WO2021233178A1 - 超声造影图像的超分辨重建预处理方法和超分辨重建方法 - Google Patents

超声造影图像的超分辨重建预处理方法和超分辨重建方法 Download PDF

Info

Publication number
WO2021233178A1
WO2021233178A1 PCT/CN2021/093327 CN2021093327W WO2021233178A1 WO 2021233178 A1 WO2021233178 A1 WO 2021233178A1 CN 2021093327 W CN2021093327 W CN 2021093327W WO 2021233178 A1 WO2021233178 A1 WO 2021233178A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
pixel
super
preprocessed
pixel point
Prior art date
Application number
PCT/CN2021/093327
Other languages
English (en)
French (fr)
Inventor
尹静宜
张珏
Original Assignee
南京超维景生物科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 南京超维景生物科技有限公司 filed Critical 南京超维景生物科技有限公司
Priority to EP21808993.6A priority Critical patent/EP4060616B1/en
Publication of WO2021233178A1 publication Critical patent/WO2021233178A1/zh
Priority to US17/837,973 priority patent/US11798150B2/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/003—Reconstruction from projections, e.g. tomography
    • G06T11/005—Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/06—Measuring blood flow
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/48—Diagnostic techniques
    • A61B8/481—Diagnostic techniques involving the use of contrast agent, e.g. microbubbles introduced into the bloodstream
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/5215—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data
    • A61B8/5223—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data for extracting a diagnostic or physiological parameter from medical diagnostic data
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/5269—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving detection or reduction of artifacts
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
    • G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/68—Analysis of geometric attributes of symmetry
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10132—Ultrasound image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30101—Blood vessel; Artery; Vein; Vascular
    • G06T2207/30104—Vascular flow; Blood flow; Perfusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30241—Trajectory
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2210/00—Indexing scheme for image generation or computer graphics
    • G06T2210/41—Medical
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2211/00—Image generation
    • G06T2211/40—Computed tomography
    • G06T2211/404—Angiography
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2211/00—Image generation
    • G06T2211/40—Computed tomography
    • G06T2211/416—Exact reconstruction

Definitions

  • This application relates to the field of ultrasound imaging technology, and in particular to a super-resolution reconstruction pre-processing method, super-resolution reconstruction method, super-resolution reconstruction pre-processing device, super-resolution reconstruction device, electronic equipment, and computer-readable storage medium of ultrasound contrast images.
  • Ultra-micro blood flow imaging technology-Ultrasound Localization Microscope overcomes the impact of acoustic diffraction, and achieves super-resolution imaging of tiny blood vessels.
  • high-concentration microbubbles that can shorten the acquisition time are often used in clinics.
  • the spatial coupling between high-concentration microbubbles affects the accuracy of super-resolution reconstruction.
  • the interference of strong noise and tissue background signals cannot be avoided, and the accuracy of ultrasound microbubble positioning is affected.
  • the embodiments of the present application provide a super-resolution reconstruction pre-processing method, super-resolution reconstruction method, super-resolution reconstruction pre-processing device, super-resolution reconstruction device, electronic equipment, and computer-readable storage medium for ultrasound contrast images.
  • an embodiment of the present application provides a super-resolution reconstruction pre-processing method for contrast-enhanced ultrasound images. Quasi-contrast image; Obtain the gray-scale fluctuation signal of the pixel in the registration contrast image to be preprocessed; Denoise and reconstruct the to-be-preprocessed image set based on the gray-scale fluctuation signal of the co-located pixel set
  • the co-located pixel point set includes: a plurality of co-located pixel points located at the same pixel coordinates in the to-be-preprocessed registration contrast image in different frames; and based on the co-located pixel point set and the The gray-scale fluctuation signal of the associated pixel point set associated with the co-located pixel point set performs interpolation calculation on the reconstructed feature parameter image to obtain a sparse image
  • the associated pixel point set includes: the to-be-preprocessed configuration in the same frame A plurality of associated
  • an embodiment of the present application provides a super-resolution reconstruction method of ultrasound contrast images, including: selecting at least one image set to be preprocessed from the contrast data; Performing preprocessing separately to obtain at least one frame of thinned image, the preprocessing method adopts any one of the preprocessing methods described above; obtaining the pixel value and the radial symmetry estimated value of the pixel in the thinned image; Performing weighted calculation on the pixel values and the radial symmetry estimation values of the pixels in the thinned out image in the same frame to obtain at least one frame of local super-resolution image corresponding to the at least one thinned out image; and The at least one frame of local super-resolution image is superimposed to obtain a reconstructed super-resolution image.
  • an embodiment of the present application provided by an embodiment of the present application provides a super-resolution reconstruction preprocessing device for ultrasound contrast images, including: a first preprocessing acquisition module configured to acquire An image set, the image set to be preprocessed includes multiple frames of preprocessed registration radiographic images; a grayscale acquisition module configured to acquire the grayscale fluctuation signals of pixels in the preprocessed registration radiographic images; denoising The enhanced reconstruction module is configured to perform denoising and reconstruction on the image set to be preprocessed based on the gray-scale fluctuation signal of the co-located pixel point set to obtain a reconstructed feature parameter image, and the co-located pixel point set includes: A plurality of co-located pixel points located at the same pixel coordinates in the registration radiographic image to be preprocessed; and a thinning module configured to be based on the co-located pixel point set and an associated pixel point set associated with the co-located pixel point set Perform interpolation calculation on the
  • an embodiment of the present application provides a super-resolution reconstruction device for contrast-enhanced ultrasound images.
  • the reconstruction device includes: a first selection module configured to select from the contrast data At least one image set to be preprocessed; a preprocessing device configured to use the super-resolution reconstruction preprocessing method as described in any one of the above, to perform preprocessing on the at least one image set to be preprocessed to obtain at least one frame sparseness Image; the axial trajectory highlight module, configured to obtain the pixel value and the radial symmetry estimated value of the pixel in the thinned image; and the pixel value and the radial symmetry of the pixel located in the thinned image in the same frame
  • the symmetry estimation value is weighted and calculated to obtain at least one frame of local super-resolution image corresponding to the at least one frame of sparse image; and a superimposing module for superimposing the at least one frame of local super-resolution image to obtain a reconstructed super-resolution image
  • an embodiment of the present application provided by an embodiment of the present application provides an electronic device, including: a processor; a memory; and computer program instructions stored in the memory.
  • the computer program instructions are being processed.
  • the processor executes the super-resolution reconstruction pre-processing method described in any one of the above or the super-resolution reconstruction method described in any one of the above.
  • an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored on the computer-readable storage medium, and the computer program instructions cause the The processor executes the super-resolution reconstruction preprocessing method described in any one of the above or the super-resolution reconstruction method described in any one of the above.
  • the embodiment of the application provides a super-resolution reconstruction pre-processing method, super-resolution reconstruction method, super-resolution reconstruction pre-processing device, super-resolution reconstruction device, electronic equipment, and computer-readable storage medium for ultrasound contrast images.
  • the reconstructed feature parameter image with enhanced microbubble signal and weakened background noise signal is obtained; the reconstructed feature is reconstructed by the similarity of the gray-scale fluctuation signal of the co-located pixel point set and the associated pixel point set associated with the co-located pixel point set
  • Interpolation of parametric images plays a role in separating different microbubbles, realizing spatial decoupling of overlapping microbubbles, and solving the overlap of microbubbles caused by the diffraction limit.
  • the impact of high concentration of microbubbles and strong noise on the accuracy of super-resolution imaging can be effectively reduced.
  • the point spread function of the microbubble is reduced, the deformation information of the microbubble is retained, and the motion axis of the microbubble in each frame of the thinned image
  • the trajectory skeleton of the direction becomes clearer, which enhances the non-localized motion axis of the microbubbles, retains the trajectory information of the microbubbles along the direction of motion, obtains local super-resolution images, and finally integrates them into a complete ultrasound super-resolution reconstruction of small blood vessels image.
  • the non-positioning motion axis enhancement method retains the trajectory skeleton of the microbubble motion axis, thereby greatly improving the spatial resolution and reconstruction speed of the super-resolution reconstructed image, achieving rapid Efficient reconstruction of super-resolution images.
  • Fig. 1 is a schematic flow chart of a preprocessing method for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • FIG. 2 is a schematic diagram of the process of obtaining reconstructed characteristic parameter images in a pre-processing method for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • FIG. 3 is a schematic diagram of the process of obtaining a sparse image by interpolation calculation in a preprocessing method for super-resolution reconstruction of a contrast-enhanced ultrasound image according to an embodiment of the application.
  • FIG. 4 shows a method for super-resolution reconstruction of contrast-enhanced ultrasound images provided by an embodiment of the application.
  • FIG. 5 is a schematic diagram of the process of obtaining the estimated value of the radial symmetry of the pixels in the thinned image in a super-resolution reconstruction method of a contrast-enhanced ultrasound image according to an embodiment of the application.
  • FIG. 6 is a schematic flowchart of selecting at least one image set to be preprocessed from the contrast data in a super-resolution reconstruction method for ultrasound contrast images provided by an embodiment of the application.
  • Fig. 7 shows a flow chart of a method for super-resolution reconstruction of the biceps biceps of the lower limbs of a New Zealand white rabbit according to an embodiment of the application.
  • FIG. 8 is a schematic diagram of a pre-processing device for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • FIG. 9 is a schematic diagram of a pre-processing device for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • FIG. 10 is a schematic diagram of a super-resolution reconstruction device for contrast-enhanced ultrasound images according to an embodiment of the application.
  • FIG. 11 is a schematic diagram of a super-resolution reconstruction device for contrast-enhanced ultrasound images according to an embodiment of the application.
  • FIG. 12 is a schematic structural diagram of an electronic device provided by an embodiment of the application.
  • Microvascular imaging is of great benefit to the diagnosis of diseases.
  • the ultrasound microbubble contrast agent invented by Gramiak in the 1970s made ultrasound vascular imaging possible.
  • Microbubbles can flow in blood vessels and vibrate in a non-linear manner under ultrasound to produce specific harmonic characteristics, enabling them to distinguish blood vessels from background tissue signals in deeper tissues with higher sensitivity.
  • contrast-enhanced ultrasound CEUS
  • CEUS contrast-enhanced ultrasound
  • Errico et al. proposed an ultra-microscopic blood flow imaging technology based on contrast-enhanced ultrasound—ultrasound positioning microscope, which overcomes the influence of acoustic diffraction by positioning and time accumulation of single microbubbles, and realizes the detection of tiny blood vessels with a diameter of tens of microns. Super-resolution imaging.
  • the basic idea of this application is to propose a pre-processing method for super-resolution reconstruction of ultrasound contrast images, by registering the gray level of the same pixel point set in the contrast image with multiple frames to be preprocessed in the time scale.
  • the non-localized motion axis is enhanced by the radial symmetry of the pixels, and the trajectory skeleton of the microbubble motion axis is retained, thereby greatly improving the super-resolution
  • the spatial resolution and reconstruction speed of the reconstructed image can realize fast and efficient super-resolution image reconstruction.
  • Fig. 1 shows a preprocessing method for super-resolution reconstruction of contrast-enhanced ultrasound images provided by an embodiment of the application.
  • the preprocessing method includes:
  • Step 101 Obtain a set of images to be preprocessed, and the set of images to be preprocessed includes multiple frames of registered contrast images to be preprocessed;
  • the target to be imaged uses an ultrasound probe to detect the area to be imaged, start to collect contrast data when a contrast enhancement signal appears, and obtain multiple frames of contrast images; register multiple frames of contrast images to suppress the movement of the probe and tissue After the image changes, the multi-frame registration image is obtained, that is, the registration contrast image; the selected part from the multi-frame registration contrast image allocates the registration contrast image as the registration contrast image to be preprocessed; therefore, each frame needs to be preprocessed and registered
  • the contrast images are all contrast images after registration.
  • the frequency of collecting multiple frames of contrast images may be 30 Hz or 50 Hz, and the frequency of collecting contrast images only needs to meet specific clinical acquisition conditions.
  • the frequency of collecting contrast images is not specifically limited in the embodiment of the present application.
  • the selection method of selecting the quasi-contrast image by the selection part from the multi-frame registration contrast image can be 4 frames every 1 frame; 3 frames every 2 frames; 5 frames every 1 frame.
  • the specific selection method of the selection part to allocate the quasi-contrast image from the multi-frame registration angiographic image is not limited.
  • the registration method for registering the multi-frame contrast images may be to use the Morphon multi-scale registration method to register the multi-frame contrast images, and the specific method of registration is not limited in the embodiment of the present application.
  • the contrast medium injection method can be to dissolve 59 mg of sulphur hexafluoride microbubble lyophilized powder Sonovi into 5mL 0.9% sodium chloride solution to prepare an ultrasound contrast medium, and inject 0.2mL into a 3L chlorinated solution. In the sodium solution model, or continuous injection at a rate of 0.2-5.0 ⁇ L/min, the specific types of contrast agents and specific injection methods are not limited in the embodiments of the present application.
  • Step 102 Obtain gray-scale fluctuation signals of pixels in the registration contrast image to be preprocessed
  • each frame of contrast image has divided pixel coordinates, and the pixel coordinates have pixel points. Since the multi-frame radiography is collected by the same device and each frame of the registered radiographic image is registered using the same registration parameters, then the pixel coordinate division on the registered radiographic image of each frame is the same, and each frame needs to be preprocessed and registered The pixel coordinate division on the contrast image is also the same.
  • the gray-scale fluctuation signal is used to indicate the change of the gray-scale fluctuation of the current pixel in the current frame. Condition.
  • the specific method for obtaining the gray-scale fluctuation signal is not specifically limited in the embodiment of the present application.
  • Step 103 Perform denoising and reconstruction on the image set to be preprocessed based on the gray level fluctuation signal of the co-located pixel point set to obtain a reconstructed feature parameter image. Multiple co-located pixels at pixel coordinates;
  • each frame of the to-be-preprocessed registration contrast image in a to-be-preprocessed image set has the same pixel coordinates. Pixels located at the same pixel coordinates in the registration radiographs to be preprocessed in different frames are co-located pixels.
  • the gray-scale fluctuation signal of the pixels of the contrast image to be preprocessed is a one-dimensional signal, that is, the gray-scale fluctuation signal of each co-located pixel is a one-dimensional gray-scale fluctuation signal; in different frames
  • the gray-scale fluctuation signal of the co-located pixel point set at the same pixel coordinates in the pre-processed registration contrast image is a 1-times multi-dimensional gray-scale fluctuation signal, that is, a co-located pixel set is a 1-times multi-dimensional gray-scale fluctuation signal
  • the 1-time multi-dimensional gray-scale fluctuation signal is used to reflect the periodicity of the gray-scale fluctuation and the randomness of the gray-scale fluctuation of the image set to be preprocessed in the current time window.
  • the gray-scale fluctuation signal of the microbubble Since the periodicity and distribution of the gray-scale fluctuation signal corresponding to the contrast agent microbubbles and the background or noise are different, the gray-scale fluctuation signal of the microbubble has stronger periodicity and stronger randomness, based on multiple co-located pixels
  • the 1-time multi-dimensional gray-scale fluctuation signal of one point can distinguish the microbubble signal from the noise or background signal, which can enhance the microbubble signal and weaken the background or noise signal, and finally obtain a microbubble signal enhanced and background or noise signal
  • the weakened reconstruction feature parameter image Since the periodicity and distribution of the gray-scale fluctuation signal corresponding to the contrast agent microbubbles and the background or noise are different, the gray-scale fluctuation signal of the microbubble has stronger periodicity and stronger randomness, based on multiple co-located pixels
  • the 1-time multi-dimensional gray-scale fluctuation signal of one point can distinguish the microbubble signal from the noise or background signal, which can enhance the microbubble signal and weaken the background or noise
  • a set of images to be preprocessed within a 5-frame time window contains five frames of A, B, C, D and E to be preprocessed registration contrast images
  • the pixel coordinate is a pixel at the top
  • the (1, 1) pixel coordinate of the B frame to be preprocessed and registered contrast image is b pixel at the top
  • the C frame is the (1, 1) pixel of the to be preprocessed registration contrast image
  • the coordinates are at the c pixel point
  • the (1, 1) pixel coordinates of the D frame to be preprocessed registration contrast image are at the d pixel point
  • the E frame is at the (1, 1) pixel coordinates of the registration contrast image to be preprocessed
  • the above is the e pixel, then a, b, c, d, and e are the set of co-located pixels.
  • the 1-by 5-dimensional gray-scale fluctuation signal based on the composition of a, b, c, d, and e can distinguish the microbubble signal from the noise or background signal in the image set to be preprocessed in the current time window, and improve the signal-to-noise ratio Contrast with letter.
  • the number of co-located pixel point sets in the image set to be preprocessed is the same as the number of identical pixel coordinates, and the number of identical pixel coordinates in the image set to be preprocessed is determined by the device that collects the contrast image.
  • Step 104 Perform interpolation calculation on the reconstructed feature parameter image based on the gray-scale fluctuation signals of the co-located pixel point set and the associated pixel point set associated with the co-located pixel point set to obtain a sparse image.
  • the associated pixel point set includes: in the same frame A plurality of associated pixels located at the same pixel coordinates in the registration radiographic image to be preprocessed are adjacent to the same pixel in the radiographic image to be preprocessed and located at the same pixel coordinates in the registration radiographic image to be preprocessed.
  • the diffraction limit Due to the existence of the diffraction limit, the microbubbles will overlap on each frame of the pre-processed registration contrast image.
  • Each associated pixel in the associated pixel point set is also located at the same pixel coordinate in different frames of the pre-processed registration angiographic image, that is to say, the associated pixel point set itself is also a group of co-located pixels, only relative to the selected current
  • the co-located pixel set, each pixel of the associated pixel set and each co-located pixel of the current co-located pixel set are adjacent and have the same relative position in the same frame to be preprocessed registration contrast image.
  • the gray-scale fluctuation signal of the co-located pixel is a one-dimensional gray-scale fluctuation signal, and the selected current co-located pixel set is a 1-times multi-dimensional continuous gray-scale fluctuation signal;
  • the gray-scale fluctuation signal of each associated pixel is also A one-dimensional gray-scale fluctuation signal,
  • the associated pixel set is also a 1-times multi-dimensional continuous gray-scale fluctuation signal, and
  • the associated pixel set is a 1-times multi-dimensional continuous gray-scale fluctuation signal located near the current co-located pixel set, using two
  • the similarity of a one-time multi-dimensional continuous gray-scale fluctuation signal can be used to separate different microbubbles by interpolating the reconstructed characteristic parameter image.
  • the (1, 1) pixel coordinates of the A frame to be pre-processed registration contrast image are a pixel at the top, the A frame (0, 1) pixel at the top is a1 pixel, and the A frame (1, 2) pixel
  • the coordinates are a2 pixel points
  • the (1, 1) pixel coordinates of the B frame to be pre-processed registration angiography image are b pixels
  • the B frame (0, 1) pixel coordinates are b1 pixels
  • the pixel coordinates are b2 pixels
  • the (1, 1) pixel coordinates of the C frame to be preprocessed registration contrast image are c pixels
  • the C frame (0, 1) pixel coordinates are c1 pixel
  • C frame (1, 2) pixel coordinates are c2 pixels
  • D frame to be pre-processed registration contrast image (1, 1) pixel coordinates are d pixels, D frame (0, 1) )
  • the pixel coordinate is d1 pixel
  • the pixel coordinates of E frame (0, 1) are e1 pixels, and the pixel coordinates of E frames (1, 2) are e2 pixels; then a, b, c, d and e are the selected current co-located pixels Set; a1, b1, c1, d1, and e1 are a set of associated pixels; a2, b2, c2, d2, and e2 are also a set of associated pixels.
  • the number of associated pixel point sets corresponding to each co-located pixel point set can be 4, 6, 8, 10, and 12.
  • the embodiment of the present application does not specify the number of associated pixel point sets corresponding to each co-located pixel point set. limited.
  • the microbubble signals of the to-be-preprocessed image set in the current time window will be compared with Distinguish between noise and background signal, improve the signal-to-noise ratio and signal-to-background ratio, and obtain the reconstructed characteristic parameter image with enhanced microbubble signal and weakened background noise signal; through the co-located pixel set and the associated pixel set associated with the co-located pixel set Interpolating the reconstructed characteristic parameter image based on the similarity of the gray-scale fluctuation signal can play a role in separating different microbubbles, can realize spatial decoupling of overlapping microbubbles, and solve the overlap of microbubbles caused by the diffraction limit.
  • the gray-scale fluctuation signal is a signal formed by arranging the pixel values of the pixels in chronological order; the gray-scale fluctuation signal of the co-located pixel set is the pixel value of multiple co-located pixels in the co-located pixel set according to 1 times multi-dimensional signal formed by chronological arrangement. Because the distribution of microbubbles and background or noise is different, the periodicity of microbubbles and background or noise is also different. The periodicity of microbubble signals is stronger, and the distribution of microbubble signals is more random.
  • the characteristics of gray-scale fluctuation signals are The randomness of the gray-scale fluctuation distribution and the periodicity of the gray-scale fluctuations, then the use of the gray-scale fluctuation distribution characteristics and the gray-scale fluctuation period characteristics of the pixels in the same position can distinguish the microbubble signal from the background or noise signal. Interpolating the similarity between the 1-time multi-dimensional signal formed by the co-located pixel point set and the 1-time multi-dimensional signal formed by the surrounding pixel point set can realize the spatial decoupling of the overlapping microbubbles.
  • FIG. 2 is a schematic diagram of the process of obtaining reconstructed characteristic parameter images in a pre-processing method for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • the denoising and reconstruction of the image set to be preprocessed based on the gray-level fluctuation signal of the co-located pixel point set may specifically include:
  • Step 1031 Select a plurality of first co-located pixel points located at the first pixel coordinates from multiple frames of to-be-preprocessed registration angiographic images to form a first co-located pixel point set;
  • the first co-located pixel points at the first pixel coordinates of each frame of the to-be-preprocessed registration contrast image are selected to form a first co-located pixel point set.
  • A, b, c, d, and e at the pixel coordinates of (1, 1) in each of the five frames of A, B, C, D, and E to be pre-processed registration contrast images form the first A set of pixels in the same position.
  • the first pixel coordinate is only a reference, the first pixel coordinate may be any pixel coordinate point in the image set to be preprocessed, and the specific selection of the first pixel coordinate is not limited in the embodiment of the present application.
  • Step 1032 Perform feature estimation and extraction on the gray-scale fluctuation signals of the first co-located pixels to obtain a first feature parameter.
  • the first feature parameter is used to characterize the random gray-scale fluctuation distribution of the first co-located pixel.
  • the first feature parameter is extracted by a plurality of first co-located pixels, and the first feature parameter is used to characterize the randomness of the gray-scale fluctuation distribution of the multiple co-located pixels, and the randomness of the gray-scale fluctuation signal of the pixels belonging to the microbubble Higher, the randomness of the gray-scale fluctuation signal of pixels belonging to noise or background is lower;
  • the first characteristic parameter is used to characterize the periodicity of gray-scale fluctuations of multiple co-located pixels, and the gray of pixels belonging to microbubbles
  • the periodicity of the fluctuation period is stronger, and the periodicity of the gray-scale fluctuation of the pixels belonging to the system and the background is lower, that is, the period of the gray-scale fluctuation of the pixels between the system and the background is weaker.
  • Step 1033 preset the mapping relationship between the first feature parameter and the reconstructed feature parameter image
  • first characteristic parameters can reflect the gray-scale fluctuation distribution characteristics and gray-scale fluctuation period characteristics of the entire image set to be preprocessed, and the periodicity of the gray-scale fluctuation signal corresponding to the contrast agent microbubbles and the background or noise And the distribution conditions are different.
  • the microbubble signal can be distinguished from the background or noise signal by using the gray-scale fluctuation distribution characteristics and the gray-scale fluctuation period characteristics. Then the microbubble signal can be distinguished from the background or noise signal according to the first characteristic parameter. .
  • the mapping relationship between the first feature parameter and the reconstructed feature parameter image is preset. According to the mapping relationship between the extracted first feature parameter and the reconstructed feature parameter image, the microbubble signal can be reconstructed while the background or noise is weakened. Reconstruct the characteristic parameter image.
  • the value of the first feature parameter is preset to be large in advance, it indicates that the gray level fluctuation period of the corresponding co-located pixel point set is stronger, and the probability of each co-located pixel point belonging to microbubbles is greater, then the corresponding feature to be reconstructed
  • the pixel gray value of the pixel in the parametric image should be high, and the value of the first feature parameter is preset to be small in advance, indicating that the gray fluctuation of the corresponding co-located pixel set is weakly periodic, and each co-located pixel belongs to the background or noise If the probability is greater, then the pixel gray value of the pixel in the corresponding reconstructed feature parameter image is lower.
  • mapping relationship between the first feature parameter and the reconstructed feature parameter image is preset in advance, the reconstructed feature parameter image in which the microbubble signal is strengthened but the background or noise is weakened can be reconstructed. It is assumed that the specific implementation of the mapping relationship is not limited.
  • Step 1034 Correspond the first feature parameter to the first pixel coordinate of the feature parameter image to be reconstructed, and obtain the reconstructed feature parameter image according to the mapping relationship.
  • Reconstructed characteristic parameter image According to the gray-scale fluctuation distribution characteristics and gray-scale fluctuation period characteristics reflected by the first characteristic parameter of the image set to be preprocessed, the reconstructed microbubble signal is strengthened while the background or noise is weakened ( The signal-to-noise ratio and signal-to-background ratio have been improved). Correspond the first feature parameter to the first pixel coordinate of the feature parameter image to be reconstructed. According to the mapping relationship, the microbubble signal needs to be strengthened and the background signal needs to be weakened. The coordinate alignment is accurately conveyed to achieve most of the background. And the removal of noise interference.
  • the microbubble signal is distinguished from the background or noise signal; Using the pre-preset mapping relationship between the first feature parameter and the reconstructed feature parameter image, the microbubble area signal needs to be enhanced and the background area signal needs to be weakened.
  • the coordinates are accurately communicated to the corresponding feature parameter image to be reconstructed. At the position, reconstruct the reconstructed characteristic parameter image with improved signal-to-noise ratio and signal-to-background ratio.
  • the method of feature estimation and extraction includes: autocorrelation estimation calculation or information entropy estimation calculation.
  • the autocorrelation estimation is used to measure the periodicity of the gray fluctuation signal of each co-located pixel; the randomness of the gray fluctuation signal of each co-located pixel is measured by the information entropy estimation.
  • the first feature parameter is obtained by autocorrelation estimation calculation or information entropy estimation calculation, and the multiple first feature parameters are used to reflect the gray-scale fluctuation distribution characteristics and the gray-scale fluctuation period characteristics of the entire image set to be preprocessed.
  • the microbubble signal in the image set to be preprocessed in the current time window is distinguished from the noise or background signal, the signal-to-noise ratio and the signal-to-background ratio are improved, and the reconstructed characteristic parameter image with enhanced microbubble signal and weakened background noise signal is obtained.
  • FIG. 3 is a schematic diagram of the process of obtaining a sparse image by interpolation calculation in a preprocessing method for super-resolution reconstruction of a contrast-enhanced ultrasound image according to an embodiment of the application.
  • performing interpolation calculation on the reconstructed feature parameter image to obtain a sparse image includes:
  • Step 1041 Select a plurality of second co-located pixel points located at the second pixel coordinates from multiple frames of to-be-preprocessed registration contrast images to form a second co-located pixel point set;
  • the second co-located pixel points at the second pixel coordinates of each frame to be pre-processed and registered contrast image are selected to form a second co-located pixel point set.
  • A, b, c, d, and e at the pixel coordinates of (1, 1) in each of the five frames of A, B, C, D, and E to be pre-processed registration contrast images form the first Two sets of co-located pixels.
  • the second pixel coordinate is only a reference, and it can be any pixel coordinate point in the image set to be preprocessed.
  • the second pixel coordinate may be the same coordinate as the first pixel coordinate, or may not be the same coordinate as the first pixel coordinate. This application The embodiment does not limit the specific selection of the second pixel coordinates.
  • Step 1042 Select multiple associated pixel points located at the associated pixel coordinates adjacent to the second pixel coordinate from the multi-frame to-be-preprocessed registration contrast image to form an associated pixel point set;
  • each frame of the registered contrast image to be preprocessed a plurality of associated pixel points located at the associated pixel coordinates adjacent to the second pixel coordinates are selected.
  • There is more than one associated pixel coordinate corresponding to the second pixel coordinate and the number of associated pixel point sets corresponding to each co-located pixel point set is more than one.
  • the number is not specifically limited. In the following, it is taken as an example that there are four associated pixel coordinates adjacent to the second pixel coordinate, that is, four associated pixel point sets corresponding to a co-located pixel point set.
  • the multiple associated pixel coordinates at the (1, 1) pixel coordinates of the A-frame to be pre-processed registration contrast image are (0, 1), (1, 2), (2, 1), and (1, 0)
  • the associated pixels of the second pixel a are a1, a2, a3, and a4 located at (0, 1), (1,2), (2, 1), and (1, 0); in the same way, frame B
  • the associated pixels of the second pixel b in the middle are b1, b2, b3, and b4;
  • the associated pixels of the second pixel c in the C frame are c1, c2, c3, and c4;
  • the pixels are d1, d2, d3, and d4;
  • the associated pixels of the second pixel e in the E frame are e1, e2, e3, and e4.
  • a, b, c, d, and e are co-located pixel point sets; a1, b1, c1, d1, and e1 are an associated pixel point set, a2, b2, c2, d2, and e2 are an associated pixel point set; a3, b3, c3, d3, and e3 are a set of related pixels; a4, b4, c4, d4, and e4 are a set of related pixels.
  • the set of co-located pixels (a, b, c, d, and e) is a 1-by 5-dimensional grayscale fluctuation signal;
  • the associated pixel set (a1, b1, c1, d1, and e1) is a 1-by 5-dimensional grayscale Fluctuation signal; in the same way, (a2, b2, c2, d2, and e2), (a3, b3, c3, d3, and e3) and (a4, b4, c4, d4, and e4) are all a one-by-five-dimensional gray Degree fluctuation signal.
  • Step 1043 Perform similarity quantification on the gray-scale fluctuation signal of the second co-located pixel point set and the gray-scale fluctuation signal of the associated pixel point set to obtain a similarity evaluation value;
  • Step 1044 Insert the similarity estimation value into the corresponding interpolation pixel coordinate in the reconstructed feature parameter image, and the interpolation pixel coordinate is located between the second pixel coordinate and the associated pixel coordinate;
  • the similarity between the co-located pixel point set (a, b, c, d, and e) and the associated pixel point set (a2, b2, c2, d2, and e2) is the similarity evaluation value Z2, and Z2 is inserted into the re Between the pixel coordinates (1, 1) and (1, 2) of the feature parameter image; the co-located pixel point set (a, b, c, d and e) and the associated pixel point set (a3, b3, c3, d3) And e3) the similarity to obtain the similarity estimate Z3, insert Z3 between the pixel coordinates (1, 1) and (2, 1) of the reconstructed feature parameter image; set the co-located pixel points (a, b, c, d and e) The similarity between d and e) and the associated pixel point set (a4, b4, c4, d4, and e4) obtains the similarity evaluation value Z4, and inserts Z4 into the pixel coordinates (1,
  • Step 1045 Use the position of the interpolated pixel coordinates to spatially decouple the overlapped microbubbles in the reconstructed feature parameter image to obtain a sparse image.
  • the similarity calculation is performed by the gray-scale fluctuation signals of the co-located pixel point set and the associated pixel point set, the similarity of the two signals is changed into a more intuitive similarity evaluation value, and the similarity evaluation value is inserted Reconstruct the corresponding interpolation pixel coordinates in the feature parameter image so that the intuitive similarity estimation value is presented between the pixels in the reconstructed feature parameter image.
  • the pixels with high similarity are attributed to the same microbubble, and the similarity is The pixels that are not high are separated into different microbubbles, and interpolation is used to separate the different microbubbles, so that the overlapping microbubbles are spatially decoupled, and finally a sparse image is obtained.
  • the method of similarity quantification includes: cross-entropy estimation or cross-correlation estimation. Quantitatively calculate the similarity between the two gray-scale fluctuation signals of the co-located pixel point set and the associated pixel point set through cross-entropy estimation or cross-correlation estimation, and turn the similarity of the two signals into a more intuitive similarity estimate , The similarity value is inserted into the reconstructed feature parameter image, and the overlapped microbubbles are spatially decoupled using intuitive values, and finally a sparse image is obtained.
  • the number of associated pixel point sets is 4. Select 4 associated pixel points located at the 4 associated pixel coordinates adjacent to the second pixel coordinate from the multi-frame to-be-preprocessed registration angiography image to form 4 associated pixel point sets corresponding to the co-located pixel point set; Calculate the similarity between the co-located pixel point set and each associated pixel point set in the 4 associated pixel point sets, and obtain a more detailed similarity between a pixel point and the surrounding associated pixels, so that the boundary of the interpolated overlapping microbubbles is more Clear, spatially decouple the overlapped microbubbles in the reconstructed feature parameter image to obtain a sparse image.
  • FIG. 4 shows a method for super-resolution reconstruction of contrast-enhanced ultrasound images provided by an embodiment of the application.
  • the super-resolution reconstruction method includes:
  • Step 401 Select at least one image set to be preprocessed from the contrast data
  • the contrast data is the collected multi-frame contrast images.
  • the multi-frame quasi-contrast image at least one image set to be preprocessed is selected, so that each image set to be preprocessed is continuous with each other, and each image set to be preprocessed
  • the microbubbles have a stable spatial position. As long as the set of images to be preprocessed meets the above conditions, the embodiment of the present application does not limit the specific selection method.
  • Step 402 Perform pre-processing on at least one image set to be pre-processed separately according to any of the foregoing super-resolution reconstruction pre-processing methods to obtain at least one frame of sparse image;
  • Step 405 Obtain the pixel value and the estimated value of the radial symmetry of the pixel in the thinned image
  • the radial symmetry refers to the radial symmetry of each pixel in a local area. Through the symmetry of the gradient field of the pixel, the radial symmetry of each pixel in the local area is obtained. As long as the estimated value of radial symmetry of the thinned image can be obtained, the specific algorithm of the estimated value of radial symmetry is not limited in the embodiment of the present application.
  • At least one image set to be preprocessed is preprocessed separately, at least one frame of thinned image is obtained, the pixel value of each pixel in each frame of the thinned image is obtained, and the radial degree estimation value of each pixel is obtained. Since the microbubbles are deformed along the direction of the motion track, the radial symmetry of the pixels is estimated to obtain the radial symmetry of each pixel in the local area, and then the radial symmetry estimation can retain the deformation information of the microbubbles, so Estimating the degree of the radial alignment of the pixel value of each pixel is equivalent to preserving the motion trajectory of the microbubbles.
  • Step 406 Perform a weighted calculation on the pixel values and the radial symmetry estimation values of the pixels in the same frame of the thinned image to obtain at least one frame of local super-resolution image respectively corresponding to the at least one frame of thinned image;
  • the microbubble point spread function is reduced, and the deformation information of the microbubble is retained, so that the trajectory skeleton of the microbubble movement axis in each frame of the thinning image becomes It is clearer, enhances the non-localized motion axis of the microbubbles, retains the trajectory information of the microbubbles along the motion direction, and obtains a local super-resolution image.
  • Step 407 Superimpose at least one frame of local super-resolution image to obtain a reconstructed super-resolution image.
  • the local super-resolution images of each frame are superimposed to form a complete ultrasound super-resolution reconstructed image of tiny blood vessels.
  • the impact of high concentration of microbubbles and strong noise on the accuracy of super-resolution imaging is effectively reduced;
  • the pixel value of the pixel and the estimated value of radial symmetry are weighted to realize the weight reduction of the microbubble point spread function, retain the deformation information of the microbubble, and make the trajectory skeleton of the microbubble movement axis in each frame of sparse image become clearer ,
  • the non-localized motion axis of the microbubbles is enhanced, the trajectory information of the microbubbles along the motion direction is retained, and the local super-resolution image is obtained, which is finally integrated into a complete ultrasonic super-resolution reconstruction image of the tiny blood vessels.
  • the non-positioning motion axis enhancement method retains the trajectory skeleton of the microbubble motion axis, thereby greatly improving the spatial resolution and reconstruction speed of the super-resolution reconstructed image, achieving rapid Efficient reconstruction of ultrasound super-resolution images.
  • FIG. 5 is a schematic diagram of the process of obtaining the estimated value of the radial symmetry of the pixels in the thinned image in a super-resolution reconstruction method of a contrast-enhanced ultrasound image according to an embodiment of the application.
  • obtaining the estimated value of the radial symmetry of the pixels in the thinned image includes:
  • Step 4051 Select a third pixel point located at the third pixel coordinate in the thinning image
  • the third pixel coordinate is only a reference, and can be any pixel coordinate point in the sparse image.
  • the embodiment of the present application does not limit the specific selection of the third pixel coordinate.
  • Step 4052 Select multiple surrounding pixel points located around the third pixel coordinate in the thinned image of the same frame;
  • Select multiple surrounding pixels around the third pixel in the same frame of thinned image for example, select a pixel at (1, 1) pixel coordinates and 12 surrounding a pixel in the thinned image of frame A1 Surrounding pixels.
  • the 12 surrounding pixels are located on a circle with a pixel as the center and a radius of r, and the 12 surrounding pixels are evenly distributed in this circle.
  • the number of surrounding pixels can be 6, 8, 10, 12, 15, 20, etc.
  • the embodiment of the present application does not limit the specific number of surrounding pixels.
  • Step 4053 Perform radial symmetry estimation on a plurality of pixel values surrounding the pixel to obtain the radial symmetry estimation value of the third pixel.
  • the radial symmetry is estimated on the pixel values of multiple surrounding pixels located around the third pixel coordinate in the thinned image in the same frame, and the deformation information of the microbubbles is retained, and the thinning of each frame is obtained by traversing.
  • the estimated value of the radial symmetry of each pixel on the image makes the non-localized motion of the microbubbles axially enhanced.
  • the number of surrounding pixels is 12, and the radial symmetry radius is 1.
  • FIG. 6 is a schematic flowchart of selecting at least one image set to be preprocessed from the contrast data in a super-resolution reconstruction method for ultrasound contrast images provided by an embodiment of the application. As shown in Fig. 6, selecting at least one image set to be preprocessed from the contrast data specifically includes:
  • Step 4011 Collect multiple frames of contrast images
  • the target to be imaged is injected with ultrasound contrast agent, the ultrasound probe is used to detect the area to be imaged, and when the contrast enhancement signal appears, the contrast data is collected to obtain multiple frames of contrast images.
  • the frequency of collecting multiple frames of contrast images may be 30 Hz or 50 Hz, and the frequency of collecting contrast images only needs to meet specific clinical acquisition conditions.
  • the frequency of collecting contrast images is not specifically limited in the embodiment of the present application.
  • the acquisition of multi-frame contrast images may be 500 frames, 1000 frames, or 1500 frames, etc.
  • the embodiment of the present application does not limit the specific number of frames of the multi-frame contrast images.
  • Step 4012 Perform registration on multiple frames of contrast images to suppress tissue movement interference, and obtain multiple frames of registered contrast images;
  • the registration of multiple frames of angiographic images is performed to suppress the image changes caused by the movement of the probe and the tissue, and the images after the multiple frames of registration are obtained.
  • the method of image registration can be flexible registration and rigid registration, including but not limited to median shift method, scale-invariant feature transformation method, tracking learning detection, optical flow method, and cross-correlation method.
  • the tracking learning detection is used for rigid registration
  • the optical flow method is used for flexible registration.
  • the specific method of registration is not limited in the embodiment of the present application.
  • Step 4013 In the multi-frame quasi-contrast image, select the second preset number of registered contrast images every first preset number of frames as the to-be-preprocessed registration contrast images, and obtain at least one to-be-preprocessed image set.
  • the second preset number of frames to register the contrast image every first preset number of frames, that is, select the same time window in the time dimension (the duration of the time window is the second preset number of frames) .
  • Perform preprocessing such as denoising reconstruction and interpolation calculation decoupling on the registered contrast images in the same time window.
  • W+1 frame G is less than W+1, W+1 is less than N
  • registration contrast image at every interval of G frames as the registration contrast image to be preprocessed
  • forming (NW-1 )/G+1 image sets to be preprocessed forming (NW-1 )/G+1 image sets to be preprocessed.
  • Each set of images to be preprocessed includes W+1 frames of registration contrast images to be preprocessed.
  • the image sets to be preprocessed are selected from the multi-frame quasi-contrast image according to the above method, so that each image set to be preprocessed is continuous, and the microbubbles of the preprocessed registration contrast image to be preprocessed have a stable spatial position.
  • N, G, and W are all positive integers.
  • the selection method of selecting the quasi-contrast image by the selection part from the multi-frame registration contrast image can be 4 frames every 1 frame; 3 frames every 2 frames; 5 frames every 1 frame.
  • the specific selection method of the selection part to allocate the quasi-contrast image from the multi-frame registration angiographic image is not limited.
  • the image set to be preprocessed is obtained by the above method, so that the denoising reconstruction and interpolation calculation are performed within a time window, so that each microbubble in the image set to be preprocessed has a stable spatial position. Reduce the influence of rapid flow of microbubbles on super-resolution reconstruction.
  • the first preset number is negatively related to blood flow velocity; and, the second preset number is negatively related to blood flow velocity. In one embodiment, the first preset number is positively correlated with the imaging frame rate; and, the second preset number is positively correlated with the imaging frame rate.
  • the first preset number G and the first preset number W+1 depend on the microbubble flow rate and the imaging frame rate. The faster the blood flow, the first preset It is assumed that the smaller the value of the number G and the first preset number W+1, the greater the imaging frame rate, and the larger the value of the first preset number G and the first preset number W+1.
  • the first preset number is 1; and, the second preset number of frames is 4.
  • step 7051, step 7052, and step 7052 in Figure 7 Perform radial symmetry estimation on the pixel values surrounding a pixel to obtain the estimated radial symmetry of the pixel (as shown in step 7051, step 7052, and step 7052 in Figure 7); in each frame of sparse image, traverse The pixel value of each pixel in the multi-sparse image and the estimated value of radial symmetry are weighted and calculated, so that the trajectory skeleton of the microbubble movement axis in each frame of the thinning image becomes clearer, and the non-positioning of the microbubble The movement axis is enhanced, so that the trajectory information of the microbubbles along the movement direction is retained, and a clear local super-resolution image is obtained (as shown in step 706 in Figure 7); a total of 1497 frames of local super-resolution images are obtained, and 1497 frames of local super-resolution images are obtained.
  • the images are superimposed to form a complete ultrasound super-resolution reconstructed image of
  • This application is based on the fluctuation characteristics of the gray-scale fluctuation signal of the microbubble in the time window, which not only effectively filters out the background or noise signal, and significantly highlights the microbubble signal, but also realizes the efficient decoupling of the spatially overlapping microbubbles, thereby enabling Significantly improve the super-resolution accuracy.
  • FIG. 8 is a schematic diagram of a pre-processing device for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • the super-resolution reconstruction pre-processing device 802 includes: a first pre-processing acquisition module 8021, configured to acquire a set of images to be pre-processed, and the set of images to be pre-processed includes multiple frames of registration contrast images to be pre-processed;
  • the degree acquisition module 8022 is configured to acquire the gray-scale fluctuation signals of the pixels in the registration contrast image to be preprocessed;
  • the denoising enhancement reconstruction module 8023 is configured to acquire the gray-scale fluctuation signals of the pixels in the registration contrast image to be preprocessed And based on the gray-scale fluctuation signal of the co-located pixel point set to denoise and reconstruct the to-be-processed image set to obtain a reconstructed characteristic parameter image, the co-located pixel point set includes: the to-be-preprocessed registration contrast image in
  • the denoising enhancement reconstruction module analyzes the gray-scale fluctuation signals of the co-located pixel point set to distinguish the microbubble signal from the noise or background signal in the image set to be preprocessed, thereby improving the signal-to-noise ratio With signal-to-back ratio, the reconstructed feature parameter image with enhanced microbubble signal and weakened background noise signal is obtained; the feature parameter image is reconstructed by the similarity of the gray-scale fluctuation signal of the co-located pixel point set and the associated pixel point set through the thinning module Interpolation is performed to separate different microbubbles, to achieve spatial decoupling of overlapping microbubbles, and effectively reduce the influence of strong noise and high-concentration microbubbles on reconstruction.
  • the gray-scale fluctuation signal is a signal formed by arranging the pixel values of the pixels in chronological order; the gray-scale fluctuation signal of the co-located pixel set is the pixel value of multiple co-located pixels in the co-located pixel set according to 1 times multi-dimensional signal formed by chronological arrangement.
  • FIG. 9 is a schematic diagram of a pre-processing device for super-resolution reconstruction of contrast-enhanced ultrasound images according to an embodiment of the application.
  • the denoising enhancement reconstruction module 8023 further includes: a first pixel point set acquisition unit 80231, configured to select a plurality of first pixel points located at the first pixel coordinates from the multiple frames of pre-processed registration contrast images.
  • the co-located pixels form a first co-located pixel point set;
  • the feature parameter extraction unit 80232 is configured to perform feature estimation and extraction on the gray level fluctuation signals of a plurality of first co-located pixels to obtain the first feature parameter, and the first feature
  • the parameter is used to characterize the randomness of the gray-scale fluctuation distribution and the periodicity of the gray-scale fluctuation of the first co-located pixel;
  • the preset unit 80233 is configured to preset the mapping between the first feature parameter and the reconstructed feature parameter image
  • the reconstruction unit 80234 is configured to map the first feature parameter to the first pixel coordinate of the feature parameter image to be reconstructed, and obtain the reconstructed feature parameter image according to the mapping relationship.
  • the method of feature estimation and extraction includes: autocorrelation estimation calculation or information entropy estimation calculation.
  • the thinning module 8024 includes: a second pixel point set acquiring unit 80241, configured to select a plurality of pixels located at the second pixel coordinates from the multi-frame to-be-preprocessed registration contrast image The second co-located pixel points form a second co-located pixel point set;
  • the associated pixel point acquisition unit 80242 is configured to select multiple frames located at the associated pixel coordinates adjacent to the second pixel coordinates from the multi-frame to-be-preprocessed registration radiographic image Associated pixel points to form an associated pixel point set;
  • the similarity quantification unit 80243 is configured to perform similarity quantification on the gray-scale fluctuation signal of the second co-located pixel point set and the gray-scale fluctuation signal of the associated pixel point set to obtain the similarity An estimated value;
  • an interpolation unit 80244 configured to insert a similarity estimated value into the corresponding interpolated pixel coordinate in the reconstructed feature parameter image, the interpolated pixel coordinate being located between
  • the method of similarity quantification includes: cross-entropy estimation or cross-correlation estimation.
  • the number of associated pixel point sets is 4.
  • FIG. 10 is a schematic diagram of a super-resolution reconstruction device for contrast-enhanced ultrasound images according to an embodiment of the application.
  • the super-resolution reconstruction device 800 includes: a first selection module 801, configured to select at least one image set to be preprocessed from the contrast data; The reconstruction preprocessing method performs preprocessing on at least one image set to be preprocessed to obtain at least one frame of sparsed image; axial trajectory highlighting module 803; configured to obtain the pixel value and the radial symmetry estimated value of the pixel in the sparsed image Calculate the pixel values and the estimated values of the radial symmetry of the pixels in the same frame of the thinned image to obtain at least one frame of local super-resolution images corresponding to at least one frame of the thinned image respectively; and the superimposing module 804 , Configured to superimpose at least one frame of local super-resolution image to obtain a reconstructed super-resolution image.
  • At least one preprocessed image set is preprocessed by the preprocessing device, which effectively reduces the impact of high concentration of microbubbles and strong noise on the accuracy of super-resolution imaging;
  • the pixel value and the radial symmetry and the estimated value of the radial symmetry of the pixel in the chemical image are weighted and calculated to realize the weight reduction of the microbubble point spread function, retain the deformation information of the microbubble, and retain the movement trajectory of the microbubble, so that the microbubble
  • the non-localized motion axis is enhanced, and the trajectory skeleton of the microbubble motion axis is retained, and local super-resolution images are obtained; through the superposition module, the integrated and complete ultrasound super-resolution reconstruction images of the micro blood vessels are finally obtained.
  • this reconstruction equipment retains the trajectory skeleton of the microbubble motion axis through the non-positioning motion axis enhancement method, thereby greatly improving the spatial resolution of the super-resolution reconstructed image Rate and reconstruction speed, to achieve fast and efficient ultrasound super-resolution image reconstruction.
  • FIG. 11 is a schematic diagram of a super-resolution reconstruction device for contrast-enhanced ultrasound images according to an embodiment of the application.
  • the axial trajectory highlighting module 803 includes: a third pixel point acquisition unit 8031, which selects a third pixel point located at the third pixel coordinate in the thinning image; and the surrounding pixel point acquisition unit 8032 is configured to be in the same Select a plurality of surrounding pixel points located around the third pixel coordinate in the frame thinning image; and a radial symmetry estimation unit 8033, configured to perform radial symmetry estimation on the pixel values of the plurality of surrounding pixels to obtain the third pixel The estimated value of the radial symmetry of the point.
  • the first selection module 801 further includes: an acquisition unit 8011, configured to acquire multiple frames of contrast images; a registration unit 8012, configured to register multiple frames of contrast images to suppress tissue Motion interference, obtaining multi-frame registration contrast images; and a filtering unit 8013, configured to select a second preset number of registration contrast images every first preset number of frames among the multi-frame quasi contrast images as the registration to be preprocessed Contrast images to obtain at least one image set to be preprocessed.
  • the first preset number is negatively related to blood flow velocity; and, the second preset number is negatively related to blood flow velocity.
  • the first preset number is positively correlated with the imaging frame rate; and, the second preset number is positively correlated with the imaging frame rate.
  • the first preset number is 1; and, the second preset number of frames is 4.
  • the apparatus 800 for super-resolution reconstruction of contrast-enhanced ultrasound images can be integrated into the electronic device 1200 as a software module and/or hardware module.
  • the electronic device 1200 can include the contrast-enhanced ultrasound image.
  • Super-resolution reconstruction device 800 the super-resolution reconstruction device 800 of the contrast-enhanced ultrasound image may be a software module in the operating system of the electronic device 1200, or may be an application developed for it; of course, the super-resolution reconstruction of the contrast-enhanced ultrasound image
  • the apparatus 800 may also be one of many hardware modules of the electronic device 1200.
  • the apparatus 800 for super-resolution reconstruction of ultrasound images and the electronic device 1200 may also be separate devices (for example, a server), and the apparatus 800 for super-resolution reconstruction of ultrasound images may be wired And/or the wireless network is connected to the electronic device 1200, and the interactive information is transmitted according to the agreed data format.
  • FIG. 12 is a schematic structural diagram of an electronic device provided by an embodiment of the application.
  • the electronic device 1200 includes: one or more processors 1201 and a memory 1202; and computer program instructions stored in the memory 1202.
  • the processor 1201 executes such as The super-resolution reconstruction pre-processing method of any of the above embodiments or the super-resolution reconstruction method of any of the above embodiments.
  • the processor 1201 may be a central processing unit (CPU) or other form of processing unit with data processing capability and/or instruction execution capability, and may control other components in the electronic device to perform desired functions.
  • CPU central processing unit
  • the memory 1202 may include one or more computer program products, and the computer program products may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory.
  • the volatile memory may include random access memory (RAM) and/or cache memory (cache), for example.
  • the non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, and the like.
  • One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 1701 may run the program instructions to implement the above-mentioned super-resolution reconstruction preprocessing of the present application as in any of the above-mentioned embodiments. The method or the steps of the super-resolution reconstruction method of any of the above embodiments and/or other desired functions.
  • the computer-readable storage medium may also store information such as light intensity, compensation light intensity, and the position of the filter.
  • the electronic device 1200 may further include: an input device 1203 and an output device 1204, and these components are interconnected by a bus system and/or other forms of connection mechanisms (not shown in FIG. 12).
  • the input device 1203 may also include, for example, a keyboard, a mouse, a microphone, and so on.
  • the output device 1204 can output various information to the outside, for example, it can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected to it, and so on.
  • the electronic device 1200 may also include any other appropriate components according to specific application conditions.
  • the embodiments of the present application may also be computer program products, including computer program instructions, which, when run by a processor, cause the processor to execute the super-resolution reconstruction preprocessing method as in any of the above-mentioned embodiments. Or a step in the super-resolution reconstruction method of any of the above embodiments.
  • the computer program product can be used to write program codes for performing the operations of the embodiments of the present application in any combination of one or more programming languages, the programming languages including object-oriented programming languages, such as Java, C++, etc., Including conventional procedural programming languages, such as "C" language or similar programming languages.
  • the program code can be executed entirely on the user's computing device, partly on the user's device, executed as an independent software package, partly on the user's computing device and partly executed on the remote computing device, or entirely on the remote computing device or server Executed on.
  • the embodiment of the present application may also be a computer-readable storage medium, on which computer program instructions are stored.
  • the processor executes the “exemplary super-resolution reconstruction described above” in this specification.
  • the steps of the preprocessing method according to any of the foregoing embodiments of the present application or the reconstruction method of any of the foregoing embodiments are described in the section "Preprocessing Method” or "Exemplary Super Resolution Reconstruction Method”.
  • the computer-readable storage medium may adopt any combination of one or more readable media.
  • the readable medium may be a readable signal medium or a readable storage medium.
  • the readable storage medium may include, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the above, for example. More specific examples (non-exhaustive list) of readable storage media include: electrical connections with one or more wires, portable disks, hard disks, random access memory ((RAM), read only memory (ROM), Erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • each component or each step can be decomposed and/or recombined.
  • decompositions and/or recombinations shall be regarded as equivalent solutions of this application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Surgery (AREA)
  • Medical Informatics (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Pathology (AREA)
  • Radiology & Medical Imaging (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Public Health (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Geometry (AREA)
  • Hematology (AREA)
  • Multimedia (AREA)
  • Physiology (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)
  • Image Processing (AREA)

Abstract

一种超声造影图像的超分辨重建预处理方法，包括：获取待预处理图像集；获取待预处理配准造影图像中像素点的灰度涨落信号；基于同位像素点集的灰度涨落信号对待预处理图像集进行去噪重构得到重构特征参数图像；以及基于同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号对重构特征参数图像进行插值计算获得稀疏化图像。通过对多帧待预处理配准造影图像中的同位像素点集的灰度涨落信号进行分析，提高信噪比与信背比；通过同位像素点集以及关联像素点集的灰度涨落信号的相似度对重构特征参数图像进行插值，使重叠的微泡实现空间解耦，有效降低强噪声以及高浓度微泡对重建的影响。

Description

超声造影图像的超分辨重建预处理方法和超分辨重建方法 技术领域
本申请涉及超声成像技术领域，具体涉及一种超声造影图像的超分辨重建预处理方法、超分辨重建方法、超分辨重建预处理装置、超分辨重建装置、电子设备和计算机可读存储介质。
发明背景
超微血流成像技术-超声定位显微镜(Ultrasound Localization Microscope,ULM)，克服了声学衍射带来的影响，实现了对微小血管的超分辨成像。目前临床经常采用可以缩短采集时间的高浓度微泡，高浓度的微泡之间存在空间耦合，影响超分辨重建的准确度。在临床采集过程中，强噪声以及组织背景信号的干扰无法避免，超声微泡定位的准确性被影响。受限于现有的临床采集条件，无法实现快速高效且准确地重建微小血管超分辨图像，从而为超分辨的临床应用带来极大的挑战。
发明内容
有鉴于此，本申请实施例提供了一种超声造影图像的超分辨重建预处理方法、超分辨重建方法、超分辨重建预处理装置、超分辨重建装置、电子设备和计算机可读存储介质，以解决现有技术中由于临床采集条件的限制无法实现快速高效且准确地重建微小血管的超分辨图像的问题。
根据本申请的一个方面，本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法，包括：获取待预处理图像集，所述待预处理图像集包含多帧待预处理配准造影图像；获取所述待预处理配准造影图像中像素点的灰度涨落信号；基于同位像素点集的灰度涨落信号对所述待预处理图像集进行去噪重构得到重构特征参数图像，所述同位像素点集包括：在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个同位像素点；以及基于所述同位像素点集以及与所述同位像素点集相关联的关联像素点集的灰度涨落信号对所述重构特征参数图像进行插值计算获得稀疏化图像，所述关联像素点集包括：在同帧所述待预处理配准造影图像中与所述同位像素点相邻，且在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个关联像素点。
根据本申请另一个方面，本申请一实施例提供的一种超声造影图像的超分辨重建方法，包括：从造影数据中选取至少一个待预处理图像集；对所述至少一个待预处理图像集分别进行预处理，获取至少一帧稀疏化图像，所述预处理方法采用上述任一项所述的预处理方法；获取所述稀疏化图像中像素点的像素值和径向对称度估计值；将位于同帧所述稀疏化图像中的像素点的像素值和径向对称度估计值进行加权计算，获得与所述至少一帧稀疏化图像分别对应的至少一帧局部超分辨图像；以及将所述至少一帧局部超分辨图像进行叠加获得重建的超分辨图像。
根据本申请的又一个方面，本申请一实施例提供的本申请一实施例提供了一种超声造影图像的超分辨重建预处理装置，包括：第一预处理获取模块，配置为获取待预处理图像集，所述待预处理图像集包含多帧待预处理配准造影图像；灰度获取模块，配置为获取所述待预处理配准造影图像中像素点的灰度涨落信号；去噪增强重构模块，配置为基于同位像素点集的灰度涨落信号对所述待预处理图像集进行去噪重构得到重构特征参数图像，所述同位像素点集包括：在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个同位像素点；以及稀疏化模块，配置为基于所述同位像素点集以及与所述同位像素点集相关联的关联像素点集的灰度 涨落信号对所述重构特征参数图像进行插值计算获得稀疏化图像，所述关联像素点集包括：在同帧所述待预处理配准造影图像中与所述同位像素点相邻，且在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个关联像素点。
根据本申请的又一个方面，本申请一实施例提供的本申请一实施例提供了一种超声造影图像的超分辨重建装置，该重建装置包括：第一选取模块，配置为从造影数据中选取至少一个待预处理图像集；预处理装置，配置为采用如上述任一项所述的超分辨重建预处理方法，对所述至少一个待预处理图像集分别进行预处理获取至少一帧稀疏化图像；轴向轨迹凸显模块，配置为获取所述稀疏化图像中像素点的像素值和径向对称度估计值；并将位于同帧所述稀疏化图像中的像素点的像素值和径向对称度估计值进行加权计算，获得与所述至少一帧稀疏化图像分别对应的至少一帧局部超分辨图像；以及叠加模块，将所述至少一帧局部超分辨图像进行叠加获得重建的超分辨图像。
根据本申请的又一个方面，本申请一实施例提供的本申请一实施例提供了一种电子设备，包括：处理器；存储器；以及存储在存储器中的计算机程序指令，计算机程序指令在被处理器运行时使得处理器执行如上述任一项所述的超分辨重建预处理方法或上述任一项所述的超分辨重建方法。
根据本申请的另一方面，本申请一实施例提供了一种计算机可读存储介质，所述计算机可读存储介质上存储有计算机程序指令，所述计算机程序指令在被处理器运行时使得所述处理器执行如上述任一项所述的超分辨重建预处理方法或上述任一项所述的超分辨重建方法。
本申请实施例提供的一种超声造影图像的超分辨重建预处理方法、超分辨重建方法、超分辨重建预处理装置、超分辨重建装置、电子设备和计算机可读存储介质，通过对多帧待预处理配准造影图像中的同位像素点集的灰度涨落信号进行分析，使当前时间窗内的待预处理图像集的微泡信号和噪声或背景信号区分开，提高信噪比与信背比，得到微泡信号增强与背景噪声信号削弱的重构特征参数图像；通过同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号的相似度对重构特征参数图像进行插值起到分隔不同微泡的作用，使重叠的微泡实现空间解耦，解决衍射极限造成的微泡重叠。通过对预处理图像集进行预处理，有效降低高浓度微泡以及强噪声对超分辨成像的准确性的影响。通过对每帧稀疏化图像中像素点的像素值和径向对称度估计值进行加权计算，实现微泡点扩散函数瘦身，保留微泡的变形信息，使每帧稀疏化图像中微泡运动轴向的轨迹骨架变得更加清晰，使微泡的非定位的运动轴向增强，保留了微泡沿运动方向的轨迹信息，获得局部超分辨图像，最终整合为完整的微小血管的超声超分辨重建图像。有别于传统的单点定位和累加策略，非定位的运动轴向增强方式保留了微泡运动轴向的轨迹骨架，从而大幅度提高了超分辨重建图像的空间分辨率及重建速度，实现快速高效的超分辨图像的重建。
附图简要说明
图1所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法的流程示意图。
图2所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法中获取重构特征参数图像的流程示意图。
图3所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法中插值计算获得稀疏化图像的流程示意图。
图4所示为本申请一实施例提供的一种超声造影图像的超分辨重建方法。
图5所示为本申请一实施例提供的一种超声造影图像的超分辨重建方法中获取稀疏化图像中像素点的径向对称度估计值的流程示意图。
图6所示为本申请一实施例提供的一种超声造影图像的超分辨重建方法中从造影数据中选取至少一个待预处理图像集的流程示意图。
图7所示为本申请一实施例提供的新西兰大白兔下肢二头肌的超声造影图像的超分辨重 建方法的流程图。
图8所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理建装置的示意图。
图9所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理建装置的示意图。
图10所示为本申请一实施例提供的一种超声造影图像的超分辨重建装置的示意图。
图11所示为本申请一实施例提供的一种超声造影图像的超分辨重建装置的示意图。
图12所示为本申请一实施例提供的电子设备的结构示意图。
实施本申请的方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
申请概述
微血管成像对于疾病的诊断大有裨益。上世纪七十年代Gramiak发明的超声微泡造影剂使得超声血管成像成为可能。微泡可以在血管内流动，在超声波下以非线性方式振动，产生特定的谐波特征，使它们能够在较深的组织中以较高的灵敏度将血管与背景组织信号区分开来。目前，造影增强超声(CEUS)已经在许多临床诊断中作为常规血流成像手段，但其仍然受到衍射极限的限制，无法对微小血管(血管直径<100μm)进行高空间分辨率成像。
Errico等人提出一种基于超声造影的超微血流成像技术-超声定位显微镜，通过对单个微泡进行定位及时间累加，克服了声学衍射带来的影响，实现了对直径几十微米微小血管的超分辨成像。
但基于微泡定位的超声超分辨血流成像，强噪声极易诱导产生误定位点，较深血流中采集的微泡信号易受组织信号的干扰，不易区分微泡信号与背景或噪声信号，影响超声微泡定位的准确性，而目前临床无法满足弱噪声的采集条件。其次，为了实现微泡的精准定位，超声定位显微镜需要血流中保持较低的微泡浓度，，但目前为了满足临床标准以及进行血管功能的诊断，在临床常见还是高浓度微泡，但微泡浓度高，微泡之间存在空间耦合，无法精准定位微泡，影响超分辨成像的准确性。
目前亟需一种超声造影图像的超分辨重建方法，在临床常见的高浓度微泡分布以及强噪声的条件下，可以快速重建微小血管的超分辨图像，使超声超分辨血流成像在临床上可以广泛应用。
针对上述的技术问题，本申请的基本构思是提出一种超声造影图像的超分辨重建的预处理方法，通过对时间尺度内多帧待预处理配准造影图像中的同位像素点集的灰度涨落信号进行分析，将微泡信号与背景或噪声信号区分开来，有效滤除了背景及噪声信号，显著凸显了微泡信号；通过同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号的相似度，对重构特征参数图像进行插值，实现了对空间重叠微泡的高效解耦，有效降低高浓度微泡以及强噪声对超分辨成像的准确性的影响。通过选取相同时间窗内待预处理图像集进行预处理，通过像素点的径向对称度使非定位的运动轴向增强，保留了微泡运动轴向的轨迹骨架，从而大幅度提高了超分辨重建图像的空间分辨率及重建速度，实现快速高效的超分辨图像的重建。
在介绍了本申请的基本原理之后，下面将参考附图来具体介绍本申请的各种非限制性实施例。
示例性预处理方法
图1所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法。如图1所示，该预处理方法包括：
步骤101：获取待预处理图像集，待预处理图像集包含多帧待预处理配准造影图像；
对待成像目标进行超声造影剂注射，用超声探头检测待成像区域，当出现造影增强信号时开始采集造影数据，获得多帧造影图像；对多帧造影图像进行配准，以抑制探头和组织运动造成的图像变化，获得多帧配准后的图像即配准造影图像；从多帧配准造影图像中选取部分配准造影图像作为待预处理配准造影图像；因此，每帧待预处理配准造影图像均是配准后的造影图像。
采集多帧造影图像的频率可以是30Hz也可以是50Hz，采集造影图像的频率只要满足具体临床采集条件即可，在本申请实施例对采集造影图像的频率不做具体限定。从多帧配准造影图像中选取部分配准造影图像的选取方法可以是每隔1帧选取4帧；也可以是每隔2帧选取3帧；还可以是每隔1帧选取5帧，本申请实施例从多帧配准造影图像中选取部分配准造影图像的具体选取方法不做限定。
对多帧造影图像进行配准的配准方法可以是使用Morphon多尺度配准方法对多帧造影图像进行配准，本申请实施例对配准的具体方法不做限定。造影剂注射方法可以是用六氟化硫微泡冻干粉声诺维59mg溶于5mL0.9％的氯化钠溶液中配置一份超声造影剂，一次性注射0.2mL进入装有3L氯化钠溶液的模型中，或以0.2-5.0μL/min的速率持续注射，本申请实施例对造影剂的具体种类以及具体注射方法不做限定。
步骤102：获取待预处理配准造影图像中像素点的灰度涨落信号；
基于采集图像的超声设备，每帧造影图像上具有被划分的像素坐标，像素坐标上具有像素点。由于多帧造影是由同一设备采集以及每帧配准造影图像是利用相同配准参数进行配准，那么每帧配准造影图像上的像素坐标划分是相同的，则每帧待预处理配准造影图像上的像素坐标划分也是相同的。获取待预处理图像集中每帧待预处理配准造影图像中的每个像素点的灰度涨落信号，该灰度涨落信号用于表示位于当前帧当前像素点的灰度涨落的变化情况。
只要可以获取每个像素点的灰度涨落信号即可，本申请实施例对灰度涨落信号的具体获取手段不做具体限定。
步骤103：基于同位像素点集的灰度涨落信号对待预处理图像集进行去噪重构得到重构特征参数图像，同位像素点集包括：在不同帧待预处理配准造影图像中位于相同像素坐标处的多个同位像素点；
由于每帧待预处理配准造影图像上的像素坐标划分是相同的，则一个待预处理图像集中每帧待预处理配准造影图像上都具有相同的像素坐标。在不同帧待预处理配准造影图中位于的相同像素坐标处的像素点就是同位像素点。在每帧待预处理配准造影图像的像素点的灰度涨落信号是一个一维信号，即每一个同位像素点的灰度涨落信号是一个一维灰度涨落信号；在不同帧待预处理配准造影图像中位于相同像素坐标处的同位像素点集的灰度涨落信号就是一个1乘多维灰度涨落信号，即一个同位像素点集是一个1乘多维灰度涨落信号，该1乘多维灰度涨落信号用于反映位于当前时间窗的待预处理图像集的灰度涨落的周期性以及灰度涨落的随机性。由于造影剂微泡和背景或噪声对应的灰度涨落信号的周期性以及分布情况均不同，微泡灰度涨落信号具有更强的周期性且更强的随机性，基于多个同位像素点的1乘多维灰度涨落信号，可以区分微泡信号和噪声或背景信号，起到增强微泡信号且削弱背景或噪声信号的作用，最终获得一个微泡信号被增强和背景或噪声信号被削弱的重构特征参数图像。
例如：位于5帧时间窗内的一个待预处理图像集包含A、B、C、D和E中五帧待预处理配准造影图像，A帧待预处理配准造影图像的(1、1)像素坐标处上是a像素点，B帧待预处理配准造影图像的(1、1)像素坐标处上是b像素点，C帧待预处理配准造影图像的(1、1)像素坐标处上是c像素点，D帧待预处理配准造影图像的(1、1)像素坐标处上是d像素点，E帧待预处理配准造影图像的(1、1)像素坐标处上是e像素点，那么a、b、c、d和e就是同位像素点集。基于a、b、c、d和e的构成的1乘5维灰度涨落信号可以使当前时间窗内的待预处理图像集的微泡信号和噪声或背景信号区分开，提高信噪比与信背比。
待预处理图像集中同位像素点集的个数与相同像素坐标的个数相同，待预处理图像集中相同像素坐标的个数由采集造影图像的设备确定。
步骤104：基于同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号对重构特征参数图像进行插值计算获得稀疏化图像，关联像素点集包括：在同帧待预处理配准造影图像中与同位像素点相邻，且在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个关联像素点。
当微泡尺寸小于波长时会发生衍射现象，这会导致成像的分辨率无法突破半波长，这种现象被称为衍射极限。由于衍射极限的存在，在每帧待预处理配准造影图像上，微泡会出现重叠。由于属于同一个微泡上的像素点具有近乎相同的灰度涨落信号，彼此之间相似度高，而属于不同微泡的像素点具有完全不同的灰度涨落信号，彼此之间相似度低，因此基于同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号对重构特征参数图像进行插值可以起到分隔不同微泡的作用，可以使重叠的微泡实现空间解耦。
关联像素点集中每一个关联像素点在不同帧待预处理配准造影图像中也是位于相同像素坐标处，也就是说关联像素点集的本身也是一组同位像素点，只是相对于被选中的当前同位像素集，关联像素点集的每一个像素点与当前同位像素集的每一个同位像素点在同帧待预处理配准造影图像中相邻且相对位置相同。同位像素点的灰度涨落信号是一个一维灰度涨落信号，被选中的当前同位像素集是一个1乘多维连续灰度涨落信号；每一个关联像素点的灰度涨落信号也是一个一维灰度涨落信号，关联像素点集也是一个1乘多维连续灰度涨落信号，关联像素点集是位于当前同位像素集附近的一个1乘多维连续灰度涨落信号，利用两个1乘多维连续灰度涨落信号的相似性对重构特征参数图像进行插值可以起到分隔不同微泡的作用。
例如：A帧待预处理配准造影图像的(1、1)像素坐标处上是a像素点，A帧(0、1)像素坐标处上是a1像素点，A帧(1、2)像素坐标处上是a2像素点；B帧待预处理配准造影图像的(1、1)像素坐标处上是b像素点，B帧(0、1)像素坐标处上是b1像素点，B帧(1、2)像素坐标处上是b2像素点；C帧待预处理配准造影图像的(1、1)像素坐标处上是c像素点，C帧(0、1)像素坐标处上是c1像素点，C帧(1、2)像素坐标处上是c2像素点；D帧待预处理配准造影图像的(1、1)像素坐标处上是d像素点，D帧(0、1)像素坐标处上是d1像素点，D帧(1、2)像素坐标处上是d2像素点；E帧待预处理配准造影图像的(1、1)像素坐标处上是e像素点，E帧(0、1)像素坐标处上是e1像素点，E帧(1、2)像素坐标处上是e2像素点；那么a、b、c、d和e是被选中的当前同位像素点集；a1、b1、c1、d1和e1就是一个关联像素点集；a2、b2、c2、d2和e2也是一个关联像素点集。
每一个同位像素点集对应的关联像素点集的个数可以是4、6、8、10以及12，本申请实施例对每一个同位像素点集对应的关联像素点集的个数不做具体限定。
在本申请实施例中，通过对多帧待预处理配准造影图像中的同位像素点集的灰度涨落信号进行分析，将使当前时间窗内的待预处理图像集的微泡信号和噪声或背景信号区分开，提高信噪比与信背比，得到微泡信号增强与背景噪声信号削弱的重构特征参数图像；通过同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号的相似度对重构特征参数图像进行插值可以起到分隔不同微泡的作用，可以使重叠的微泡实现空间解耦，解决衍射极限造成的微泡重叠。
在一个实施例中，灰度涨落信号为像素点的像素值按照时间顺序排列形成的信号；同位像素点集的灰度涨落信号为同位像素点集中的多个同位像素点的像素值按照时间顺序排列形成的1乘多维信号。由于微泡和背景或噪声的分布情况不同，微泡和背景或噪声的周期性也不同，微泡信号的周期性更强，微泡信号的分布更随机，而灰度涨落信号的特征是灰度涨落分布的随机性与灰度涨落的周期性，那么利用同位像素点的灰度涨落分布特性与灰度涨落周期特性可以区分微泡信号与背景或噪声信号。利用同位像素点集形成的1乘多维信号与周围像素点集形成的1乘多维信号的相似度进行插值可以使重叠的微泡实现空间解耦。
图2所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法中获取重构特征参数图像的流程示意图。如图2所示，基于同位像素点集的灰度涨落信号对待预处理图像集进行去噪重构得到重构特征参数图像可具体包括：
步骤1031：在多帧待预处理配准造影图像中选取位于第一像素坐标处的多个第一同位像 素点，形成第一同位像素点集；
选取每帧待预处理配准造影图像的第一像素坐标处的第一同位同位像素点，形成第一同位像素点集。例如：在A、B、C、D和E中五帧待预处理配准造影图像中的每帧上的(1、1)的像素坐标处的a、b、c、d和e，形成第一同位像素点集。
第一像素坐标只是一个代指，第一像素坐标可以是待预处理图像集中任何一个像素坐标点，本申请实施例对第一像素坐标的具体选取不做限定。
步骤1032：对多个第一同位像素点的灰度涨落信号进行特征估计提取得到第一特征参数，第一特征参数的用于表征第一同位像素点的灰度涨落分布的随机性与灰度涨落的周期性；
多个第一同位像素点进行第一特征参数提取，用第一特征参数表征多个同位像素点的灰度涨落分布的随机性，属于微泡的像素点的灰度涨落信号随机性更高，属于噪声或背景的像素点的灰度涨落信号的随机性更低；用第一特征参数表征多个同位像素点的灰度涨落的周期性，属于微泡的像素点的灰度涨落周期的周期性更强，属于系统与背景的像素点的灰度涨落周期性更低，也就是系统与背景的像素点的灰度涨落周期比较弱。遍历待预处理图像中每个像素坐标处的每个同位像素点集，提取出每个同位像素点集的对应的第一特征参数，利用多个第一特征参数反映出整个待预处理图像集的灰度涨落分布特性与灰度涨落周期特性。
步骤1033：预设第一特征参数与重构特征参数图像的映射关系；
由于多个第一特征参数可以反映出整个待预处理图像集的灰度涨落分布特性与灰度涨落周期特性，并且造影剂微泡和背景或噪声对应的灰度涨落信号的周期性以及分布情况均不同，利用灰度涨落分布特性与灰度涨落周期特性又可以区分微泡信号与背景或噪声信号，那么根据第一特征参数就能实现区分微泡信号与背景或噪声信号。预设第一特征参数与重构特征参数图像的映射关系，根据提取的第一特征参数与重构特征参数图像的映射关系，就能重构出微泡信号被加强而背景或噪声被削弱的重构特征参数图像。
例如：提前预设第一特征参数的值大，表明对应的同位像素点集的灰度涨落周期性更强，每个同位像素点属于微泡的概率更大，那么对应的待重构特征参数图像中的像素点的像素灰度值要高，提前预设第一特征参数的值小，表明对应的同位像素点集的灰度涨落周期性弱，每个同位像素点属于背景或者噪声的概率更大，那么对应的重构特征参数图像中的像素点的像素灰度值要低。只要根据第一特征参数与重构特征参数图像的提前预设映射关系，就能重构出微泡信号被加强而背景或噪声被削弱的重构特征参数图像即可，本申请实施例对预设映射关系的具体实现方式不做限定。
步骤1034：将第一特征参数对应到待重构特征参数图像的第一像素坐标处，根据映射关系，得到重构特征参数图像。
重构特征参数图像：是根据待预处理图像集的第一特征参数反映的灰度涨落分布特性与灰度涨落周期特性，重新构建得到的微泡信号被加强而背景或噪声被削弱(信噪比与信背比被改善)的图像。将第一特征参数对应到待重构特征参数图像的第一像素坐标处，根据映射关系，微泡信号需要被加强以及背景信号需要被削弱通过坐标的对准被准确地传达，实现大部分背景及噪声干扰的去除。
本申请实施例中，通过提取反映出整个待预处理图像集的灰度涨落分布特性与灰度涨落周期特性的多个第一特征参数，使得微泡信号与背景或噪声信号被区分；利用第一特征参数与重构特征参数图像的提前预设的映射关系，使得微泡区域信号需要被加强以及背景区域信号需要被削弱通过坐标被准确地传达到待重构特征参数图像的对应的位置处，重新构建得到信噪比与信背比被改善的重构特征参数图像。
在一个实施例中，特征估计提取的方法包括：自相关估计计算或信息熵估计计算。通过自相关估计来度量每个同位像素点的灰度涨落信号的周期性强弱；通过信息熵估计来度量每个同位像素点灰度涨落信号的分布的随机性。利用自相关估计计算或信息熵估计计算得到第一特征参数，利用多个第一特征参数反映出整个待预处理图像集的灰度涨落分布特性与灰度涨落周期特性。将使当前时间窗内的待预处理图像集的微泡信号和噪声或背景信号区分开，提高信噪比与信背比，得到微泡信号增强与背景噪声信号削弱的重构特征参数图像。
图3所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理方法中插值计算获得稀疏化图像的流程示意图。如图3所示，基于同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号对重构特征参数图像进行插值计算获得稀疏化图像包括：
步骤1041：在多帧待预处理配准造影图像中选取位于第二像素坐标处的多个第二同位像素点，形成第二同位像素点集；
选取每帧待预处理配准造影图像的第二像素坐标处的第二同位像素点，形成第二同位像素点集。例如：在A、B、C、D和E中五帧待预处理配准造影图像中的每帧上的(1、1)的像素坐标处的a、b、c、d和e，形成第二同位像素点集。
第二像素坐标只是一个代指，可以是待预处理图像集中任何一个像素坐标点，第二像素坐标可以与第一像素坐标是同一个坐标，可以与第一像素坐标不是同一个坐标，本申请实施例对第二像素坐标的具体选取不做限定。
步骤1042：在多帧待预处理配准造影图像中选取位于与第二像素坐标相邻的关联像素坐标处的多个关联像素点，形成关联像素点集；
在每帧待预处理配准造影图像中，选取位于第二像素坐标相邻的关联像素坐标处多个关联像素点。第二像素坐标对应的关联像素坐标不止一个，每一个同位像素点集对应的关联像素点集的个数也不止一个，本申请实施例对每一个同位像素点集对应的关联像素点集的个数不做具体限定。下面以第二像素坐标相邻的关联像素坐标有四个，即一个同位像素点集对应的四个关联像素点集为例。例如，A帧待预处理配准造影图像的(1、1)像素坐标处的多个关联像素坐标为(0、1)、(1、2)、(2、1)和(1、0)，第二像素点a的关联像素点分别是位于(0、1)、(1、2)、(2、1)和(1、0)的a1、a2、a3和a4；同理，B帧中第二像素点b的关联像素点是b1、b2、b3和b4；C帧中第二像素点c的关联像素点是c1、c2、c3和c4；D帧中第二像素点d的关联像素点是d1、d2、d3和d4；E帧中第二像素点e的关联像素点是e1、e2、e3和e4。那么a、b、c、d和e是同位像素点集；a1、b1、c1、d1和e1是一个关联像素点集，a2、b2、c2、d2和e2就是一个关联像素点集；a3、b3、c3、d3和e3就是一个关联像素点集；a4、b4、c4、d4和e4就是一个关联像素点集。同位像素点集(a、b、c、d和e)是一个1乘5维灰度涨落信号；关联像素点集(a1、b1、c1、d1和e1)是一个1乘5维灰度涨落信号；同理，(a2、b2、c2、d2和e2)，(a3、b3、c3、d3和e3)和(a4、b4、c4、d4和e4)均是一个1乘5维灰度涨落信号。
步骤1043：对第二同位像素点集的灰度涨落信号与关联像素点集的灰度涨落信号进行相似度量化得到相似度估量值；
利用每帧待预处理配准造影图像中第二同位像素点和与第二同位像素点相关联的关联像素点(第二同位像素点集与关联像素点的)的灰度涨落信号进行相似度计算，即两个1乘多维连续灰度涨落信号的相似性得到量化，将两个信号的相似度变为更直观的相似度估量值。
步骤1044：将相似度估量值插入重构特征参数图像中对应的插值像素坐标处，插值像素坐标位于第二像素坐标与关联像素坐标之间；
将第二同位像素点集与关联像素点的相似度估量值插入重构特征参数图像中对应的插值像素坐标处，使得直观的相似度估量值呈现在重构特征参数图像中的各个像素之间。
例如：计算同位像素点集(a、b、c、d和e)与关联像素点集(a1、b1、c1、d1和e1)的相似度得到相似度估量值Z1，将Z1插入重构特征参数图像的像素坐标(1、1)与(0、1)之间。以此类推，将同位像素点集(a、b、c、d和e)与关联像素点集(a2、b2、c2、d2和e2)的相似度得到相似度估量值Z2，将Z2插入重构特征参数图像的像素坐标(1、1)与(1、2)之间；将同位像素点集(a、b、c、d和e)与关联像素点集(a3、b3、c3、d3和e3)的相似度得到相似度估量值Z3，将Z3插入重构特征参数图像的像素坐标(1、1)与(2、1)之间；将同位像素点集(a、b、c、d和e)与关联像素点集(a4、b4、c4、d4和e4)的相似度得到相似度估量值Z4，将Z4插入重构特征参数图像的像素坐标(1、1)与(1、0)之间。
步骤1045：利用插值像素坐标的位置，将重构特征参数图像中重叠的微泡进行空间解耦，得到稀疏化图像。
将相似度估量值插入重构特征参数图像中对应的插值像素坐标处，使得直观的相似度估量值呈现在重构特征参数图像中的各个像素之间，将相似度高的像素点归属于同一微泡，将相似度不高的像素点分隔为不同的微泡，利用插值起到分隔不同微泡的作用，可以使重叠的微泡实现空间解耦。
本申请实施例中，通过同位像素点集与关联像素点集的灰度涨落信号进行相似度计算，将两个信号的相似度变为更直观的相似度估量值，将相似度估量值插入重构特征参数图像中对应的插值像素坐标处，使得直观的相似度估量值呈现在重构特征参数图像中的各个像素之间，将相似度高的像素点归属于同一微泡，将相似度不高的像素点分隔为不同的微泡，利用插值起到分隔不同微泡的作用，使重叠的微泡实现空间解耦，最终得到稀疏化图像。
在一个实施例中，相似度量化的方法包括：交叉熵估计或互相关估计。通过交叉熵估计或互相关估计定量地计算同位像素点集与关联像素点集的两个灰度涨落信号之间的相似度，将两个信号的相似度变为更直观的相似度估量值，将相似度值插入到重构特征参数图像，利用直观的数值将重叠的微泡进行空间解耦，最终得到稀疏化图像。
在一个实施例中，关联像素点集的个数为4个。在多帧待预处理配准造影图像中选取位于与第二像素坐标相邻的分别位于4个关联像素坐标处的4个关联像素点，形成同位像素点集对应的4个关联像素点集；分别计算同位像素点集与4个关联像素点集中每个关联像素点集的相似度，更详尽的获取一个像素点与周围的关联像素点的相似度，使插值后的重叠微泡的边界更加清晰，使重构特征参数图像中重叠的微泡进行空间解耦，得到稀疏化图像。
示例性重建方法
图4所示为本申请一实施例提供的一种超声造影图像的超分辨重建方法。如图4所示，该超分辨重建方法包括：
步骤401：从造影数据中选取至少一个待预处理图像集；
造影数据是被采集的多帧造影图像，在多帧准造影图像选取至少一个待预处理图像集，使得每一个待预处理图像集彼此之间是连续的，且每一个待预处理图像集的微泡具有稳定的空间位置。只要待预处理图像集中满足上述条件即可，本申请实施例对具体选取方法不做限定。
步骤402：按照前述任一超分辨重建预处理方法对至少一个待预处理图像集分别进行预处理，获取至少一帧稀疏化图像；
通过前述任一超分辨重建预处理方法对待预处理图像集进行预处理，有效降低高浓度微泡以及强噪声对超分辨成像的准确性的影响。通过对同位像素点集的灰度涨落信号进行分析，将微泡信号和噪声或背景信号区分开，提高信噪比与信背比；通过同位像素点集与关联像素点集的灰度涨落信号的相似度对重构特征参数图像进行插值，使重叠微泡实现空间解耦，解决衍射极限造成的微泡重叠。
步骤405：获取稀疏化图像中像素点的像素值和径向对称度估计值；
径向对称度是指每个像素点在局部区域内的径向对称度。通过像素点的梯度场的对称性，得到每个像素点在局部区域内的径向对称度。只要能获得稀疏化图像的径向对称度估计值即可，本申请实施例对径向对称度估计值的具体算法不做限定。
对至少一个待预处理图像集分别进行预处理，获取至少一帧稀疏化图像，获取每帧稀疏化图像中每一个像素点的像素值，以及获取每一个像素点的径向度估计值。由于微泡沿着运动轨迹方向会有变形，像素的径向对称度估计得到每个像素点在局部区域内的径向对称度，那么径向对称度估计就可以保留微泡的变形信息，因此对每个像素的像素值进行径向对程度估计就等价于保留微泡的运动轨迹。
步骤406：将位于同帧稀疏化图像中的像素点的像素值和径向对称度估计值进行加权计算，获得与至少一帧稀疏化图像分别对应的至少一帧局部超分辨图像；
在实际场景中，由于微泡是运动的，因此微泡会出现“拖尾”现象。通过将像素点的像素值和径向对称度估计值进行加权计算，可得到微泡区域的中心区域被增强，而微泡区域的边缘区域被弱化的局部超分辨图像。通过将像素点的像素值和径向对称度估计值进行加权计算，实现微泡点扩散函数瘦身，保留微泡的变形信息，使每帧稀疏化图像中微泡运动轴向的轨迹骨架变 得更加清晰，使微泡的非定位的运动轴向增强，保留了微泡沿运动方向的轨迹信息，获得局部超分辨图像。
步骤407：将至少一帧局部超分辨图像进行叠加获得重建的超分辨图像。
将每一帧局部超分辨图像叠加，形成完整的微小血管的超声超分辨重建图像。
本申请实施例中，通过对至少一个预处理图像集进行预处理，有效降低高浓度微泡以及强噪声对超分辨成像的准确性的影响；通过对预处理之后得到的每帧稀疏化图像中像素点的像素值和径向对称度估计值进行加权计算，实现微泡点扩散函数瘦身，保留微泡的变形信息，使每帧稀疏化图像中微泡运动轴向的轨迹骨架变得更加清晰，使微泡的非定位的运动轴向增强，保留了微泡沿运动方向的轨迹信息，获得局部超分辨图像，最终整合为完整的微小血管的超声超分辨重建图像。有别于传统的单点定位和累加策略，非定位的运动轴向增强方式保留了微泡运动轴向的轨迹骨架，从而大幅度提高了超分辨重建图像的空间分辨率及重建速度，实现快速高效的超声超分辨图像的重建。
图5所示为本申请一实施例提供的一种超声造影图像的超分辨重建方法中获取稀疏化图像中像素点的径向对称度估计值的流程示意图。如图5所示，获取稀疏化图像中像素点的径向对称度估计值包括：
步骤4051：在稀疏化图像中选取位于第三像素坐标的第三像素点；
第三像素坐标只是一个代指，可以是稀疏化图像中任何一个像素坐标点，本申请实施例对第三像素坐标的具体选取不做限定。
步骤4052：在同帧稀疏化图像中选取位于第三像素坐标周围的多个围绕像素点；
在同帧稀疏化图像中选取第三像素点周围的多个围绕像素点；例如：选取A1帧稀疏化图像的(1、1)像素坐标处的a像素点以及围绕在a像素点周围的12个围绕像素点。12个围绕像素点位于以a像素点为中心且半径为r的一个圆上，并且12个围绕像素点在此圆内均匀分布。
围绕像素点的个数可以6、8、10、12、15以及20等，本申请实施例对围绕像素点的具体个数不做限定。
步骤4053：对多个围绕像素点的像素值进行径向对称度估计，获得第三像素点的径向对称度估计值。
对a像素点以及围绕在a像素点的像素值和a像素点的周围的12个像素点的像素值进行径向对称度估计，获得a像素点的径向对称度估计值。
本申请实施例中，通过对同帧稀疏化图像中位于第三像素坐标的周围的多个围绕像素点的像素值进行径向对称度估计，保留微泡的变形信息，遍历获得每帧稀疏化图像上每一个像素点的径向对称度估计值，使微泡的非定位的运动轴向增强。
在一个实施例中，围绕像素点的个数为12，径向对称度半径为1。
图6所示为本申请一实施例提供的一种超声造影图像的超分辨重建方法中从造影数据中选取至少一个待预处理图像集的流程示意图。如图6所示，从造影数据中选取至少一个待预处理图像集具体包括：
步骤4011：采集多帧造影图像；
对待成像目标进行超声造影剂注射，用超声探头检测待成像区域，当出现造影增强信号时开始采集造影数据，获得多帧造影图像。
采集多帧造影图像的频率可以是30Hz也可以是50Hz，采集造影图像的频率只要满足具体临床采集条件即可，在本申请实施例对采集造影图像的频率不做具体限定。采集多帧造影图像可以是500帧、1000帧、或者1500帧等，本申请实施例对多帧造影图像的具体帧数不做限定。
步骤4012：对多帧造影图像进行配准以抑制组织运动干扰，获得多帧配准造影图像；
对多帧造影图像进行配准，以抑制探头和组织运动造成的图像变化，获得多帧配准后的图像。图像配准的方法可以是柔性配准和刚性配准，包括但不限于中值漂移法、尺度不变特征变换法、跟踪学习检测、光流法、互相关法。优选的，使用跟踪学习检测进行刚性配准，使用光流法进行柔性配准。只要可以对多帧造影图像进行配准，本申请实施例对配准的具体方法不 做限定。
步骤4013：在多帧准造影图像中每隔第一预设数量帧选取第二预设数量帧配准造影图像作为待预处理配准造影图像，获得至少一个待预处理图像集。
在多帧准造影图像中每隔第一预设数量帧选取第二预设数量帧配准造影图像，也就是在时间维度上选取相同时间窗(时间窗的时长为第二预设数量帧)，对相同时间窗内的配准造影图像进行去噪重构与插值计算解耦等预处理。例如：在N帧配准造影图像每间隔G帧帧选取W+1帧(G小于W+1，W+1小于N)配准造影图像作为待预处理配准造影图像，形成(N-W-1)/G+1个待预处理图像集。每个待预处理图像集包括W+1帧待预处理配准造影图像。按照上述方法在多帧准造影图像选取的待预处理图像集，使得每一个待预处理图像集是连续的，且进行预处理的待预处理配准造影图像的微泡具有稳定的空间位置。
N、G以及W都是正整数。从多帧配准造影图像中选取部分配准造影图像的选取方法可以是每隔1帧选取4帧；也可以是每隔2帧选取3帧；还可以是每隔1帧选取5帧，本申请实施例从多帧配准造影图像中选取部分配准造影图像的具体选取方法不做限定。
本申请实施例中，通过上述方法获取待预处理图像集，使得进行去噪重构以及插值计算是在一个时间窗内进行，使得每一个待预处理图像集中的微泡具有稳定的空间位置，降低微泡快速流动对超分辨重建的影响。
在一个实施例中，第一预设数量与血流速度负相关；和，第二预设数量与血流速度负相关。在一个实施例中，第一预设数量与成像帧率正相关；和，第二预设数量与成像帧率正相关。为了使微泡在一定时间窗内具有空间位置的稳定性，第一预设数量G和第一预设数量W+1取决于微泡流速和成像帧率，血流速度越快，第一预设数量G和第一预设数量W+1的取值越小，成像帧率越大，第一预设数量G和第一预设数量W+1的取值越大。
在一个实施例中，第一预设数量为1；和，第二预设数量帧数为4。
图7所示为本申请一实施例提供的新西兰大白兔下肢二头肌的超声造影图像的超分辨重建方法的流程图。其中，N＝1500，W+1＝4，G＝3。
采集新西兰大白兔下肢二头肌的1500帧超声造影图像(如图7中步骤7011所示)。采集造影数据，获得1500帧造影图像；对1500帧造影图像进行配准，以抑制探头和组织运动造成的图像变化，获得多帧1500帧配准造影图像(如图7中步骤7012所示)。选取时间窗时长为4帧，滑动时间窗，在1500帧准造影图像中每隔1帧选取为4帧配准造影图像作为待预处理配准造影图像，获得1497个待预处理图像集(如图7中步骤7013所示)。获取每个待预处理图像集中4帧待预处理配准造影图像的每个像素点的灰度涨落信号(如图7中步骤702所示)。通过4帧待预处理配准造影图像的每个同位像素点集的灰度涨落信号(如图7中步骤7031所示)；进行自相关估计或信息熵估计，提取出每个同位像素点集的对应的第一特征参数(如图7中步骤7032所示)；根据提取的第一特征参数与重构特征参数图像的映射关系(如图7中步骤7033所示)；就能重构出微泡信号被加强而背景或噪声被削弱的重构特征参数图像(如图7中步骤7034所示)；遍历1497个待预处理图像集中4帧待预处理配准造影图像中每个像素坐标，将1497个待预处理图像集去噪重构获得1497帧信噪比与信背比被改善的重构特征参数图像，实现大部分背景及噪声干扰的去除。
将4帧待预处理配准造影图像的每个同位像素点集的灰度涨落信号和与每个同位像素点集相邻的4个关联像素点集的灰度涨落信号(如图7中步骤7041与步骤7042所示)；通过交叉熵估计或互相关估计定量地计算，将两个信号的相似度变为更直观的相似度估量值(如图7中步骤7043所示)；遍历4帧待预处理配准造影图像中每个像素坐标，获得每个同位像素点集与相邻的4个关联像素点集的相似度估量值，将相似度估量值插入重构特征参数图像中对应的插值像素坐标处(如图7中步骤7044所示)，使得直观的相似度估量值呈现在重构特征参数图像中的各个像素之间；将重构特征参数图像中重叠的微泡进行空间解耦，得到稀疏化图像(如图7中步骤7045所示)。利用1497个每个待预处理图像集中4帧待预处理配准造影图像中同位像素点及与关联像素点集的相似度估量值，对1497帧重构特征参数图像进行插值，将1497帧重构特征参数图像中重叠的微泡实现空间解耦，最终得到1497帧稀疏化图像。
获取1497帧稀疏化图像中每帧稀疏化图像中每个像素点的像素值(如图7中步骤7050所示)；通过径向对称度估量对同帧稀疏化图像中与像素点对应的12个围绕像素点的像素值进行径向对称度估计获得该像素点的径向对称度估计值(如图7中步骤7051、步骤7052和步骤7052所示)；在每帧稀疏化图像中，遍历多稀疏化图像中每个像素点的像素值和径向对称度估计值进行加权计算，使每帧稀疏化图像中微泡运动轴向的轨迹骨架变得更加清晰，使微泡的非定位的运动轴向增强，从而保留了微泡沿运动方向的轨迹信息，获得清晰的局部超分辨图像(如图7中步骤706所示)；共计获得1497帧局部超分辨图像，将1497帧局部超分辨图叠加，形成完整的微小血管的超声超分辨重建图像(如图7中步骤707所示)。
本申请基于微泡在时间窗内的灰度涨落信号的波动特性，不仅有效滤除了背景或噪声信号，显著凸显了微泡信号，而且实现了对空间重叠微泡的高效解耦，从而能够大幅度提高超分辨精度。
示例性超分辨重建预处理装置
图8所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理装置的示意图。如图8所示，该超分辨重建预处理装置802包括：第一预处理获取模块8021，配置为获取待预处理图像集，待预处理图像集包含多帧待预处理配准造影图像；灰度获取模块8022，配置为获取待预处理配准造影图像中像素点的灰度涨落信号；去噪增强重构模块8023，配置为获取待预处理配准造影图像中像素点的灰度涨落信号；并基于同位像素点集的灰度涨落信号对待预处理图像集进行去噪重构得到重构特征参数图像，同位像素点集包括：在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个同位像素点；以及稀疏化模块8024，配置为基于同位像素点集以及与同位像素点集相关联的关联像素点集的灰度涨落信号对重构特征参数图像进行插值计算获得稀疏化图像，关联像素点集包括：在同帧待预处理配准造影图像中与同位像素点相邻，且在不同帧待预处理配准造影图像中位于相同像素坐标处的多个关联像素点。
在本申请实施例中，通过去噪增强重构模块对同位像素点集的灰度涨落信号进行分析，使待预处理图像集的微泡信号和噪声或背景信号区分开，提高信噪比与信背比，得到微泡信号增强与背景噪声信号削弱的重构特征参数图像；通过稀疏化模块对同位像素点集以及关联像素点集的灰度涨落信号的相似度对重构特征参数图像进行插值，使不同微泡分隔开，使重叠的微泡实现空间解耦，有效降低强噪声以及高浓度微泡对重建的影响。
在一个实施例中，灰度涨落信号为像素点的像素值按照时间顺序排列形成的信号；同位像素点集的灰度涨落信号为同位像素点集中的多个同位像素点的像素值按照时间顺序排列形成的1乘多维信号。
图9所示为本申请一实施例提供的一种超声造影图像的超分辨重建预处理装置的示意图。如图9所示，去噪增强重构模块8023进一步包括：第一像素点集获取单元80231，配置为在多帧待预处理配准造影图像中选取位于第一像素坐标处的多个第一同位像素点，形成第一同位像素点集；特征参数提取单元80232，配置为对多个第一同位像素点的灰度涨落信号进行特征估计提取得到第一特征参数，第一特征参数的用于表征第一同位像素点的灰度涨落分布的随机性与灰度涨落的周期性；预设单元80233，配置为预设第一特征参数与重构特征参数图像的映射关系；以及重构单元80234，配置为将第一特征参数对应到待重构特征参数图像的第一像素坐标处，根据映射关系，得到重构特征参数图像。
在一个实施例中，特征估计提取的方法包括：自相关估计计算或信息熵估计计算。
在一个实施例中，如图9所示，稀疏化模块8024包括：第二像素点集获取单元80241，配置为在多帧待预处理配准造影图像中选取位于第二像素坐标处的多个第二同位像素点，形成第二同位像素点集；关联像素点获取单元80242，配置为在多帧待预处理配准造影图像中选取位于与第二像素坐标相邻的关联像素坐标处的多个关联像素点，形成关联像素点集；相似度量化单元80243，配置为对第二同位像素点集的灰度涨落信号与关联像素点集的灰度涨落信号进行相似度量化得到相似度估量值；插值单元80244，配置为将相似度估量值插入所述重构特征参数图像中对应的插值像素坐标处，插值像素坐标位于所述第二像素坐标与所述关联像素坐标之间；以及稀疏解耦单元80245，配置为利用插值像素坐标的位置，将重构特征参数图像中重 叠的微泡进行空间解耦，得到稀疏化图像。
在一个实施例中，相似度量化的方法包括：交叉熵估计或互相关估计。
在一个实施例中，关联像素点集的个数为4个。
示例性超分辨重建装置
图10所示为本申请一实施例提供的一种超声造影图像的超分辨重建装置的示意图。如图10所示，该超分辨重建装置800包括：第一选取模块801，配置为从造影数据中选取至少一个待预处理图像集；超分辨重建预处理装置802，配置采用前述任一超分辨重建预处理方法对至少一个待预处理图像集分别进行预处理获取至少一帧稀疏化图像；轴向轨迹凸显模块803；配置为获取稀疏化图像中像素点的像素值和径向对称度估计值；并将位于同帧稀疏化图像中的像素点的像素值和径向对称度估计值进行加权计算，获得与至少一帧稀疏化图像分别对应的至少一帧局部超分辨图像；以及叠加模块804，配置为将至少一帧局部超分辨图像进行叠加获得重建的超分辨图像。
本申请实施例中，通过预处理装置对至少一个预处理图像集进行预处理，有效降低高浓度微泡以及强噪声对超分辨成像的准确性的影响；通过轴向轨迹凸显模块对每帧稀疏化图像中像素点的像素值和径向对称度和径向对称度估计值进行加权计算，实现微泡点扩散函数瘦身，保留微泡的变形信息，保留微泡的运动轨迹，从而使微泡的非定位的运动轴向增强，进而保留了微泡运动轴向的轨迹骨架，获得局部超分辨图像；通过叠加模块，最终获得整合的完整的微小血管的超声超分辨重建图像。有别于传统的单点定位和累加策略的重建设备，本重建设备通过非定位的运动轴向增强方式保留了微泡运动轴向的轨迹骨架，从而大幅度提高了超分辨重建图像的空间分辨率及重建速度，实现快速高效的超声超分辨图像的重建。
图11所示为本申请一实施例提供的一种超声造影图像的超分辨重建装置的示意图。如图11所示，轴向轨迹凸显模块803包括：第三像素点获取单元8031，在稀疏化图像中选取位于第三像素坐标的第三像素点；围绕像素点获取单元8032，配置为在同帧稀疏化图像中选取位于第三像素坐标周围的多个围绕像素点；以及径向对称度估计单元8033，配置为对多个围绕像素点的像素值进行径向对称度估计，获得第三像素点的径向对称度估计值。
在一个实施例中，如图11所示，第一选取模块801进一步包括：采集单元8011，配置为采集多帧造影图像；配准单元8012，配置为对多帧造影图像进行配准以抑制组织运动干扰，获得多帧配准造影图像；以及筛选单元8013，配置为在多帧准造影图像中每隔第一预设数量帧选取第二预设数量帧配准造影图像作为待预处理配准造影图像，获得至少一个待预处理图像集。
在一个实施例中，第一预设数量与血流速度负相关；和，第二预设数量与血流速度负相关。
在一个实施例中，第一预设数量与成像帧率正相关；和，第二预设数量与成像帧率正相关。
在一个实施例中，第一预设数量为1；和，第二预设数量帧数为4。
上述超声造影图像的超分辨重建装置中的预处理装置的具体功能和操作已经在上面参考图1到图3所示的描述的超分辨重建预处理方法进行了详细介绍；重建装置中其他各个模块的具体功能和操作图4到图6描述的超分辨重建方法中进行了详细介绍，因此，这里将省略其重复描述。
需要说明的是，根据本申请实施例的超声造影图像的超分辨重建装置800可以作为一个软件模块和/或硬件模块而集成到电子设备1200中，换言之，该电子设备1200可以包括该超声造影图像的超分辨重建装置800。例如，该超声造影图像的超分辨重建装置800可以是该电子设备1200的操作系统中的一个软件模块，或者可以是针对于其所开发的一个应用程序；当然，该超声造影图像的超分辨重建装置800同样可以是该电子设备1200的众多硬件模块之一。
在本申请另一实施例中，该超声造影图像的超分辨重建装置800与该电子设备1200也可以是分立的设备(例如，服务器)，并且该超声造影图像的超分辨重建装置800可以通过有线和/或无线网络连接到该电子设备1200，并且按照约定的数据格式来传输交互信息。
示例性电子设备
图12所示为本申请一实施例提供的电子设备的结构示意图。如图12所示，该电子设备1200包括：一个或多个处理器1201和存储器1202；以及存储在存储器1202中的计算机程序指令，计算机程序指令在被处理器1201运行时使得处理器1201执行如上述任一实施例的超分辨重建预处理方法或上述任一实施例的超分辨重建方法。
处理器1201可以是中央处理单元(CPU)或者具有数据处理能力和/或指令执行能力的其他形式的处理单元，并且可以控制电子设备中的其他组件以执行期望的功能。
存储器1202可以包括一个或多个计算机程序产品，所述计算机程序产品可以包括各种形式的计算机可读存储介质，例如易失性存储器和/或非易失性存储器。所述易失性存储器例如可以包括随机存取存储器(RAM)和/或高速缓冲存储器(cache)等。所述非易失性存储器例如可以包括只读存储器(ROM)、硬盘、闪存等。在所述计算机可读存储介质上可以存储一个或多个计算机程序指令，处理器1701可以运行所述程序指令，以实现上文所述的本申请如上述任一实施例的超分辨重建预处理方法或上述任一实施例的超分辨重建方法的步骤以及/或者其他期望的功能。在所述计算机可读存储介质中还可以存储诸如光线强度、补偿光强度、滤光片的位置等信息。
在一个示例中，电子设备1200还可以包括：输入装置1203和输出装置1204，这些组件通过总线系统和/或其他形式的连接机构(图12中未示出)互连。
此外，该输入设备1203还可以包括例如键盘、鼠标、麦克风等等。
该输出装置1204可以向外部输出各种信息，例如可以包括例如显示器、扬声器、打印机、以及通信网络及其所连接的远程输出设备等等。
当然，为了简化，图12中仅示出了该电子设备1200中与本申请有关的组件中的一些，省略了诸如总线、输入装置/输出接口等组件。除此之外，根据具体应用情况，电子设备1200还可以包括任何其他适当的组件。
示例性计算机程序产品和计算机可读存储介质
除了上述方法和设备以外，本申请的实施例还可以是计算机程序产品，包括计算机程序指令，计算机程序指令在被处理器运行时使得处理器执行如上述任一实施例的超分辨重建预处理方法或上述任一实施例的超分辨重建方法的中的步骤。
计算机程序产品可以以一种或多种程序设计语言的任意组合来编写用于执行本申请实施例操作的程序代码，所述程序设计语言包括面向对象的程序设计语言，诸如Java、C++等，还包括常规的过程式程序设计语言，诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算设备上执行、部分地在用户设备上执行、作为一个独立的软件包执行、部分在用户计算设备上部分在远程计算设备上执行、或者完全在远程计算设备或服务器上执行。
此外，本申请的实施例还可以是计算机可读存储介质，其上存储有计算机程序指令，所述计算机程序指令在被处理器运行时使得所述处理器执行本说明书上述“示例性超分辨重建预处理方法”或“示例性超分辨重建方法”部分中描述的根据本申请上述任一实施例的预处理方法或上述任一实施例的重建方法的步骤。
所述计算机可读存储介质可以采用一个或多个可读介质的任意组合。可读介质可以是可读信号介质或者可读存储介质。可读存储介质例如可以包括但不限于电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式盘、硬盘、随机存取存储器((RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。
以上结合具体实施例描述了本申请的基本原理，但是，需要指出的是，在本申请中提及的优点、优势、效果等仅是示例而非限制，不能认为这些优点、优势、效果等是本申请的各个实施例必须具备的。另外，上述公开的具体细节仅是为了示例的作用和便于理解的作用，而非限制，上述细节并不限制本申请为必须采用上述具体的细节来实现。
本申请中涉及的器件、装置、设备、系统的方框图仅作为例示性的例子并且不意图要求或暗示必须按照方框图示出的方式进行连接、布置、配置。如本领域技术人员将认识到的，可 以按任意方式连接、布置、配置这些器件、装置、设备、系统。诸如“包括”、“包含”、“具有”等等的词语是开放性词汇，指“包括但不限于”，且可与其互换使用。这里所使用的词汇“或”和“和”指词汇“和/或”，且可与其互换使用，除非上下文明确指示不是如此。这里所使用的词汇“诸如”指词组“诸如但不限于”，且可与其互换使用。
还需要指出的是，在本申请的装置、设备和方法中，各部件或各步骤是可以分解和/或重新组合的。这些分解和/或重新组合应视为本申请的等效方案。
提供所公开的方面的以上描述以使本领域的任何技术人员能够做出或者使用本申请。对这些方面的各种修改对于本领域技术人员而言是非常显而易见的，并且在此定义的一般原理可以应用于其他方面而不脱离本申请的范围。因此，本申请不意图被限制到在此示出的方面，而是按照与在此公开的原理和新颖的特征一致的最宽范围。
为了例示和描述的目的已经给出了以上描述。此外，此描述不意图将本申请的实施例限制到在此公开的形式。尽管以上已经讨论了多个示例方面和实施例，但是本领域技术人员将认识到其某些变型、修改、改变、添加和子组合。
以上所述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所作的任何修改、等同替换等，均应包含在本申请的保护范围之内。

Claims (19)

  1. 一种超声造影图像的超分辨重建预处理方法，包括：
    获取待预处理图像集，所述待预处理图像集包含多帧待预处理配准造影图像；
    获取所述待预处理配准造影图像中像素点的灰度涨落信号；
    基于同位像素点集的灰度涨落信号对所述待预处理图像集进行去噪重构得到重构特征参数图像，所述同位像素点集包括：在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个同位像素点；以及
    基于所述同位像素点集以及与所述同位像素点集相关联的关联像素点集的灰度涨落信号对所述重构特征参数图像进行插值计算获得稀疏化图像，所述关联像素点集包括：在同帧所述待预处理配准造影图像中与所述同位像素点相邻，且在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个关联像素点。
  2. 根据权利要求1所述的超分辨重建预处理方法，其中，所述灰度涨落信号为所述像素点的像素值按照时间顺序排列形成的信号；所述同位像素点集的灰度涨落信号为所述同位像素点集中的多个同位像素点的像素值按照时间顺序排列形成的1乘多维信号。
  3. 根据权利要求1或2所述的超分辨重建预处理方法，其中，所述基于同位像素点集的灰度涨落信号对所述待预处理图像集进行去噪重构得到重构特征参数图像包括：
    在所述多帧待预处理配准造影图像中选取位于第一像素坐标处的多个第一同位像素点，形成第一同位像素点集；
    对所述多个第一同位像素点的灰度涨落信号进行特征估计提取得到第一特征参数，所述第一特征参数的用于表征所述第一同位像素点的灰度涨落分布的随机性与灰度涨落的周期性；
    预设所述第一特征参数与重构特征参数图像的映射关系，以及
    将所述第一特征参数对应到待重构特征参数图像的所述第一像素坐标处，根据所述映射关系，得到所述重构特征参数图像。
  4. 根据权利要求3所述的超分辨重建预处理方法，其中，所述特征估计提取的方法包括：自相关估计计算或信息熵估计计算。
  5. 根据权利要求3或4所述的超分辨重建预处理方法，其中，所述第一特征参数的值越大，所述重构特征参数图像中与所述第一特征参数对应的像素点的灰度值越高。
  6. 根据权利要求1至5中的任一项所述的超分辨重建预处理方法，其中，所述基于所述同位像素点集以及与所述同位像素点集相关联的关联像素点集的灰度涨落信号对所述重构特征参数图像进行插值计算获得稀疏化图像包括：
    在所述多帧待预处理配准造影图像中选取位于第二像素坐标处的多个第二同位像素点，形成第二同位像素点集；
    在所述多帧待预处理配准造影图像中选取位于与所述第二像素坐标相邻的关联像素坐标处的多个关联像素点，形成所述关联像素点集；
    对所述第二同位像素点集的灰度涨落信号与所述关联像素点集的灰度涨落信号进行相似度量化得到相似度估量值；
    将所述相似度估量值插入所述重构特征参数图像中对应的插值像素坐标处，所述插值像素坐标位于所述第二像素坐标与所述关联像素坐标之间；以及
    利用所述插值像素坐标的位置，将所述重构特征参数图像中重叠的微泡进行空间解耦，得到所述稀疏化图像。
  7. 根据权利要求6所述的超分辨重建预处理方法，其中，所述相似度量化的方法包括：交叉熵估计或互相关估计。
  8. 根据权利要求1至7中任一项所述的超分辨重建预处理方法，其中，所述关联像素点集的个数为4个。
  9. 一种超声造影图像的超分辨重建方法，包括：
    从造影数据中选取至少一个待预处理图像集；
    对所述至少一个待预处理图像集分别进行预处理，获取至少一帧稀疏化图像，所述预处理的方法采用如权利要求1至8中任一项所述的超分辨重建预处理方法；
    获取所述稀疏化图像中像素点的像素值和径向对称度估计值；
    将位于同帧所述稀疏化图像中的像素点的像素值和径向对称度估计值进行加权计算，获得与所述至少一帧稀疏化图像分别对应的至少一帧局部超分辨图像；以及
    将所述至少一帧局部超分辨图像进行叠加获得重建的超分辨图像。
  10. 据权利要求9所述的超分辨重建方法，其中，获取所述稀疏化图像中像素点的径向对称度估计值包括：
    在所述稀疏化图像中选取位于第三像素坐标的第三像素点；
    在同帧稀疏化图像中选取位于所述第三像素坐标周围的多个围绕像素点；以及
    对所述多个围绕像素点的像素值进行径向对称度估计，获得所述第三像素点的径向对称度估计值。
  11. 据权利要求10所述的超分辨重建方法，其中，所述围绕像素点的个数为12个。
  12. 据权利要求9至11中任一项所述的超分辨重建方法，其中，所述从造影数据中选取至少一个待预处理图像集包括：
    采集多帧造影图像；
    对所述多帧造影图像进行配准以抑制组织运动干扰，获得多帧配准造影图像；以及
    在所述多帧准造影图像中每隔第一预设数量帧选取第二预设数量帧配准造影图像作为待预处理配准造影图像，获得至少一个待预处理图像集。
  13. 根据权利要求12所述的超分辨重建方法，其中，
    所述第一预设数量与血流速度负相关；和，所述第二预设数量与血流速度负相关。
  14. 根据权利要求12或13所述的超分辨重建方法，其中，所述第一预设数量与成像帧率正相关；和，所述第二预设数量与成像帧率正相关。
  15. 根据权利要求12至14中任一项所述的超分辨重建方法，其中，
    所述第一预设数量为1；和，所述第二预设数量帧数为4。
  16. 一种超声造影图像的超分辨重建预处理装置，包括：
    第一预处理获取模块，配置为获取待预处理图像集，所述待预处理图像集包含多帧待预处理配准造影图像；
    灰度获取模块，配置为获取所述待预处理配准造影图像中像素点的灰度涨落信号；
    去噪增强重构模块，配置为基于同位像素点集的灰度涨落信号对所述待预处理图像集进行去噪重构得到重构特征参数图像，所述同位像素点集包括：在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个同位像素点；以及
    稀疏化模块，配置为基于所述同位像素点集以及与所述同位像素点集相关联的关联像素点集的灰度涨落信号对所述重构特征参数图像进行插值计算获得稀疏化图像，所述关联像素点集包括：在同帧所述待预处理配准造影图像中与所述同位像素点相邻，且在不同帧所述待预处理配准造影图像中位于相同像素坐标处的多个关联像素点。
  17. 一种超声造影图像的超分辨重建装置，包括：
    第一选取模块，配置为从造影数据中选取至少一个待预处理图像集；
    预处理装置，配置为采用如权利要求1至8中任一项所述的超分辨重建预处理方法，对所述至少一个待预处理图像集分别进行预处理获取至少一帧稀疏化图像；
    轴向轨迹凸显模块，配置为获取所述稀疏化图像中像素点的像素值和径向对称度估计值；并将位于同帧所述稀疏化图像中的像素点的像素值和径向对称度估计值进行加权计算，获得与所述至少一帧稀疏化图像分别对应的至少一帧局部超分辨图像；以及
    叠加模块，配置为将所述至少一帧局部超分辨图像进行叠加获得重建的超分辨图像。
  18. 一种电子设备，包括：
    处理器；以及
    存储器，在所述存储器中存储有计算机程序指令，所述计算机程序指令在被所述处理器运行时使得所述处理器执行如权利要求1至8中任一项所述的超分辨重建预处理方法或如权利要求9至15中任一项所述的超分辨重建方法。
  19. 一种计算机可读存储介质，所述计算机可读存储介质上存储有计算机程序指令，所述计算机程序指令在被处理器运行时使得所述处理器执行如权利要求1至8中任一项所述的超分辨重建预处理方法或如权利要求9至15中任一项所述的超分辨重建方法。
PCT/CN2021/093327 2020-05-18 2021-05-12 超声造影图像的超分辨重建预处理方法和超分辨重建方法 WO2021233178A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP21808993.6A EP4060616B1 (en) 2020-05-18 2021-05-12 Super-resolution reconstruction preprocessing method and super-resolution reconstruction method for ultrasound contrast image
US17/837,973 US11798150B2 (en) 2020-05-18 2022-06-10 Super-resolution reconstruction preprocessing method and super-resolution reconstruction method for contrast-enhanced ultrasound images

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010419625.7 2020-05-18
CN202010419625.7A CN111598965B (zh) 2020-05-18 2020-05-18 超声造影图像的超分辨重建预处理方法和超分辨重建方法

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/837,973 Continuation US11798150B2 (en) 2020-05-18 2022-06-10 Super-resolution reconstruction preprocessing method and super-resolution reconstruction method for contrast-enhanced ultrasound images

Publications (1)

Publication Number Publication Date
WO2021233178A1 true WO2021233178A1 (zh) 2021-11-25

Family

ID=72187282

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/093327 WO2021233178A1 (zh) 2020-05-18 2021-05-12 超声造影图像的超分辨重建预处理方法和超分辨重建方法

Country Status (4)

Country Link
US (1) US11798150B2 (zh)
EP (1) EP4060616B1 (zh)
CN (1) CN111598965B (zh)
WO (1) WO2021233178A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111598965B (zh) * 2020-05-18 2023-07-14 南京超维景生物科技有限公司 超声造影图像的超分辨重建预处理方法和超分辨重建方法
CN113240587B (zh) * 2021-07-12 2021-09-24 深圳华声医疗技术股份有限公司 超分辨率扫描变换方法、装置、超声设备及存储介质
CN113855235B (zh) * 2021-08-02 2024-06-14 应葵 用于肝脏部位的微波热消融手术中磁共振导航方法及装置
CN114866766A (zh) * 2022-07-04 2022-08-05 广东欧谱曼迪科技有限公司 灵敏度评价方法、测试方法、装置、电子设备及存储介质
CN116805284B (zh) * 2023-08-28 2023-12-19 之江实验室 基于特征迁移的三维磁共振平面间超分辨重建方法和系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7744534B2 (en) * 2002-06-07 2010-06-29 Verathon Inc. 3D ultrasound-based instrument for non-invasive measurement of amniotic fluid volume
CN103876776A (zh) * 2012-12-24 2014-06-25 深圳迈瑞生物医疗电子股份有限公司 一种超声造影成像方法及装置
CN107361791A (zh) * 2017-07-21 2017-11-21 北京大学 一种快速超分辨血流成像方法
CN109215093A (zh) * 2018-07-27 2019-01-15 深圳先进技术研究院 低剂量pet图像重建方法、装置、设备及存储介质
CN111598965A (zh) * 2020-05-18 2020-08-28 南京超维景生物科技有限公司 超声造影图像的超分辨重建预处理方法和超分辨重建方法

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104106079A (zh) * 2011-09-09 2014-10-15 帕那莫夫公司 图像处理系统和方法
CN103167218B (zh) * 2011-12-14 2016-04-06 北京大学 一种基于非局部性的超分辨率重建方法和设备
EP2890300B1 (en) * 2012-08-31 2019-01-02 Kenji Suzuki Supervised machine learning technique for reduction of radiation dose in computed tomography imaging
US9646366B2 (en) * 2012-11-30 2017-05-09 Change Healthcare Llc Method and apparatus for enhancing medical images
EP3025166B1 (en) * 2013-07-23 2020-06-17 Regents of the University of Minnesota Ultrasound image formation and/or reconstruction using multiple frequency waveforms
EP3125765A4 (en) * 2014-03-04 2018-04-18 The Trustees of Columbia University in the City of New York Regularization of images
WO2017042280A1 (en) * 2015-09-09 2017-03-16 Technische Universiteit Eindhoven Imaging of dispersion and velocity of contrast agents
US11589840B2 (en) * 2017-05-31 2023-02-28 Mayo Foundation For Medical Education And Research Methods for super-resolution ultrasound imaging of microvessels
GB201716606D0 (en) * 2017-10-10 2017-11-22 Heriot-Watt Univ Tracking in ultrasound imaging
WO2020081915A1 (en) * 2018-10-19 2020-04-23 Mayo Foundation For Medical Education And Research Systems and methods for kalman filter-based microvessel inpainting for super-resolution imaging
US20220011270A1 (en) * 2018-11-09 2022-01-13 Georgia Tech Research Corporation Systems and methods for ultrasound imaging and focusing
WO2020141127A1 (en) * 2019-01-03 2020-07-09 Koninklijke Philips N.V. Systems and methods for contrast enhanced imaging
CN110772285B (zh) * 2019-10-31 2022-05-17 南京景瑞康分子医药科技有限公司 一种超声超分辨成像方法
US20210321988A1 (en) * 2020-04-20 2021-10-21 Shenzhen Mindray Bio-Medical Electronics Co., Ltd. System and method for contrast enhanced ultrasound quantification imaging

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7744534B2 (en) * 2002-06-07 2010-06-29 Verathon Inc. 3D ultrasound-based instrument for non-invasive measurement of amniotic fluid volume
CN103876776A (zh) * 2012-12-24 2014-06-25 深圳迈瑞生物医疗电子股份有限公司 一种超声造影成像方法及装置
CN107361791A (zh) * 2017-07-21 2017-11-21 北京大学 一种快速超分辨血流成像方法
CN109215093A (zh) * 2018-07-27 2019-01-15 深圳先进技术研究院 低剂量pet图像重建方法、装置、设备及存储介质
CN111598965A (zh) * 2020-05-18 2020-08-28 南京超维景生物科技有限公司 超声造影图像的超分辨重建预处理方法和超分辨重建方法

Also Published As

Publication number Publication date
CN111598965A (zh) 2020-08-28
US11798150B2 (en) 2023-10-24
EP4060616A4 (en) 2023-06-14
EP4060616A1 (en) 2022-09-21
CN111598965B (zh) 2023-07-14
EP4060616B1 (en) 2024-04-03
US20220301132A1 (en) 2022-09-22

Similar Documents

Publication Publication Date Title
WO2021233178A1 (zh) 超声造影图像的超分辨重建预处理方法和超分辨重建方法
CN107361791B (zh) 一种快速超分辨血流成像方法
US9179881B2 (en) Physics based image processing and evaluation process of perfusion images from radiology imaging
JP2023138502A (ja) Ｘ線画像特徴検出および位置合わせのシステムおよび方法
US10039501B2 (en) Computer-aided diagnosis (CAD) apparatus and method using consecutive medical images
Wong et al. Cardiac flow analysis applied to phase contrast magnetic resonance imaging of the heart
WO2021232867A1 (zh) 超分辨流场重建方法及其装置
US8313437B1 (en) Vascular ultrasound intima-media thickness (IMT) measurement system
Raza et al. Brain image representation and rendering: A survey
US20120063656A1 (en) Efficient mapping of tissue properties from unregistered data with low signal-to-noise ratio
JP2022515464A (ja) 人工知能に基づく血流区間の分類方法及びシステム
Cao et al. Multiparameter synchronous measurement with IVUS images for intelligently diagnosing coronary cardiac disease
CN109726752A (zh) 基于时间信号曲线的灌注动态影像的分割方法及系统
EP4059442B1 (en) Image processing method and apparatus based on ultrasound contrast image
Chen et al. All answers are in the images: A review of deep learning for cerebrovascular segmentation
Storve et al. Realtime automatic assessment of cardiac function in echocardiography
CN112651984A (zh) 血管管腔内膜轮廓提取方法、装置及超声设备和存储介质
Hennersperger et al. Vascular 3D+ T freehand ultrasound using correlation of doppler and pulse-oximetry data
JP6934948B2 (ja) 流体解析装置および流体解析装置の作動方法並びに流体解析プログラム
JP2023152885A (ja) 医用画像処理装置、医用画像処理方法、及び、医用画像処理プログラム
WO2018159708A1 (ja) 血流解析装置および方法並びにプログラム
Liu et al. Multislice left ventricular ejection fraction prediction from cardiac MRIs without segmentation using shared SptDenNet
CN110097502B (zh) 眼底无灌注区的测量方法及装置及图像处理方法
WO2023189261A1 (ja) コンピュータプログラム、情報処理装置及び情報処理方法
Kanwal et al. IA2SKAbs: Intelligent automated and accurate system for classification of kidney abnormalities

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21808993

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2021808993

Country of ref document: EP

Effective date: 20220615

NENP Non-entry into the national phase

Ref country code: DE