Endoscopic Ultrasound Ringing Artifact Removal Method Based on Time-Frequency Analysis Filtering

Abstract

The present invention discloses an adaptive single-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering. The method is as follows: Obtain the original data of endoscopic single-element ultrasound imaging, and perform time-frequency spectrum analysis on adjacent frame signals; Obtain the mutation situation of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal according to the optical flow method; Determine the threshold coefficient and the spectrum distribution situation factor, and determine the parameters of the time-frequency filter through these two parameters; Perform time-frequency filtering on the signal to be processed by the initially calculated time-frequency filter and perform time-domain recovery; Calculate the indexes using the data after removing the artifacts by time-frequency filtering and the original data, and optimize and adjust the threshold parameters according to whether the indexes meet the standards; Use the time-frequency filter parameters with qualified indexes to perform time-frequency filtering, time-domain reconstruction, and ultrasound endoscopic image reconstruction on the data. This method can almost completely remove the ringing artifacts, restore the effective echo signals, and maintain the structural integrity during near-distance imaging.

Description

Technical Field

[0001] The present invention belongs to the field of interventional ultrasonic imaging, and relates to a method for removing ringing noise in the field of interventional single-element ultrasonic imaging, and particularly relates to an adaptive single-element endoscopic ultrasonic ringing artifact removal method based on time-frequency analysis filtering. Background Art

[0002] At present, interventional single-element ultrasonic imaging is often interfered by various artifacts. Taking intravascular ultrasound (IVUS) as an example, ultrasonic ringing artifacts are relatively typical artifacts in interventional single-element ultrasonic imaging represented by IVUS.

[0003] Ringing artifacts are artifacts generated when the ultrasonic detection signal emitted by the ultrasonic transducer after being excited by the ultrasonic transceiver is received by the ultrasonic transducer itself. The appearance of the ringing artifact phenomenon will cause a series of dense concentric circular bright spots to remain in the central area of the ultrasonic image. This makes it impossible for the interventional single-element ultrasonic probe to effectively obtain the tissue structure information due to the covering effect of the ringing artifact when it is close to the tissue to be imaged, greatly affecting the overall imaging ability of the system.

[0004] Regarding the characteristics of ringing artifacts, there are already some research conclusions. Due to the zero drift generated by the ultrasonic transceiver during use and the electromagnetic interference of external devices, the ultrasonic detection signal will have a certain degree of difference. When the ultrasonic detection signal reaches the tissue to be imaged and generates an ultrasonic echo signal, a part of the original detection signal itself will enter the acquisition of the entire ultrasonic signal through the ultrasonic transducer, and this part of the ultrasonic detection signal after coupling is the ultrasonic ringing artifact signal.

[0005] At present, the methods for removing ringing artifacts in the world are relatively single and cannot well distinguish artifacts from normal echo signals. Due to the particularity of ringing artifacts and their relatively fixed appearance areas, most current methods remove ringing artifacts and improve the overall image contrast by circularly cropping a fixed area of the reconstructed interventional ultrasonic image. However, this method is based on image-end processing, ignores the size of the ringing artifact area, and cannot be adjusted accordingly for changes in external noise and transducer parameters. Moreover, when the area to be imaged is close to the transducer and the effective echo signal is submerged in the ringing artifact, the conventional cropping method will remove the effective imaging information close to the transducer as well. Another method is to perform frequency-domain filtering on the signal and then perform image reconstruction. Such a method can only suppress ringing artifacts to a certain extent and also has an inhibitory effect on ultrasonic echo signals.

[0006] The above-mentioned conventional processing methods may cause incorrect reconstruction of relatively narrow and catheter-proximal parts in interventional single-element ultrasonic endoscopy imaging such as IVUS, ultimately affecting the acquisition of the overall cavity wall structure information. Summary of the Invention

[0007] In order to solve the problem that there are ringing artifacts in intrusive single - element ultrasonic endoscopy imaging, which seriously affect the imaging effect and the existing methods for removing them are not ideal, the present invention provides a method for removing ringing artifacts in endoscopic ultrasonic imaging based on time - frequency analysis filtering. This method can almost completely remove the ringing artifacts, restore the effective echo signal, and maintain the structural integrity in near - distance imaging.

[0008] The object of the present invention is achieved by the following technical solutions:

[0009] A method for removing ringing artifacts in endoscopic ultrasonic imaging based on time - frequency analysis filtering, comprising the following steps:

[0010] Step 1: Obtain the original data of intrusive single - element ultrasonic imaging, perform time - frequency spectrum analysis on adjacent frame signals in the ultrasonic data with ultrasonic echo signals to obtain the time - frequency spectrum of the signals;

[0011] Step 2: Obtain the mutation situation of the time - frequency spectrum of the signal to be processed relative to the time - frequency spectrum of the reference signal according to the optical flow method;

[0012] Step 3: Determine the threshold coefficient and the factor of the spectrum distribution situation, and determine the parameters of the time - frequency filter through these two parameters;

[0013] Step 4: Perform time - frequency filtering on the signal to be processed through the initially calculated time - frequency filter and perform time - domain restoration;

[0014] Step 5: Calculate the indexes of the data after removing artifacts by time - frequency filtering and the original data, and optimize and adjust the threshold parameters according to whether the indexes meet the standards;

[0015] Step 6: Perform time - frequency filtering, time - domain reconstruction, and ultrasonic endoscopic image reconstruction on the data using the parameters of the time - frequency filter with qualified indexes to obtain an ultrasonic endoscopic image without ringing artifacts.

[0016] An adaptive single - element endoscopic ultrasonic ringing artifact removing device based on time - frequency analysis filtering, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the computer program to implement the above - mentioned adaptive single - element endoscopic ultrasonic ringing artifact removing method based on time - frequency analysis filtering.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] The present invention can accurately and effectively reduce the "ringing" artifact effect in intrusive single - element ultrasonic endoscopy imaging, eliminate the artifact coverage situation, improve the signal - to - noise ratio of proximal imaging, and greatly improve the overall structural imaging quality. Brief Description of the Drawings

[0019] Figure 1 It is the composition of the original signal of a single-array-element ultrasound;

[0020] Figure 2 It is the flow chart of the endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering;

[0021] Figure 3 It is the time-frequency spectrum image of the reference signal;

[0022] Figure 4 It is the time-frequency spectrum image of the signal to be processed;

[0023] Figure 5 It is the image of the changing part of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal;

[0024] Figure 6 It is the control matrix (image form) used for comparison with the changing part of the time-frequency spectrum;

[0025] Figure 7 It is the time-frequency filter image obtained by preliminary calculation;

[0026] Figure 8 It is the ultrasound echo signal containing the ultrasound ringing artifact and the ultrasound echo signal image after removing the ultrasound ringing artifact;

[0027] Figure 9 It is the image containing the ringing artifact reconstructed by the method of the present invention;

[0028] Figure 10 It is the image reconstructed by the method of the present invention after removing the ringing artifact. Detailed implementation manners

[0029] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings, but it is not limited thereto. Any modification or equivalent replacement of the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention shall be covered by the protection scope of the present invention.

[0030] Since in a single-array-element imaging system, usually only a "line"-shaped signal can be obtained in one "transmission-detection", it is thus called an A-Line. The present invention will use the A-Line signal to represent the signal obtained in one "transmission-detection".

[0031] Through research on the generation method of the existing "ringing" artifact, it is found that the generated "ringing" artifact is regular. The "ringing" artifact is generated because the transmitted detection ultrasonic wave is reversely coupled into the received signal. Its appearance position is relatively fixed in the whole signal, and the amplitude and shape will change to a certain extent due to hardware temperature and parameter fluctuations, but the adjacent spaced ringing artifacts have extremely strong similarity. The mixing process of the ultrasound echo signal and the ringing artifact signal is asFigure 1 As shown, the ultrasonic excitation signal US exciting (t) forms the ringing artifact signal k·C P (t) through the coupling effect of the ultrasonic transducer and the influence of temperature and environment on the ringing artifact signal, and is mixed with the ultrasonic echo signal S P (t) reflected by the tissue under the ultrasonic excitation signal to form the finally sampled ultrasonic signal S(t), where k is the coupling effect of the ultrasonic transducer and the temperature and environment proportionality coefficient.

[0032] The present invention provides a method for removing endoscopic ultrasonic ringing artifacts based on time-frequency analysis filtering. The method is based on time-frequency spectrum analysis. According to the strong correlation between the ringing artifacts at two adjacent A-line intervals, the relatively unchanged part is removed in the time-frequency spectrum, and the changing part is retained to remove the relatively fixed signal of the ringing artifacts. As Figure 2 shown, the specific steps are as follows:

[0033] Step 1: Obtain the original data of endoscopic single-element ultrasonic imaging, respectively collect the ultrasonic data with ultrasonic echo signals and the ultrasonic data containing only ultrasonic ringing artifacts, and perform time-frequency spectrum analysis on the signals at two adjacent A-line intervals in the ultrasonic data with ultrasonic echo signals. The specific steps are as follows:

[0034] Step 11: Collect a frame of B-scan ultrasonic data through an endoscopic ultrasonic imaging system. The ultrasonic data contains H groups of ultrasonic A-line signals. Each ultrasonic A-line signal is composed of ultrasonic ringing artifacts and ultrasonic echo signals. The ultrasonic data is converted from acoustic signals to electrical signals by an ultrasonic transducer and finally collected by a data acquisition card.

[0035] Step 12: Extract the signals at two adjacent A-line intervals from the H groups of ultrasonic A-line signals, and perform continuous wavelet transform or short-time Fourier transform on them to obtain the time-frequency spectra of the signals at two adjacent A-line intervals, which are S e (w m ,t n ) M×N and S r (w m ,t n ) M×N , as Figure 3 and Figure 4 shown, where w m represents the ordinate frequency of the time-frequency spectrum, and t n represents the abscissa sampling points of the time-frequency spectrum. The calculated time-frequency spectrum is a matrix of M×N, and M and N are the frequency resolution size and time resolution size of the time-frequency spectrum respectively.

[0036] Step 2: Obtain the mutation situation of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal according to the optical flow method. The specific steps are as follows:

[0037] Step 21: According to the time-frequency spectrum of the signal obtained in Step 1, select one frame of the time-frequency spectrum as the reference signal S e (w m ,t n ) M×N ;

[0038] Step 22: Calculate the relative change part of S r (w m ,t n ) M×N in the time-frequency spectrum according to the optical flow method. For the relative change part of S e (w m ,t n ) M×N , obtain the moving part D of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal. The calculated moving part of the time-frequency spectrum is as shown in M×N Figure 5 , and the calculation formula is D M×N =S e (w,t(w,t n ) M×N -S r (w,t n ) M×N .

[0039] Step 3: Determine the threshold coefficient and the factor of the spectrum distribution situation, and determine the parameters of the time-frequency filter through these two parameters. The specific steps are as follows:

[0040] Step 3.1: Determine the threshold coefficient eff to decide that the part where the relative change rate of the time-frequency spectrum is greater than the threshold coefficient is retained, and the part where the relative change rate is less than the threshold coefficient will be eliminated.

[0041] Step 3.2: At the same time, add the factor Γ(w m ) of the spectrum distribution situation of the ultrasonic ringing artifact:

[0042]

[0043] Γ(w m ) is the Fourier transform of the ringing artifact u0(t) itself, representing the distribution of the ringing artifact in frequency. By adding this factor, the purpose of increasing the probability of eliminating the part with a wider frequency distribution is achieved.

[0044] Step 3.3: Multiply the reference time-frequency spectrum S e (w m ,t n ) M×NMultiply by the threshold coefficient eff and add the factor Γ(w of the spectral distribution m ) to determine the final comparison matrix M com (m,n):

[0045] M com (m,n) = eff·S e (w m ,t n )·e -Γ(wm) .

[0046] The calculated comparison matrix is as Figure 6 shown.

[0047] Step 3.4: Compare the comparison matrix M com (m,n) with the moving part D of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal M×N and set the elements in D M×N that are larger than M com (m,n) to 1 and the smaller elements to 0 to obtain the preliminary time-frequency filter A(m,n):

[0048]

[0049] The image of the preliminarily calculated time-frequency filter is as Figure 7 shown.

[0050] Step 4: Perform time-frequency filtering on the signal to be processed using the preliminarily calculated time-frequency filter and perform time-domain recovery. The specific steps are as follows:

[0051] Step 41: Use the preliminarily obtained time-frequency filter A(m,n) to perform time-frequency filtering on the time-frequency spectrum of the signal to be processed to obtain the time-frequency spectrum S result (w,t n ):

[0052] S result (w m ,t n ) = A(m,n)·S e (w m ,t n )

[0053] Step 42: Perform an inverse time-frequency transform on the time-frequency spectrum S result (w,t n ) to restore it to a time-domain signal to obtain an ultrasonic echo signal without ultrasonic ringing artifacts. The restored time-domain signal is as Figure 8 shown.

[0054] Step 5: After using time-frequency filtering to filter out the artifacts, calculate the metrics for the filtered data and the original data, and optimize and adjust the threshold parameters according to whether the metrics meet the standards. The specific steps are as follows:

[0055] Step 51: Remove the ultrasonic ringing artifact signals from all ultrasonic data with ultrasonic echo signals to obtain an ultrasonic echo signal u clear num (t) that is the num-th among all A-line signals, and superimpose and mix it with the ultrasonic data containing only the ringing artifacts to simulate the actual ultrasonic echo signal u mix num (t);

[0056] Step 52: Repeat the ultrasonic ringing artifact removal method to remove the ultrasonic ringing artifacts from the simulated actual ultrasonic echo signal u mix num (t) to obtain the ultrasonic echo signal u cleaned num (t) after artifact removal. Calculate the correlation coefficient between this signal and the ultrasonic echo signal u clear num (t) without ringing artifacts. Repeat the filtering process and optimize the correlation coefficient between the restored ultrasonic echo signal and the original ultrasonic echo signal by adjusting the value of the threshold coefficient eff.

[0057] Step 6: Use the time-frequency filter parameters that meet the metrics to perform time-frequency filtering on the data, reconstruct in the time domain, and perform ultrasonic endoscopy image reconstruction and compare the artifact removal effects Figure 9 is the image containing ringing artifacts reconstructed by the method of the present invention, Figure 10 is the image with ringing artifacts removed reconstructed by the method of the present invention. From Figure 9 and Figure 10 , it can be seen that after removing the ringing artifacts, the image in the area near the center of the image can be restored and is no longer affected by the ultrasonic ringing artifacts.

[0058] The present invention also provides an adaptive single-array endoscopic ultrasonic ringing artifact removal device based on time-frequency analysis filtering. The device includes a memory and a processor, where the memory stores a computer program, and the processor is configured to execute the computer program to implement the above-mentioned adaptive single-array endoscopic ultrasonic ringing artifact removal method based on time-frequency analysis filtering.

Claims

1. An adaptive single-array-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering, characterized in that The method includes the following steps: Step 1: Obtain the original data of endoscopic single-element ultrasonic imaging, perform time-frequency spectrum analysis on adjacent frame signals in the ultrasonic data with ultrasonic echo signals to obtain the time-frequency spectrum of the signals; Step 2: Obtain the mutation situation of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal according to the optical flow method; Step 3: Determine the threshold coefficient and the factor of the spectrum distribution situation, and determine the parameters of the time-frequency filter through these two parameters; Step 4: Perform time-frequency filtering on the signal to be processed by the initially calculated time-frequency filter and perform time-domain restoration; Step 5: Calculate the indexes of the data after removing artifacts by time-frequency filtering and the original data, and optimize and adjust the threshold parameters according to whether the indexes meet the standards; Step 6: Perform time-frequency filtering, time-domain reconstruction, and ultrasonic endoscopic image reconstruction on the data with the time-frequency filter parameters that meet the indexes to obtain an ultrasonic endoscopic image without ringing artifacts.

2. The adaptive single-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering according to claim 1, wherein The specific steps of Step 1 are as follows: Step 11: Collect a frame of ultrasonic B-scan data through an ultrasonic endoscopic imaging system. The ultrasonic data contains H groups of ultrasonic A-line signals, and each ultrasonic A-line signal is composed of ultrasonic ringing artifacts and ultrasonic echo signals; Step 12: Extract the signals at two adjacent A-line intervals from the H-group ultrasound A-line signals, and perform continuous wavelet transform or short-time Fourier transform on them to obtain the time-frequency spectra of the signals at two adjacent A-line intervals, which are S e (w m ,t n ) M×N and S r (w m ,t n ) M×N , where w m represents the ordinate frequency of the time-frequency spectrum, and t n represents the abscissa sampling points of the time-frequency spectrum. The calculated time-frequency spectrum is a matrix of M×N, where M and N are the frequency resolution and time resolution of the time-frequency spectrum, respectively.

3. The adaptive single-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering according to claim 1, characterized in that The specific steps of Step 2 are as follows: Step 21: Select one frame of the time-frequency spectrum as the reference signal S according to the time-frequency spectrum of the signal obtained in Step 1 e (w m ,t n ) M×N ; Step 22: Calculate the S in the time-frequency spectrum according to the optical flow method r (w m ,t n ) M×N For S e (w m ,t n ) M×N For the relatively changing part, obtain the moving part D of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal M×N .

4. The adaptive single-array-element endoscopic ultrasonic ringing artifact removal method based on time-frequency analysis filtering according to claim 3, characterized in that The said D M×N The calculation formula is as follows: D M×N = S e (w, t n ) M×N - S r (w, t n ) M×N .

5. The adaptive single-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering according to claim 1, wherein The specific steps of Step 3 are as follows: Step 3.1: Determine the threshold coefficient eff to decide that the part where the relative change rate of the time-frequency spectrum is greater than the threshold coefficient is retained, and the part where the relative change rate is less than the threshold coefficient will be removed; Step 3.2: Add the factor Γ(w m ) of the spectral distribution of the ultrasonic ringing artifact at the same time u0(t) is the ringing artifact; Step 3.3: Multiply the reference time-frequency spectrum S e (w m ,t n ) M×N by the threshold coefficient eff, and at the same time add the spectrum distribution factor Γ(w m ) to determine the final comparison matrix M com (m,n): Step 3.4: Compare the comparison matrix M com (m,n) with the moving part D of the time-frequency spectrum of the signal to be processed relative to the time-frequency spectrum of the reference signal M×N and set the elements in D M×N that are larger than M com (m,n) to 1 and the smaller elements to 0 to obtain the preliminary time-frequency filter A(m,n):

6. The adaptive single-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering according to claim 1, wherein The specific steps of Step 4 are as follows: Step 41: Use the initially obtained time-frequency filter A(m,n) to perform time-frequency filtering on the time-frequency spectrum of the signal to be processed, and obtain the time-frequency spectrum S after time-frequency filtering result (w,t n ): S result (w m ,t n ) = A(m,n)·S e (w m ,t n ) Step 42: Perform time-frequency inverse transformation on the time-frequency spectrum S result (w, t n ) to restore it to a time-domain signal and obtain an ultrasonic echo signal without ultrasonic ringing artifacts.

7. The adaptive single-array-element endoscopic ultrasound ringing artifact removal method based on time-frequency analysis filtering according to claim 1, wherein The specific steps of Step 5 are as follows: Step 51: Remove the ultrasonic ringing artifact signals from all ultrasonic data with ultrasonic echo signals to obtain an ultrasonic echo signal u without ringing artifacts clear num (t), which is the num-th among all A-line signals, and is superimposed and mixed with the ultrasonic data containing only ringing artifacts to simulate the actual ultrasonic echo signal u mix num (t); Step 52: Repeat the ultrasonic ringing artifact removal method to remove the ultrasonic ringing artifacts from the simulated actual ultrasonic echo signal u mix num (t), and obtain the ultrasonic echo signal u cleaned num (t) after removing the artifacts. Calculate the correlation coefficient between this signal and the ultrasonic echo signal u clear num (t) without ringing artifacts. Repeat the filtering process and optimize the correlation coefficient between the restored ultrasonic echo signal and the original ultrasonic echo signal by adjusting the value of the threshold coefficient eff.

8. An adaptive single-element endoscopic ultrasound ringing artifact removal device based on time-frequency analysis filtering, characterized in that The device includes a memory and a processor, wherein: The memory stores a computer program; The processor is used to execute the computer program to implement the adaptive single-element endoscopic ultrasonic ringing artifact removal method based on time-frequency analysis filtering according to any one of claims 1-7.

Images