Y k Z ) and data spectrum wavenumber (k ux k uy k vx k vy The relationship between ω and ω:

[0080] k X =k ux +k vx .

[0081] k Y =k uy +k vy .

[0082] .

[0083] S25, adjust the spectrum data P(k) according to the relationship in step S24. ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis.

[0084] S30, the spectrum data P(k) after Stolt interpolation mapping in step S25 is processed. ux k uy k vx k vy Coherent superposition is performed by k, ω) X and k Y Summation in dimension (k) X=k ux +k vx k Y =k uy +k vy ), thereby synthesizing three-dimensional image spectral data F(k) X k Y k Z ).

[0085] S40, for the three-dimensional image spectral data F(k) X k Y k Z After performing a three-dimensional inverse fast Fourier transform (3D-IFFT), a three-dimensional ultrasound image in the spatial domain is reconstructed.

[0086] .

[0087] In the above formula, f(x, y, z) represents the reconstructed three-dimensional ultrasound image in the spatial domain. This represents the three-dimensional inverse fast Fourier transform.

[0088] Figure 3 This is a schematic diagram of the architecture of a matrix array full-matrix capture three-dimensional ultrasound imaging system based on the frequency-wavenumber domain in an embodiment of the present invention.

[0089] like Figure 3 As shown, this embodiment also provides a matrix array full-matrix capture three-dimensional ultrasound imaging system 100 based on the frequency-wavenumber domain. It uses the matrix array full-matrix capture three-dimensional ultrasound imaging method based on the frequency-wavenumber domain in the embodiment, including a two-dimensional full sampling matrix array 10, a frequency domain conversion module 20, a wavenumber mapping module 30, a superposition and synthesis module 40, an inverse transformation module, and a three-dimensional imaging module 50.

[0090] The two-dimensional full sampling matrix array 10 is used to perform full matrix capture of the medium according to the method in step S10, thereby acquiring five-dimensional spatiotemporal domain radio frequency data P(u x u y v x v y ,t).

[0091] Frequency domain conversion module 20 is connected to two-dimensional full sampling matrix array 10, and is used to convert five-dimensional spatiotemporal domain radio frequency data P(u) according to the method in step S10. x u y v x v y Perform a five-dimensional fast Fourier transform on t) to convert it to the frequency-wavenumber domain to obtain the spectral data P(k). ux k uy k vx kvy ,ω).

[0092] Wavenumber mapping module 30 is connected to frequency domain conversion module 20, and is used to map the angular frequency ω and the corresponding axial wavenumber k according to the method in step S20. Z Relationship For the spectral data P(k ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis.

[0093] The overlay synthesis module 40 is connected to the wavenumber mapping module 30, and is used to define the spatial wavenumber coordinates of the image as k according to the method in step S30. X =k ux +k vx And k Y =k uy +k vy Then, the spectral data after Stolt interpolation mapping by the wavenumber mapping module are coherently superimposed to synthesize the three-dimensional image spectral data F(k). X k Y k Z ).

[0094] The inverse transform module is connected to the 3D imaging module 50 and the superposition and synthesis module 40, and is used to process the 3D image spectral data F(k) according to the method in step S50. X k Y k Z After performing a three-dimensional inverse fast Fourier transform, a three-dimensional ultrasound image f(x, y, z) in the spatial domain is reconstructed.

[0095] Test case

[0096] Figure 4 This is a schematic diagram illustrating the principle of TFM-DAS beamforming under existing technology.

[0097] like Figure 4 As shown, in the existing technology, in order to reconstruct the amplitude of any voxel point (x, y, z) in the traditional TFM-DAS reconstruction method, the DAS algorithm needs to calculate the amplitude of the ultrasonic wave from each transmitting element (u). x u y The propagation reaches the voxel point and is then reflected back to each receiving element (v). x v yTotal flight time (TOF).

[0098] The TOF is given by the following formula:

[0099] .

[0100] Then, the DAS algorithm extracts the signal amplitude corresponding to time t from the acquired FMC data, and performs analysis on all N×N transmitting array elements and N×N receiving array elements (a total of N). 4 The signals from each channel are coherently superimposed to obtain the final amplitude I(x, y, z) of the voxel point:

[0101] .

[0102] This method repeats the process for every voxel in the imaging region, resulting in extremely high computational complexity.

[0103] This test case uses the frequency-wavenumber domain-based matrix array full-matrix capture three-dimensional ultrasound imaging system 100 described in the embodiment. The actual test is carried out according to the frequency-wavenumber domain-based matrix array full-matrix capture three-dimensional ultrasound imaging method described in the embodiment, and it is compared with the aforementioned prior art DAS method.

[0104] (1) Simulation experiment

[0105] The simulation was performed using the Verasonics Vantage system in simulation mode. A 32×32 matrix array was set up with a center frequency of 3.47 MHz. Five point target media were placed at axial distances ranging from 10 mm to 30 mm. The results are as follows: Figure 5 As shown.

[0106] Figure 5 This document presents the reconstructed images and quantitative comparison results of simulated point targets in the test examples of this invention. Part (a) represents the 3D image of the simulated point target reconstructed using the existing TFM-DAS method; parts (b), (c), and (d) represent the lateral-vertical slices, lateral-axial slices at y=0mm, and longitudinal-axial slices at x=0mm of the TFM-DAS reconstructed image, respectively; part (e) represents the lateral amplitude profile comparison curve at a depth of 20mm; part (f) represents the 3D image of the simulated point target reconstructed using the method of this embodiment; parts (g), (h), and (i) represent the lateral-vertical slices, lateral-axial slices at y=0mm, and longitudinal-axial slices at x=0mm of the reconstructed image, respectively; and part (j) represents a quantitative comparison histogram of the lateral half-peak full width (FWHM) of the two methods at different depths.

[0107] like Figure 5 As shown in section (f), the main lobe width of the point target reconstructed by the method (MA-FMC-fk) in the embodiment remains stable at different depths, while the main lobe of the DAS method widens with increasing depth. Figure 5 As can be seen from the quantitative FWHM (-6dB full width at half maximum) comparison in part (j), the average FWHM of DAS is 0.96mm, while the average of the method in the example is 0.72mm, which improves the lateral resolution by 25.00%.

[0108] (2) Phantom Experiment

[0109] Experiments were conducted using an 8×32 matrix array (center frequency 3.47MHz) and a CIRS Model 040GSE phantom.

[0110] Figure 6 This document presents the reconstructed images and quantitative comparison results of the phantom line target in the test examples of this invention. Part (a) represents the 3D image of the phantom line target reconstructed using the prior art TFM-DAS method; parts (b), (c), and (d) represent the lateral-longitudinal slices, axial-lateral slices, and axial-longitudinal slices at depth z=19.20mm, respectively, of the TFM-DAS reconstructed image; part (e) represents the lateral amplitude profile comparison curve at depth 28.59mm; part (f) represents the 3D image of the phantom line target reconstructed using the method of this embodiment; parts (g), (h), and (i) represent the lateral-longitudinal slices, axial-lateral slices, and axial-longitudinal slices at depth z=19.20mm, respectively, of the reconstructed image; and part (j) represents the quantitative comparison histogram of the lateral half-peak full width (FWHM) of the two methods at different depths.

[0111] like Figure 6 As shown, it displays the reconstruction results of the wire targets. Figure 6 The (f) and (h) parts in the text Figure 6 As shown in parts (a) and (c) of the document, the main lobe width of the method in this embodiment is significantly narrower than that of the DAS method. Figure 6 The quantitative analysis in section (j) shows that the average FWHM of DAS is 0.73 mm, while the average FWHM of the method in the example is 0.59 mm, representing a resolution improvement of 19.18%.

[0112] Figure 7This is a reconstructed image of the phantom cyst target in the test example of the present invention. Part (a) represents a three-dimensional image of the phantom cyst target reconstructed by the prior art TFM-DAS method; parts (b), (c), and (d) represent transverse-longitudinal slices, axial-transverse slices at depth z=14.90 mm, and axial-longitudinal slices at x=0 mm, respectively, of the image reconstructed by the TFM-DAS method; part (e) represents a three-dimensional image of the phantom cyst target reconstructed by the method of the embodiment of the present invention; parts (f), (g), and (h) represent transverse-longitudinal slices, axial-transverse slices at y=5.1 mm, and axial-longitudinal slices, respectively, of the image reconstructed by the method of the embodiment of the present invention.

[0113] like Figure 7 As shown, it displays the reconstruction results of the cyst target. (As shown...) Figure 7 As shown in section (c), the DAS method exhibits significant noise artifacts within the cyst, with blurred edges. Figure 7 As shown in section (g), the method of the embodiment effectively suppresses internal noise and produces clearer edges. As shown in Table 1, the method of the embodiment has a CNR (contrast-to-noise ratio) of 2.19 and an sSNR (spot signal-to-noise ratio) of 2.39; compared with the DAS method's 1.96 and 2.15, these represent improvements of 11.73% and 11.16%, respectively.

[0114] Table 1 (CR, CNR, sSNR and gCNR of experimental cyst phantom and human carotid artery)

[0115]

[0116] (3) In vivo experiments

[0117] The carotid arteries of healthy volunteers were scanned using the same equipment as in the phantom experiment.

[0118] Figure 8 These are reconstructed images of the in vivo carotid artery in a test example of the present invention. Part (a) represents a three-dimensional image of the in vivo carotid artery reconstructed using the prior art TFM-DAS method; parts (b) and (c) represent transverse-axial and longitudinal-axial slices of the image reconstructed using the TFM-DAS method, respectively; part (d) represents a three-dimensional image of the in vivo carotid artery reconstructed using the method of the present invention; and parts (e) and (f) represent transverse-axial and longitudinal-axial slices of the image reconstructed using the method of the present invention, respectively.

[0119] like Figure 8As shown in sections (d) and (e), the images reconstructed by the method of the embodiment exhibit smoother speckle patterns and clearer tissue boundaries. As shown in Table 1 above, the method of the embodiment has a CNR of 0.78 and an sSNR of 0.91; compared with the DAS method's 0.66 and 0.76, these represent improvements of 18.18% and 19.74%, respectively.

[0120] Furthermore, as shown in Table 1, the method of this embodiment performs comparably to the existing DAS method in terms of contrast ratio (CR) and generalized contrast-to-noise ratio (gCNR). Specifically, in the phantom cyst experiment, the CR of the embodiment method was 20.02 and the gCNR was 0.9674, which are essentially the same as the values ​​of the DAS method (21.59 and 0.9666); in the human carotid artery experiment, the CR of the embodiment method was 17.13 and the gCNR was 0.5662, which are also almost identical to the DAS method (16.71 and 0.5607). This indicates that the method of the embodiment can effectively maintain the original contrast characteristics of the tissue while significantly improving the image signal-to-noise ratio and edge sharpness.

[0121] (4) Computational efficiency

[0122] The computation time of the two algorithms was compared on a MATLAB workstation (Intel Xeon CPU, 1 TB RAM).

[0123] Figure 9 This is a graph comparing the computation time of the embodiment method in the test example of the present invention with that of the TFM-DAS method in the prior art.

[0124] like Figure 9 As shown, the computation time of the method in the embodiment (MA-FMC-fk) is significantly lower than that of the DAS method. The computational advantage of the method in the embodiment becomes more pronounced as the total number of voxels processed increases; for example, in 2... 18 When individual units were used, the speed increased by 4.2 times, in 2 22 When individual elements are used, the speed increases by 11.6 times.

[0125] In summary, the method of the embodiment (MA-FMC-fk) achieves a several-fold to ten-fold improvement in computational efficiency compared to the traditional DAS method, while significantly improving the spatial resolution, contrast-to-noise ratio, and speckle signal-to-noise ratio of the image, providing an effective technical approach for achieving high-resolution real-time three-dimensional ultrasound imaging.

[0126] Those skilled in the art should understand that this invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to this invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A three-dimensional ultrasound imaging method based on full matrix capture using a matrix array in the frequency-wavenumber domain, characterized in that, Includes the following steps: S10, five-dimensional spatiotemporal radio frequency data P(u) is acquired by performing full matrix capture of the medium through a two-dimensional full sampling matrix array. x u y v x v y After t), a five-dimensional fast Fourier transform is performed on it to convert it to the frequency-wavenumber domain, obtaining the spectral data P(k). ux k uy k vx k vy ,ω), Among them, (u x u y (v) represents the coordinates of the transmitting element. x v y ) represents the coordinates of the receiving array element, t represents time, and k represents the time. ux k uy k represents the transverse wave number emitted. vx k vy This indicates the received transverse wavenumber, and ω represents the angular frequency. S20, based on the angular frequency ω and the corresponding axial wave number k Z Relationship For the spectrum data P(k) ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis, Where c represents the speed of sound in the medium, k uz k represents the transmitted axial wave number. vz Indicates the received axial wavenumber; S30, define the spatial wavenumber coordinates of the image as: k X =k ux +k vx k Y =k uy +k vy The spectrum data obtained after Stolt interpolation mapping in step S20 are coherently superimposed to synthesize three-dimensional image spectrum data F(k). X k Y k Z ); S40, for the three-dimensional image spectrum data F(k) X k Y k Z After performing a three-dimensional inverse fast Fourier transform, a three-dimensional ultrasound image f(x, y, z) in the spatial domain is reconstructed.

2. The three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture according to claim 1, characterized in that: in, In step S10, the two-dimensional full sampling matrix array includes multiple array elements. The array elements are arranged in a rectangular grid pattern along the x and y axes. Multiple array elements sequentially transmit ultrasonic waves as transmitting array elements, and all of the array elements simultaneously receive echo signals as receiving array elements.

3. The three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture according to claim 1, characterized in that: in, In step S10, before performing the five-dimensional fast Fourier transform, the five-dimensional spatiotemporal domain radio frequency data P(u) is also processed. x u y v x v y ,t) is zero-padded in its five dimensions.

4. The three-dimensional ultrasound imaging method based on frequency-wavenumber domain matrix array full matrix capture according to claim 1, characterized in that: in, Step S20 includes the following sub-steps: S21, According to Weyl identity, the three-dimensional Green's function from the origin to the spatial point (x, y, z) can be expressed as: , Where, k x k represents the wave number in the x-axis direction. y k represents the wave number in the y-axis direction. z This represents the wave number along the z-axis, where i represents the imaginary unit. ; S22, based on the Green's function, the ultrasonic wave is scattered from the transmitting array element to the theoretical scattering point f. theory (x, y, z) and the complete pulse echo process returning the received array element is represented by the following spatial integral: ； S23, after converting the spatial integral to the frequency-wavenumber domain, the spectrum data P(k) of the received signal is obtained. ux k uy k vx k vy ,ω) and the theoretical scattering point f theory The theoretical value F of the three-dimensional image spectrum data corresponding to (x, y, z) theory (k X k Y k Z The relationship between ) ； S24, Establish the image spectral wavenumber (k) based on the relationship in step S23. X k Y k Z ) and data spectrum wavenumber (k ux k uy k vx k vy The relationship between ω and ω: k X =k ux +k vx , k Y =k uy +k vy , ； S25, adjust the spectrum data P(k) according to the relationship in step S24. ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis.

5. A matrix array full-matrix capture three-dimensional ultrasound imaging system based on the frequency-wavenumber domain, characterized in that, The method for three-dimensional ultrasound imaging based on a matrix array with full matrix capture in the frequency-wavenumber domain, as described in any one of claims 1 to 4, includes: The two-dimensional full sampling matrix array is used to perform full matrix capture of the medium to acquire five-dimensional spatiotemporal radio frequency data P(u x u y v x v y ,t); The frequency domain conversion module is used to convert the five-dimensional spatiotemporal radio frequency data P(u x u y v x v y Perform a five-dimensional fast Fourier transform on t) to convert it to the frequency-wavenumber domain to obtain the spectral data P(k). ux k uy k vx k vy ,ω); The wavenumber mapping module is used to map the angular frequency ω and the corresponding axial wavenumber k. Z Relationship For the spectrum data P(k) ux k uy k vx k vy ,ω) performs Stolt interpolation mapping, thereby transforming each sample point (k) into a single sample. ux k uy k vx k vy Data along the ω-axis is resampled to a uniformly distributed k. Z On the axis; The overlay and synthesis module is used to define the spatial wavenumber coordinates of the image as k. X =k ux +k vx And k Y =k uy +k vy Then, the spectral data after Stolt interpolation mapping by the wavenumber mapping module are coherently superimposed to synthesize three-dimensional image spectral data F(k). X k Y k Z );as well as The inverse transform module and the three-dimensional imaging module are used to process the three-dimensional image spectral data F(k) X k Y k Z After performing a three-dimensional inverse fast Fourier transform, a three-dimensional ultrasound image f(x, y, z) in the spatial domain is reconstructed.

Images