CN-122004938-A - Noise suppression processing method and system for ultrasonic microbubble imaging

CN122004938ACN 122004938 ACN122004938 ACN 122004938ACN-122004938-A

Abstract

The invention discloses a noise suppression processing method and a system for ultrasonic microbubble imaging, and relates to the field of medical image processing, wherein the method comprises the steps of obtaining peak sparse pressure and shell elasticity parameters, and obtaining tissue depth information and fat content indexes; calculating local effective sound pressure after tissue attenuation according to peak sparse pressure, tissue depth information and fat content indexes, determining a basic rupture threshold, correcting the basic rupture threshold to obtain a corrected rupture threshold, comparing the local effective sound pressure with the corrected rupture threshold to obtain a microbubble rupture risk coefficient, generating a rupture risk distribution map, performing time-frequency analysis on an ultrasonic echo signal in the rupture risk distribution map, identifying candidate rupture noise, filtering through coherence verification to obtain a noise-filtered echo signal, and reconstructing an ultrasonic microbubble image. The method solves the problems that noise and nonlinear harmonic signals are seriously aliased, image quality is interfered, and specific noise cannot be accurately separated.

Inventors

ZHAO PING
WANG LI

Assignees

中国人民解放军空军军医大学

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. A noise suppression processing method for ultrasonic microbubble imaging, the method comprising: Acquiring peak sparse pressure of emitted ultrasonic pulses and shell elasticity parameters of an ultrasonic microbubble contrast agent, and acquiring tissue depth information and fat content indexes of each pixel point in an imaging region; Calculating local effective sound pressure after tissue attenuation at each pixel point according to the peak sparse pressure, the tissue depth information of each pixel point and the corresponding fat content index; Determining a basic rupture threshold according to the shell elasticity parameter, and correcting the basic rupture threshold according to the fat content index of each pixel point to obtain a corrected rupture threshold of each pixel point; Comparing the local effective sound pressure with the corrected fracture threshold value for each pixel point in the imaging area, calculating to obtain a microbubble fracture risk coefficient, integrating the microbubble fracture risk coefficients of all the pixel points, and generating a fracture risk distribution map; based on the fracture risk distribution map, the fat content index and the tissue depth information, performing feature analysis and noise identification on an ultrasonic echo signal to identify candidate fracture noise; based on coherence calculation, judging real fracture noise from the candidate fracture noise, filtering to obtain a pure echo signal, and reconstructing and generating an ultrasonic microbubble image by using the pure echo signal.
2. The noise suppression processing method for ultrasonic microbubble imaging according to claim 1, wherein acquiring the peak sparse pressure of the transmitted ultrasonic pulse, the shell elasticity parameter of the ultrasonic microbubble contrast agent, and the tissue depth information and the fat content index of each pixel point in the imaging region, comprises: Acquiring a peak sparse pressure of an emitted ultrasonic pulse and a shell elasticity parameter of an ultrasonic microbubble contrast agent; acquiring an ultrasonic echo signal, and determining ultrasonic echo propagation time corresponding to each pixel point based on the ultrasonic echo signal; Multiplying the ultrasonic echo propagation time with a preset tissue average sound velocity, dividing the multiplied ultrasonic echo propagation time by two, and calculating to obtain tissue depth information of each pixel point; Setting a plurality of analysis windows in an imaging region; For each analysis window, respectively acquiring a near-field region ultrasonic radio frequency echo signal close to the ultrasonic probe in the window as a shallow radio frequency signal and a far-field region ultrasonic radio frequency echo signal far from the ultrasonic probe as a deep radio frequency signal; Performing Fourier transform on the shallow radio frequency signal and the deep radio frequency signal respectively to obtain a shallow power spectrum and a deep power spectrum; Calculating a logarithmic difference spectrum between the deep power spectrum and the shallow power spectrum; performing linear fitting on the logarithmic difference spectrum by adopting a least square method, and taking a linear slope value obtained by fitting as an acoustic attenuation coefficient slope of an analysis window; And generating fat content indexes corresponding to all pixel points in the imaging region through spatial interpolation calculation based on the slopes of the sound attenuation coefficients of all the analysis windows.
3. The noise suppression processing method for ultrasonic microbubble imaging according to claim 1, characterized in that calculating a local effective sound pressure after tissue attenuation at each pixel point according to the peak sparse pressure, tissue depth information of each pixel point, and a corresponding fat content index, comprising: Inquiring a mapping relation table between a preset fat content index and tissue sound attenuation coefficients according to the fat content index of each pixel point, and determining the tissue sound attenuation coefficients corresponding to each pixel point; substituting the peak sparse pressure, the tissue depth information of each pixel point and the corresponding tissue sound attenuation coefficient into a sound pressure attenuation formula to calculate so as to obtain local effective sound pressure at each pixel point.
4. The noise suppression processing method for ultrasonic microbubble imaging according to claim 1, characterized in that determining a base rupture threshold according to the shell elasticity parameter, and correcting the base rupture threshold according to a fat content index of each pixel point to obtain a corrected rupture threshold of each pixel point, comprising: determining a basal collapse threshold of the microbubbles based on the shell elasticity parameter; Calculating depth-fat correlation characteristic values and local spatial gradient characteristic values corresponding to each pixel point based on tissue depth information and fat content indexes of each pixel point; calculating correction factors corresponding to all pixel points by adopting a linear weighting formula according to the depth-fat correlation characteristic values and the local spatial gradient characteristic values; and multiplying the basic rupture threshold value by a correction factor corresponding to each pixel point to obtain a correction rupture threshold value of each pixel point.
5. The noise suppression processing method for ultrasonic microbubble imaging according to claim 4, characterized in that calculating a depth-fat associated feature value and a local spatial gradient feature value corresponding to each pixel point based on tissue depth information and a fat content index of each pixel point, comprising: For each pixel point, multiplying the tissue depth information value of the pixel point by the fat content index value, and multiplying the tissue depth information value by a logarithmic value obtained by adding one to the tissue depth information value to obtain a depth-fat correlation characteristic value of the pixel point; And taking the pixel points as the centers, and calculating standard deviation of fat content indexes of all the pixel points in a preset square neighborhood window to be used as local spatial gradient characteristic values of the pixel points.
6. The noise suppression processing method for ultrasonic microbubble imaging according to claim 1, characterized in that for each pixel point in an imaging area, comparing the local effective sound pressure with the corrected fracture threshold value, calculating a microbubble fracture risk coefficient, and integrating the microbubble fracture risk coefficients of all the pixel points to generate a fracture risk profile, comprising: For each pixel point in an imaging area, calculating the ratio of the local effective sound pressure of the pixel point to the corrected cracking threshold value as an initial risk ratio; based on the depth-fat associated characteristic value and the local spatial gradient characteristic value corresponding to the pixel point, obtaining a comprehensive weighting factor through normalization and weighting calculation; multiplying the initial risk ratio by the comprehensive weighting factor to obtain a microbubble rupture risk coefficient of the pixel point; And collecting microbubble rupture risk coefficients of all pixel points in the imaging area, and generating a rupture risk distribution map according to the spatial position arrangement of each pixel point.
7. The noise suppression processing method for ultrasonic microbubble imaging according to claim 1, characterized in that performing feature analysis and noise discrimination on an ultrasonic echo signal based on the fracture risk profile, the fat content index, and the tissue depth information, identifying candidate fracture noise, comprising: For each pixel point, extracting a microbubble rupture risk coefficient, a fat content index, tissue depth information, a depth-fat associated characteristic value and a local spatial gradient characteristic value of the pixel point to form a tissue characteristic parameter vector; performing time-frequency transformation on the ultrasonic echo signals corresponding to the pixel points, extracting time-frequency domain features, and forming signal feature vectors, wherein the time-frequency domain features comprise peak frequency, frequency bandwidth, time domain peak amplitude and frequency domain energy; Respectively performing Z-score standardization processing on the tissue characteristic parameter vector and the signal characteristic vector of the pixel point, and then splicing to form a fusion characteristic vector; inputting the fusion feature vector into a pre-trained noise classification model, and outputting a breaking noise probability that a pixel point signal is breaking noise; Based on the tissue characteristic parameter vector and the basic discrimination threshold of the pixel point, obtaining a dynamic discrimination threshold of the pixel point through weighted calculation; And comparing the fracture noise probability with the dynamic discrimination threshold, marking the signal component with the fracture noise probability larger than the dynamic discrimination threshold as candidate fracture noise, and recording corresponding time information and position information.
8. The noise suppression processing method for ultrasonic microbubble imaging according to claim 7, characterized in that the construction process of the noise classification model comprises: acquiring a plurality of groups of sample ultrasonic microbubble imaging data acquired under different imaging conditions; For each pixel point in each group of sample ultrasonic microbubble imaging data, respectively calculating and extracting corresponding sample tissue characteristic parameter vectors and sample signal characteristic vectors, and integrating to form a sample fusion characteristic vector set; Judging whether the ultrasonic echo signal at each pixel point simultaneously meets the transient broadband characteristic and is incoherent with the adjacent pulse, if so, marking the ultrasonic echo signal as a real fracture noise label, otherwise, marking the ultrasonic echo signal as a pure signal label to form a sample supervision label set; Constructing a noise classification model based on the deep neural network; and performing supervised training on the noise classification model by adopting the sample fusion feature vector set and the sample supervision tag set until verification converges to obtain a trained noise classification model.
9. The noise suppression processing method for ultrasonic microbubble imaging according to claim 1, characterized in that based on coherence computation, real fracture noise is discriminated from the candidate fracture noise and filtered out, a clean echo signal is obtained, and an ultrasonic microbubble image is reconstructed and generated by using the clean echo signal, comprising: On the time sequence, based on the time information and the position information of the candidate fracture noise, carrying out spatial position registration on the ultrasonic echo signals of the current frame and the ultrasonic echo signals of the adjacent frames containing the candidate fracture noise through an image registration algorithm; calculating complex correlation coefficients between the current frame signal and adjacent frame signals by adopting a complex correlation coefficient formula at the space position corresponding to each candidate fracture noise, and taking the complex correlation coefficients as coherence coefficients; Judging the candidate cracking noise with the coherence coefficient lower than a preset coherence threshold as the real cracking noise; Filtering all signal components judged to be real cracking noise from the ultrasonic echo signal of the current frame to obtain a pure echo signal; and carrying out an imaging processing flow of beam synthesis, envelope detection and logarithmic compression on the clean echo signals, and reconstructing to generate an ultrasonic microbubble image.
10. A noise-suppression processing system for ultrasound microbubble imaging, for implementing a noise-suppression processing method for ultrasound microbubble imaging as recited in any one of claims 1-9, the system comprising: The data acquisition module is used for acquiring peak sparse pressure of the transmitted ultrasonic pulse and shell elasticity parameters of the ultrasonic microbubble contrast agent, and acquiring tissue depth information and fat content indexes of each pixel point in an imaging area; The index calculation module is used for calculating the local effective sound pressure of each pixel point after tissue attenuation according to the peak sparse pressure, the tissue depth information of each pixel point and the corresponding fat content index; the threshold value correction module is used for determining a basic rupture threshold value according to the shell elasticity parameter, correcting the basic rupture threshold value according to the fat content index of each pixel point, and obtaining a corrected rupture threshold value of each pixel point; The risk coefficient acquisition module is used for comparing the local effective sound pressure with the corrected fracture threshold value for each pixel point in the imaging area, calculating to obtain a microbubble fracture risk coefficient, integrating the microbubble fracture risk coefficients of all the pixel points, and generating a fracture risk distribution map; the fracture noise screening module is used for carrying out feature analysis and noise identification on the ultrasonic echo signals based on the fracture risk distribution diagram, the fat content index and the tissue depth information, and identifying candidate fracture noise; And the fracture noise filtering module is used for judging the real fracture noise from the candidate fracture noise based on coherence calculation and filtering the real fracture noise to obtain a pure echo signal, and reconstructing and generating an ultrasonic microbubble image by using the pure echo signal.

Description

Noise suppression processing method and system for ultrasonic microbubble imaging Technical Field The invention relates to the field of medical image processing, in particular to a noise suppression processing method and system for ultrasonic microbubble imaging. Background Ultrasonic microbubble imaging is an advanced functional imaging mode based on intravascular ultrasound contrast agents, and is achieved by intravenous injection of micron-sized microbubbles containing inert gas, which vibrate non-linearly in an ultrasound field, producing harmonic signals with intensities far higher than surrounding tissue. By receiving and selectively processing these harmonic signals, the linear echo of stationary tissue can be suppressed almost completely. The traditional method has the problems that the parameters are single, the individual difference is ignored, the cracking when the mechanical stability threshold value is exceeded can generate transient and broadband strong noise, the noise and nonlinear harmonic signals are seriously mixed, the image quality is interfered, the specific noise cannot be separated accurately, and the noise identification is inaccurate. In view of the foregoing, there is a strong need for a novel processing method and system that can deeply fuse individual tissue characteristics from the physical source of noise generation, and realize adaptive and precise recognition and suppression of ultrasonic microbubble rupture noise. Disclosure of Invention The application provides a noise suppression processing method and system for ultrasonic microbubble imaging, and aims to solve the problems that noise and nonlinear harmonic signals are seriously aliased, image quality is interfered, and specific noise cannot be accurately separated in the prior art. In view of the above, the present application provides a noise suppression processing method and system for ultrasonic microbubble imaging. In a first aspect, the present application provides a noise suppression processing method for ultrasonic microbubble imaging, including: Acquiring peak sparse pressure of emitted ultrasonic pulses and shell elasticity parameters of an ultrasonic microbubble contrast agent, and acquiring tissue depth information and fat content indexes of each pixel point in an imaging region; Calculating local effective sound pressure after tissue attenuation at each pixel point according to the peak sparse pressure, the tissue depth information of each pixel point and the corresponding fat content index; Determining a basic rupture threshold according to the shell elasticity parameter, and correcting the basic rupture threshold according to the fat content index of each pixel point to obtain a corrected rupture threshold of each pixel point; Comparing the local effective sound pressure with the corrected fracture threshold value for each pixel point in the imaging area, calculating to obtain a microbubble fracture risk coefficient, integrating the microbubble fracture risk coefficients of all the pixel points, and generating a fracture risk distribution map; based on the fracture risk distribution map, the fat content index and the tissue depth information, performing feature analysis and noise identification on an ultrasonic echo signal to identify candidate fracture noise; based on coherence calculation, judging real fracture noise from the candidate fracture noise, filtering to obtain a pure echo signal, and reconstructing and generating an ultrasonic microbubble image by using the pure echo signal. In a second aspect, the present application provides a noise suppression processing system for ultrasound microbubble imaging, comprising: The data acquisition module is used for acquiring peak sparse pressure of the transmitted ultrasonic pulse and shell elasticity parameters of the ultrasonic microbubble contrast agent, and acquiring tissue depth information and fat content indexes of each pixel point in an imaging area; The index calculation module is used for calculating the local effective sound pressure of each pixel point after tissue attenuation according to the peak sparse pressure, the tissue depth information of each pixel point and the corresponding fat content index; the threshold value correction module is used for determining a basic rupture threshold value according to the shell elasticity parameter, correcting the basic rupture threshold value according to the fat content index of each pixel point, and obtaining a corrected rupture threshold value of each pixel point; The risk coefficient acquisition module is used for comparing the local effective sound pressure with the corrected fracture threshold value for each pixel point in the imaging area, calculating to obtain a microbubble fracture risk coefficient, integrating the microbubble fracture risk coefficients of all the pixel points, and generating a fracture risk distribution map; the fracture noise screening module is used for carrying out feature ana