An ultrasonic endoscope image processing method and apparatus

Abstract

The application provides an endoscopic ultrasound image processing method and device, applied to the technical field of image processing, through obtaining an ultrasonic echo signal and performing frequency domain analysis, low-frequency energy information is extracted for artifact detection and gain adjustment. By pre-setting a low-frequency energy baseline value, then, calculating an artifact detection dynamic threshold according to the low-frequency energy baseline value and the energy fluctuation range, calculating a low-frequency gain suppression factor according to the low-frequency energy change rate, the low-frequency energy change rate reflects the speed of the appearance of artifacts, which is used for dynamically adjusting the gain suppression intensity. Finally, the real-time low-frequency gain coefficient is applied to the ultrasonic echo signal in the frequency domain for gain adjustment, and the adjusted frequency domain signal is converted into a time domain signal to reconstruct and generate an endoscopic ultrasound image, thereby improving the quality of the endoscopic ultrasound image.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to an endoscopic ultrasound image processing method and device. BACKGROUND

[0002] In minimally invasive endoscopic ultrasound surgery, endoscopic doctors use an endoscope to diagnose and treat in the body cavity. Doctors guide endoscopic operation by observing ultrasound images and identify lesions. However, when the endoscope passes through a region rich in blood vessels or high-density tissue, the quality of the ultrasound image is prone to decrease. Bleeding produces artifacts, and tissue attenuation weakens ultrasound signals, and these factors superimposed make the effect of traditional image processing methods limited to improve image quality in complex situations. The decline in image quality makes it difficult for doctors to accurately determine the location and nature of the lesion, directly affecting the efficiency of the operation and the safety of the patient. Therefore, under complex physiological conditions, how to provide high-quality ultrasound images to meet the real-time diagnosis needs is crucial. Existing image processing techniques, such as fixed gain compensation or linear filtering, are difficult to effectively cope with dynamic image degradation caused by complex physiological environments, especially when bleeding artifacts and tissue attenuation occur at the same time, the image quality improvement effect is not good, and it cannot fully meet the dual needs of real-time and image quality for minimally invasive surgery.

[0003] In minimally invasive endoscopic ultrasound surgery, real-time degradation of ultrasound image quality caused by superimposed bleeding artifacts and tissue attenuation in complex physiological environments is a key problem that restricts diagnostic accuracy and surgical efficiency. Existing image processing techniques are difficult to effectively cope with dynamic changes in physiological environments and cannot provide continuous high-quality image support. Therefore, how to design an ultrasound image real-time processing method that can adapt to the dynamic changes of complex physiological environments and effectively suppress artifacts and compensate for attenuation without increasing the burden of additional equipment and the complexity of operation has become a problem to be solved. SUMMARY

[0004] In view of the deficiencies of the prior art described above, the present application provides an endoscopic ultrasound image processing method and device, which is applied to the technical field of image processing and has the advantage of improving the quality of endoscopic ultrasound images.

[0005] In a first aspect, an endoscopic ultrasound image processing method is used to improve the quality of ultrasound images in minimally invasive endoscopic ultrasound surgery, the method comprising the steps of:

[0006] S1: Acquire ultrasound echo signals, perform frequency domain analysis, and extract low-frequency band energy information, the low-frequency band energy information including at least: low-frequency band energy value, low-frequency band energy fluctuation range, and low-frequency band energy change rate;

[0007] S2: Obtain a pre-set low-frequency band energy baseline value;

[0008] S3: calculating an artifact detection dynamic threshold according to the low-frequency energy baseline value and the low-frequency energy fluctuation range;

[0009] S4: calculating a low-frequency gain suppression factor according to the low-frequency energy change rate;

[0010] S5: calculating a real-time low-frequency gain coefficient according to the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value;

[0011] S6: applying the real-time low-frequency gain coefficient to the frequency domain ultrasonic echo signal for gain adjustment, converting the gain-adjusted frequency domain signal into a time domain signal, performing image reconstruction, and generating an ultrasonic endoscope image.

[0012] The ultrasonic endoscope processing method provided in the present application extracts low-frequency energy information including energy value, energy fluctuation range, and energy change rate by acquiring ultrasonic echo signals and performing frequency domain analysis. These information are used for subsequent artifact detection and gain adjustment. The present scheme pre-sets a low-frequency energy baseline value as an energy reference under normal circumstances. Then, an artifact detection dynamic threshold is calculated according to the low-frequency energy baseline value and the energy fluctuation range. This artifact detection dynamic threshold can adapt to signal fluctuations and more accurately detect artifacts. The scheme also calculates a low-frequency gain suppression factor according to the low-frequency energy change rate. The low-frequency energy change rate reflects the speed of artifact appearance and is used to dynamically adjust the gain suppression strength. The scheme calculates a real-time low-frequency gain coefficient by comprehensively considering the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the current low-frequency energy value, and realizes dynamic adjustment of the gain. Finally, the scheme applies the real-time low-frequency gain coefficient to the frequency domain ultrasonic echo signal for gain adjustment, converts the adjusted frequency domain signal into a time domain signal, and reconstructs to generate an ultrasonic endoscope image, thereby ultimately improving the quality of the ultrasonic endoscope image. Therefore, by analyzing and utilizing the low-frequency energy information of the ultrasonic echo signal, the present scheme realizes dynamic gain adjustment of the ultrasonic image, thereby effectively improving the quality of the ultrasonic image in minimally invasive ultrasonic endoscopic surgery.

[0013] Further, step S3 includes:

[0014] S31: acquiring a coupling degree parameter between the ultrasonic endoscope probe and the lesion area, the coupling degree parameter at least including ultrasonic echo signal strength;

[0015] S32: calculating a coupling degree confidence factor according to the ultrasonic echo signal strength;

[0016] S33: calculating the artifact detection dynamic threshold according to the coupling degree confidence factor, the low-frequency energy baseline value, and the low-frequency energy fluctuation range.

[0017] The application provides an endoscopic ultrasound processing method, a coupling degree confidence factor is calculated according to an ultrasonic echo signal intensity, so as to quantize a coupling degree parameter into the coupling degree confidence factor, and the coupling degree confidence factor is introduced into calculation of a dynamic threshold value, so that the dynamic threshold value depends not only on low-frequency energy information, but also on a probe coupling state, the accuracy and reliability of the artifact detection dynamic threshold value are improved, and the performance of the endoscopic ultrasound image processing method is improved.

[0018] Further, in step S33, a formula for calculating the artifact detection dynamic threshold value according to the coupling degree confidence factor, the low-frequency energy baseline value and the low-frequency energy fluctuation range is: DT = BLV * (1 + α * FR / BLV) * exp(-β * (1 - CCF)); wherein, DT is the artifact detection dynamic threshold value, BLV is the low-frequency energy baseline value, FR is the low-frequency energy fluctuation range, and CCF is the coupling degree confidence factor.

[0019] α is an influence factor of the low-frequency energy fluctuation range, representing a positive influence degree of the low-frequency energy fluctuation range on the artifact detection dynamic threshold value.

[0020] β is a nonlinear influence degree of the coupling degree confidence factor, representing a nonlinear negative influence degree of the coupling degree confidence factor on the artifact detection dynamic threshold value.

[0021] The endoscopic ultrasound processing method provided by the application specifically realizes the calculation method of the artifact detection dynamic threshold value through the formula DT = BLV * (1 + α * FR / BLV) * exp(-β * (1 - CCF)), and the dynamic adjustment strategy enables the artifact detection dynamic threshold value to be adaptively adjusted according to the change of the coupling degree, thereby improving the accuracy and sensitivity of artifact detection.

[0022] Further, step S4 includes:

[0023] S41: performing normalization processing on the low-frequency energy change rate to obtain a normalized low-frequency energy change rate.

[0024] S42: calculating the low-frequency gain suppression factor through a nonlinear mapping function according to the normalized low-frequency energy change rate; wherein, the nonlinear mapping function is a Sigmoid function, and the calculation formula is:

[0025] Lf_inhibition = 1 / (1 + exp(-k * Normalized_rate)), wherein, Lf_inhibition is the low frequency gain inhibition factor, k is a preset Sigmoid function steepness adjustment parameter; Normalized_rate is the normalized low frequency energy change rate.

[0026] The ultrasound endoscope processing method provided in the present application can obtain a more reasonable low frequency gain inhibition factor through normalization processing and Sigmoid nonlinear mapping, thereby being better used in the subsequent gain adjustment step to improve image quality.

[0027] Further, the step S5 comprises:

[0028] S51: judging whether the low frequency energy value exceeds the artifact detection dynamic threshold value, and if yes, determining that there is a bleeding artifact;

[0029] S52: if the low frequency energy value does not exceed the artifact detection dynamic threshold value, obtaining a preset low frequency reference gain, and taking the low frequency reference gain as the real-time low frequency gain coefficient;

[0030] 53: if the low frequency energy value exceeds the artifact detection dynamic threshold value, calculating a real-time low frequency gain coefficient according to the artifact detection dynamic threshold value, the low frequency gain inhibition factor and the low frequency energy value.

[0031] Further, in the step S53, the formula for calculating the real-time low frequency gain coefficient according to the artifact detection dynamic threshold value, the low frequency gain inhibition factor and the low frequency energy value is:

[0032] Gain_lf = Base_gain - Lf_inhibition * (Current_lf - DT), wherein, Gain_lf is the real-time low frequency gain coefficient, Base_gain is a preset low frequency reference gain, Lf_inhibition is the low frequency gain inhibition factor, Current_lf is the low frequency energy value, and DT is the artifact detection dynamic threshold value.

[0033] Further, the step S6 comprises:

[0034] S61: extracting high frequency band energy information according to the ultrasound echo signal, wherein the high frequency band energy information at least comprises a high frequency band energy value, a high frequency band energy fluctuation range and a high frequency band energy change rate;

[0035] S62: obtaining a preset high frequency band energy baseline value;

[0036] S63: calculating a decay detection dynamic threshold according to the high frequency band energy baseline value and the high frequency band energy fluctuation range;

[0037] S64: calculating a high frequency gain amplification factor according to the high frequency energy change rate;

[0038] S65: calculating a real-time high frequency gain coefficient according to the decay detection dynamic threshold, the high frequency gain amplification factor and the high frequency band energy value;

[0039] S66: applying the real-time low frequency gain coefficient and the real-time high frequency gain coefficient to the frequency domain ultrasonic echo signal to perform gain adjustment, converting the gain-adjusted frequency domain signal into a time domain signal, performing image reconstruction to generate an ultrasonic endoscope image.

[0040] Further, step S66 comprises:

[0041] S661: multiplying the real-time low frequency gain coefficient with the low frequency component of the ultrasonic echo signal to realize gain adjustment of the low frequency component; multiplying the real-time high frequency gain coefficient with the high frequency component of the ultrasonic echo signal to realize gain adjustment of the high frequency component;

[0042] S662: performing Gaussian filtering on the high frequency component and the low frequency component to smooth noise to obtain the gain-adjusted frequency domain signal;

[0043] S663: performing inverse Fourier transform on the gain-adjusted frequency domain signal to convert the frequency domain signal into a time domain signal;

[0044] S664: performing envelope detection on the time domain signal, and performing logarithmic compression on the envelope-detected time domain signal to enhance image contrast;

[0045] S665: performing scan conversion on the logarithmically compressed time domain signal to convert the polar coordinate form of the ultrasonic echo signal into a rectangular coordinate form, performing image interpolation to generate an ultrasonic endoscope image.

[0046] Further, step S662 comprises:

[0047] S6621: respectively decomposing the high frequency component and the low frequency component into real part signals and imaginary part signals;

[0048] S6622: respectively performing Gaussian filtering on the real part signals and the imaginary part signals to obtain filtered real part signals and filtered imaginary part signals, wherein the standard deviation of the Gaussian filter is adaptively adjusted according to a preset high frequency noise level, and the size of the Gaussian filter matches the resolution of the ultrasonic endoscope image;

[0049] S6623: combine the filtered real part signal and the filtered imaginary part signal into the filtered high frequency component and the filtered low frequency component, and combine the filtered high frequency component and the filtered low frequency component as the gain-adjusted frequency domain signal.

[0050] In a second aspect, an ultrasound endoscope image processing device is applied in the steps of any one of the above-mentioned ultrasound endoscope image processing methods, and the device comprises:

[0051] A first obtaining module is configured to obtain an ultrasound echo signal, perform frequency domain analysis, and extract low frequency band energy information, wherein the low frequency band energy information at least includes a low frequency band energy value, a low frequency band energy fluctuation range, and a low frequency band energy change rate.

[0052] A second obtaining module is configured to obtain a pre-set low frequency band energy baseline value.

[0053] A dynamic threshold adjusting module is configured to calculate an artifact detection dynamic threshold according to the low frequency band energy baseline value and the low frequency band energy fluctuation range.

[0054] A dynamic gain adjusting module is configured to calculate a low frequency gain suppression factor according to the low frequency band energy change rate.

[0055] A gain coefficient adjusting module is configured to calculate a real-time low frequency gain coefficient according to the artifact detection dynamic threshold, the low frequency gain suppression factor, and the low frequency energy value.

[0056] An image reconstruction module is configured to apply the real-time low frequency gain coefficient to the frequency domain ultrasound echo signal, perform gain adjustment, convert the gain-adjusted frequency domain signal into a time domain signal, perform image reconstruction, and generate an ultrasound endoscope image.

[0057] Beneficial effects: The ultrasonic endoscope image processing method and device provided by the application extracts low-frequency energy information including energy value, energy fluctuation range and energy change rate through obtaining ultrasonic echo signals and performing frequency domain analysis. These information are used for subsequent artifact detection and gain adjustment. The scheme pre-sets a low-frequency energy baseline value as the energy reference under normal circumstances. Then, the artifact detection dynamic threshold is calculated according to the low-frequency energy baseline value and the energy fluctuation range. The artifact detection dynamic threshold can adapt to signal fluctuation and more accurately detect artifacts. The scheme also calculates a low-frequency gain suppression factor according to the low-frequency energy change rate. The low-frequency energy change rate reflects the speed of artifact appearance and is used for dynamically adjusting the gain suppression intensity. The scheme combines the artifact detection dynamic threshold, the low-frequency gain suppression factor and the current low-frequency energy value to calculate a real-time low-frequency gain coefficient, and realizes dynamic adjustment of the gain. Finally, the scheme applies the real-time low-frequency gain coefficient to the frequency domain ultrasonic echo signals for gain adjustment, converts the adjusted frequency domain signals into time domain signals, reconstructs to generate ultrasonic endoscope images, and finally achieves the effect of improving the quality of ultrasonic endoscope images. Therefore, by analyzing and utilizing the low-frequency energy information of ultrasonic echo signals, the scheme realizes dynamic gain adjustment of ultrasonic images, thereby effectively improving the quality of ultrasonic images in minimally invasive ultrasonic endoscopic surgery. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 A flowchart of an ultrasonic endoscope image processing method provided by the application.

[0059] Figure 2 A structural diagram of an ultrasonic endoscope image processing device provided by the application.

[0060] Label explanation: 201, image acquisition module; 202, point cloud construction module; 203, blood vessel segmentation module; 204, parameter calculation module; 205, blood vessel grading module. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. The components of the embodiments of the application described and indicated in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the application provided in the drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the application. Based on the embodiments of the application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the application.

[0062] It should be noted that similar reference numerals and letters refer to like items in the accompanying drawings, and that, once an item is defined in one drawing, it should not require further defining or explaining in subsequent drawings. Also, in the description of the present application, the terms "first", "second", and so on are used only to distinguish descriptions, and should not be understood as indicating or implying relative importance.

[0063] In the complex physiological environment of minimally invasive endoscopic surgery, the real-time degradation of ultrasound image quality caused by superimposed bleeding artifacts and tissue attenuation is a key problem that restricts the accuracy of diagnosis and the efficiency of surgery. Existing image processing techniques are difficult to effectively respond to the dynamic changes of the physiological environment, and cannot provide continuous high-quality image support. Therefore, the present application proposes an endoscopic ultrasound image processing method and device, as follows:

[0064] Please refer to Figure 1 , in a first aspect, an endoscopic ultrasound image processing method is used to improve the quality of ultrasound images in minimally invasive endoscopic surgery, the method comprising the steps of:

[0065] S1: Acquire the ultrasound echo signal, perform frequency domain analysis, and extract the low-frequency band energy information, which at least includes: low-frequency band energy value, low-frequency band energy fluctuation range, and low-frequency band energy change rate;

[0066] S2: Obtain a pre-set low-frequency band energy baseline value;

[0067] S3: Calculate the artifact detection dynamic threshold according to the low-frequency band energy baseline value and the low-frequency band energy fluctuation range;

[0068] S4: Calculate the low-frequency gain suppression factor according to the low-frequency band energy change rate;

[0069] S5: Calculate the real-time low-frequency gain coefficient according to the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value;

[0070] S6: Apply the real-time low-frequency gain coefficient to the frequency domain ultrasound echo signal, perform gain adjustment, convert the gain-adjusted frequency domain signal into a time domain signal, perform image reconstruction, and generate an endoscopic ultrasound image.

[0071] In step S1, after the ultrasound echo signal is acquired, frequency domain analysis is performed, for example, by fast Fourier transform. On the basis of frequency domain analysis, low-frequency band energy information is extracted, which at least includes low-frequency band energy value, low-frequency band energy fluctuation range, and low-frequency band energy change rate. The low-frequency band energy value can reflect the intensity of the current low-frequency signal. The low-frequency band energy fluctuation range can reflect the change amplitude of the low-frequency signal within a period of time. The low-frequency band energy change rate can reflect the speed of the low-frequency signal change.

[0072] In step S2, a pre-set low-frequency energy baseline value is obtained, which is a low-frequency energy reference value under normal tissue or no artifact condition, and can be obtained by experiment or experience in advance.

[0073] In step S3, the calculation of the artifact detection dynamic threshold considers the volatility of the low-frequency energy signal, so that the artifact detection dynamic threshold can adapt to the change of the low-frequency energy signal, thereby more accurately detecting artifacts. For example, the artifact detection dynamic threshold can be calculated by the formula DT = BLV * (1 + a * FR / BLV), where DT is the artifact detection dynamic threshold, BLV is the low-frequency energy baseline value, FR is the low-frequency energy fluctuation range, and a is an influence factor.

[0074] In step S4, the low-frequency gain inhibition factor is used to dynamically adjust the gain inhibition strength. The greater the energy change rate, the higher the possibility of artifact occurrence, and the low-frequency gain inhibition factor should also be increased accordingly. For example, the low-frequency gain inhibition factor can be calculated by the Sigmoid function Lf_inhibition = 1 / (1 + exp(-k * Normalized_rate)), where Lf_inhibition is the low-frequency gain inhibition factor, k is the steepness adjustment parameter, and Normalized_rate is the normalized low-frequency energy change rate.

[0075] In step S5, if the current low-frequency energy value exceeds the artifact detection dynamic threshold, it is determined that there may be a bleeding artifact, and at this time the gain coefficient needs to be dynamically adjusted according to the low-frequency gain inhibition factor and the low-frequency energy value to suppress the artifact. If the low-frequency energy value does not exceed the threshold, it is determined that there is no bleeding artifact, and the pre-set low-frequency reference gain can be used as the real-time low-frequency gain coefficient. For example, the real-time low-frequency gain coefficient can be calculated by the formula Gain_lf = Base_gain - Lf_inhibition * (Current_lf - DT), where Gain_lf is the real-time low-frequency gain coefficient, Base_gain is the low-frequency reference gain, Lf_inhibition is the low-frequency gain inhibition factor, Current_lf is the low-frequency energy value, and DT is the artifact detection dynamic threshold.

[0076] In step S6, the real-time low-frequency gain coefficient is applied to the frequency domain ultrasonic echo signal for gain adjustment. The gain-adjusted frequency domain signal is converted into a time domain signal, for example, by inverse Fourier transform. Finally, the time domain signal is used for image reconstruction to generate an ultrasonic endoscopic image. The image reconstruction process can include envelope detection, logarithmic compression, scan conversion and other steps.

[0077] Specifically, the scheme aims to improve the quality of ultrasound images in minimally invasive endoscopic surgery. To address the problem of image quality degradation in complex physiological environments, a dynamic gain adjustment method based on low-frequency energy information is proposed. First, the ultrasound echo signal is collected and analyzed in the frequency domain, and the low-frequency energy information is extracted to provide data support for subsequent artifact detection and gain adjustment. The pre-set low-frequency energy baseline value is used as the energy reference under normal conditions for dynamic threshold calculation. The artifact detection dynamic threshold is calculated based on the low-frequency energy baseline value and the energy fluctuation range, which realizes the adaptive adjustment of the artifact detection threshold and improves the accuracy of artifact detection. The low-frequency gain suppression factor is calculated based on the low-frequency energy change rate, which realizes the dynamic adjustment of the gain suppression intensity and can adaptively adjust the gain suppression intensity according to the speed of artifact appearance. The real-time low-frequency gain coefficient considers the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the current low-frequency energy value, which realizes the real-time and dynamic adjustment of the gain coefficient and can dynamically reduce the gain to suppress the impact of artifacts on image quality when artifacts are detected. Finally, the real-time low-frequency gain coefficient is applied to the frequency domain ultrasound echo signal for gain adjustment, and the adjusted frequency domain signal is converted to the time domain signal to reconstruct and generate endoscopic ultrasound images, effectively improving the quality of ultrasound images in minimally invasive endoscopic surgery. Through the above steps, the scheme can dynamically adjust the gain of ultrasound images according to the low-frequency energy information of the ultrasound echo signal, effectively suppress the bleeding artifacts, and improve the image quality to provide clearer ultrasound images for doctors and assist doctors in more accurate diagnosis and treatment.

[0078] In some embodiments, the frequency domain analysis is realized by short-time Fourier transform, and the low-frequency band is set to 0MHz-2MHz band. The low-frequency energy value is obtained by calculating the sum of the square of the spectrum amplitude in the frequency band. The low-frequency energy fluctuation range is obtained by calculating the difference between the maximum and minimum values of the low-frequency energy value through a sliding window, and the length of the sliding window is set to 5 frames. The low-frequency energy change rate is obtained by calculating the difference between the current frame and the previous frame low-frequency energy value.

[0079] The low-frequency energy baseline value is pre-set to 500 units of energy value.

[0080] The artifact detection dynamic threshold is calculated by the formula DT = BLV * (1 + 0.5 * FR / BLV) * exp(-2 * (1- CCF)), where the coupling degree confidence factor CCF can be simplified to 1.

[0081] The normalized low-frequency energy change rate is obtained by dividing the low-frequency energy change rate by the maximum possible energy change range, and the steepness adjustment parameter k can be pre-set to 5.

[0082] The low-frequency reference gain can be preset to 0dB.

[0083] The gain-adjusted frequency domain signal is converted into a time domain signal using inverse short-time Fourier transform. The image reconstruction process employs a commonly used B-mode ultrasound imaging reconstruction method. Through the specific parameters and methods described above, this scheme was implemented and its effectiveness in improving the quality of endoscopic ultrasound images was verified.

[0084] Furthermore, step S3 includes:

[0085] S31: Obtain the coupling parameters between the endoscopic ultrasound probe and the lesion area. The coupling parameters include at least the ultrasound echo signal intensity.

[0086] S32: Calculate the coupling confidence factor based on the ultrasonic echo signal intensity;

[0087] S33: Calculate the dynamic threshold for artifact detection based on the coupling confidence factor, the low-frequency energy baseline value, and the low-frequency energy fluctuation range.

[0088] In step S31, the coupling degree parameter represents the signal transmission efficiency between the endoscopic ultrasound probe and the lesion area. The coupling degree parameter is obtained through a sensor configured to measure the physical contact characteristics between the probe and the tissue. The ultrasound echo signal intensity reflects the effective transfer of ultrasound energy from the probe to the tissue.

[0089] In step S32, the calculation of the coupling confidence factor is a process of quantifying the degree of coupling. The ultrasonic echo signal intensity is normalized or mapped to a range of 0 to 1 to generate the coupling confidence factor. The higher the signal intensity, the closer the confidence factor is to 1, indicating a better coupling. The lower the signal intensity, the closer the confidence factor is to 0, indicating a worse coupling.

[0090] In step S33, the dynamic threshold for artifact detection is calculated using the formula: DT = BLV * (1 + α * FR / BLV) * exp(-β * (1 - CCF)). The formula introduces the coupling confidence factor CCF to adjust the dynamic threshold DT. When the coupling confidence factor is high, the exp(-β * (1 - CCF)) term is close to 1, and the dynamic threshold for artifact detection is mainly determined by the low-frequency energy baseline value BLV and the low-frequency energy fluctuation range FR. When the coupling confidence factor is low, the exp(-β * (1 - CCF)) term decreases, and the dynamic threshold DT decreases accordingly. Therefore, the dynamic threshold for artifact detection can be adaptively adjusted according to the coupling degree.

[0091] Further, in step S33, according to the coupling confidence factor, the low-frequency energy baseline value, and the low-frequency energy fluctuation range, a formula for calculating the artifact detection dynamic threshold value is: DT = BLV * (1 + a * FR / BLV) * exp(-b * (1 - CCF)); wherein, DT is the artifact detection dynamic threshold value, BLV is the low-frequency energy baseline value, FR is the low-frequency energy fluctuation range, and CCF is the coupling confidence factor;

[0092] a is an influence factor of the low-frequency energy fluctuation range, representing the positive influence degree of the low-frequency energy fluctuation range on the artifact detection dynamic threshold value;

[0093] b is a non-linear influence degree of the coupling confidence factor, representing the non-linear negative influence degree of the coupling confidence factor on the artifact detection dynamic threshold value.

[0094] In some embodiments, the low-frequency energy baseline value BLV can be preset as 100, the low-frequency energy fluctuation range FR can be preset as 20, the influence factor a can be set as 0.5, and the non-linear influence degree factor b can be set as 2. When the coupling confidence factor CCF is 0.9, DT = 100 * (1 + 0.5 * 20 / 100) * exp(-2 * (1 - 0.9)) ≈ 110 * exp(-0.2) ≈ 90.2 is obtained by substituting the formula. When the coupling confidence factor CCF is reduced to 0.5, DT = 100 * (1 + 0.5 * 20 / 100) * exp(-2 * (1 - 0.5)) ≈ 110 * exp(-1) ≈ 40.4. As can be seen, as the coupling confidence factor CCF decreases, the artifact detection dynamic threshold value DT also decreases significantly. This adjustment of the artifact detection dynamic threshold value can adapt to the artifact detection requirements under different coupling conditions, and can still maintain high artifact detection sensitivity when the coupling degree is poor.

[0095] Further, step S4 comprises:

[0096] S41: performing normalization processing on the low-frequency energy change rate to obtain a normalized low-frequency energy change rate;

[0097] S42: calculating a low-frequency gain suppression factor according to the normalized low-frequency energy change rate through a non-linear mapping function; wherein, the non-linear mapping function is a Sigmoid function, and the calculation formula is:

[0098] Lf_inhibition = 1 / (1 + exp(-k * Normalized_rate)), wherein, Lf_inhibition is a low-frequency gain inhibition factor, k is a preset Sigmoid function steepness adjustment parameter; Normalized_rate is a normalized low-frequency energy change rate.

[0099] In step S41, the low-frequency energy change rate is normalized, and a linear transformation method can be used to map the low-frequency energy change rate to the interval [0, 1]. For example, assuming that the original value range of the low-frequency energy change rate is [Rate_min, Rate_max], the calculation formula of the normalized low-frequency energy change rate can be: Normalized_rate = (Current_rate - Rate_min) / (Rate_max - Rate_min), wherein, Current_rate is the current low-frequency energy change rate.

[0100] In step S42, the Sigmoid function is a kind of S-shaped curve function, and its output value range is (0, 1), which is very suitable for generating an inhibition factor with a value between 0 and 1. The steepness adjustment parameter k controls the shape of the Sigmoid function. The greater the value of k, the steeper the Sigmoid function, and the more sensitive the low-frequency gain inhibition factor to the change of the normalized low-frequency energy change rate. The value of k can be preset according to the actual application scenario and empirical data, for example, the value of k can be set to 5. By adjusting the value of k, the response degree of the low-frequency gain inhibition factor to the energy change rate can be finely adjusted, so as to optimize the image artifact suppression effect.

[0101] The present scheme combines normalization processing and Sigmoid nonlinear mapping to obtain a more reasonable low-frequency gain inhibition factor, which can be better used for subsequent gain adjustment to improve image quality.

[0102] Further, step S5 includes:

[0103] S51: determining whether the low-frequency energy value exceeds the artifact detection dynamic threshold, and if so, determining that there is a bleeding artifact;

[0104] S52: if the low-frequency energy value does not exceed the artifact detection dynamic threshold, obtaining a pre-set low-frequency reference gain, and taking the low-frequency reference gain as a real-time low-frequency gain coefficient;

[0105] 53: if the low-frequency energy value exceeds the artifact detection dynamic threshold, calculating a real-time low-frequency gain coefficient according to the artifact detection dynamic threshold, the low-frequency gain inhibition factor and the low-frequency energy value.

[0106] In step S51, the system determines whether the low-frequency energy value exceeds the artifact detection dynamic threshold. If the low-frequency energy value exceeds the artifact detection dynamic threshold, the system determines that there is a bleeding artifact in the current endoscopic ultrasound image. This determination provides a basis for whether to perform gain adjustment subsequently.

[0107] In step S52, it is determined that the low-frequency energy value does not exceed the artifact detection dynamic threshold, which means that the current image can not be affected by a significant bleeding artifact. As a processing method, the system can directly use the pre-set low-frequency reference gain as the real-time low-frequency gain coefficient. The low-frequency reference gain is a pre-set stable gain value, which is used when no bleeding artifact is detected. The purpose is to maintain the original state of the image and avoid unnecessary gain adjustment.

[0108] In step S53, it is determined that the low-frequency energy value exceeds the artifact detection dynamic threshold, which means that there is a bleeding artifact. According to the artifact detection dynamic threshold, the low-frequency gain inhibition factor and the low-frequency energy value, the real-time low-frequency gain coefficient is calculated. The calculated real-time low-frequency gain coefficient will be used for subsequent gain adjustment to achieve the effect of suppressing the bleeding artifact.

[0109] In some embodiments, in a minimally invasive endoscopic ultrasound surgery, after the ultrasound probe acquires the ultrasound echo signal, the device first performs frequency domain analysis to extract the low-frequency energy value. Assuming that the low-frequency energy value of the current frame of ultrasound image is 80 units, and the pre-calculated artifact detection dynamic threshold is 70 units. In step S51, the system compares 80 and 70, and determines that the low-frequency energy value exceeds the artifact detection dynamic threshold, so it is determined that there may be a bleeding artifact. At this time, step S52 is skipped and step S53 is directly executed. In step S53, assuming that the low-frequency gain inhibition factor calculated in step S4 is 0.5, and the pre-set low-frequency reference gain is 10 dB. According to the formula Gain_lf = Base_gain - Lf_inhibition * (Current_lf -DT), the real-time low-frequency gain coefficient Gain_lf = 10 - 0.5 * (80 - 70) = 5 dB is calculated. Finally, the system will use 5 dB as the real-time low-frequency gain coefficient to perform gain adjustment on the low-frequency component of the ultrasound echo signal to suppress the bleeding artifact and improve the image quality. Conversely, if the low-frequency energy value is 60 units and does not exceed the threshold of 70 units, step S52 is executed and the low-frequency reference gain of 10 dB is directly used without performing gain adjustment related to artifact suppression. Thus, the technical effect of maintaining the original gain of the image when there is no bleeding artifact and performing gain adjustment when a bleeding artifact is detected is achieved.

[0110] By judging whether a bleeding artifact appears or not in the present application, the gain adjustment is realized on demand, and the gain adjustment is only performed when the bleeding artifact is detected, and the original gain is maintained when there is no artifact. This selective gain adjustment strategy can ensure the artifact suppression effect while preserving as much normal image information as possible, improving the efficiency and pertinence of image processing.

[0111] Further, in step S53, the formula for calculating the real-time low-frequency gain coefficient according to the artifact detection dynamic threshold, the low-frequency gain inhibition factor, and the low-frequency energy value is:

[0112] Gain_lf = Base_gain - Lf_inhibition * (Current_lf - DT), wherein Gain_lf is the real-time low-frequency gain coefficient, Base_gain is the preset low-frequency reference gain, Lf_inhibition is the low-frequency gain inhibition factor, Current_lf is the low-frequency energy value, and DT is the artifact detection dynamic threshold.

[0113] The calculation formula of the real-time low-frequency gain coefficient embodies the calculation method of the real-time low-frequency gain coefficient. Specifically, the calculation method takes the preset low-frequency reference gain as the initial value, and dynamically adjusts according to the difference between the current low-frequency energy value and the artifact detection dynamic threshold. When the low-frequency energy value exceeds the artifact detection dynamic threshold, the difference is positive, and the real-time low-frequency gain coefficient will be reduced based on the low-frequency reference gain, realizing the suppression of the bleeding artifact. The low-frequency gain inhibition factor is used to adjust the suppression degree, realizing fine control of the gain suppression.

[0114] Through the above formula, the real-time low-frequency gain coefficient Gain_lf can be adaptively adjusted according to the degree of artifact, realizing fine gain control and improving the quality of endoscopic ultrasound images in complex physiological environments.

[0115] Further, step S6 includes:

[0116] S61: extracting high-frequency band energy information from the ultrasonic echo signal, the high-frequency band energy information at least including: a high-frequency band energy value, a high-frequency band energy fluctuation range, and a high-frequency band energy change rate;

[0117] S62: obtaining a pre-set high-frequency band energy baseline value;

[0118] S63: calculating a decay detection dynamic threshold according to the high-frequency band energy baseline value and the high-frequency band energy fluctuation range;

[0119] S64: calculating a high-frequency gain amplification factor according to the high-frequency energy change rate;

[0120] S65: calculating real-time high-frequency gain factor according to attenuation detection dynamic threshold, high-frequency gain amplification factor and high-frequency band energy value;

[0121] S66: applying real-time low-frequency gain factor and real-time high-frequency gain factor to frequency domain ultrasonic echo signal, performing gain adjustment, converting gain-adjusted frequency domain signal into time domain signal, performing image reconstruction, and generating ultrasonic endoscopic image.

[0122] In step S61, after receiving the ultrasonic echo signal, the signal is decomposed into components of different frequencies by frequency domain analysis method, such as Fourier transform. High-frequency band energy information extraction is to calculate the energy in the pre-set high-frequency range of the frequency domain signal. The high-frequency band energy value can be obtained by calculating the sum or average of the square of the high-frequency component amplitude. The high-frequency band energy fluctuation range can be obtained by calculating the difference between the maximum and minimum values of the high-frequency band energy value in a period of time. The high-frequency band energy change rate can be obtained by calculating the change amount of the high-frequency band energy value per unit time.

[0123] In step S62, the pre-set high-frequency band energy baseline value represents the high-frequency band energy level under ideal tissue state, i.e. without obvious attenuation. This baseline value can be calibrated by experiment, set by experience or automatically obtained during system initialization.

[0124] In step S63, the attenuation detection dynamic threshold can be a threshold adaptive to the degree of tissue attenuation. The formula for calculating the attenuation detection dynamic threshold uses the high-frequency band energy baseline value and the high-frequency band energy fluctuation range, aiming to set a reasonable attenuation detection standard according to the characteristics of the tissue itself and the signal fluctuation. For example, the formula can be designed as dynamic threshold equal to high-frequency band energy baseline value plus a term related to high-frequency band energy fluctuation range, so as to adapt to the energy fluctuation characteristics of different tissues.

[0125] In step S64, the high-frequency gain amplification factor is a coefficient for compensating the attenuation of high-frequency signal. The high-frequency energy change rate reflects the speed or degree of high-frequency signal attenuation, so the high-frequency gain amplification factor can be dynamically adjusted according to the high-frequency energy change rate. For example, when the high-frequency energy change rate indicates that the signal is rapidly attenuating, the high-frequency gain amplification factor is increased, and vice versa. The calculation of high-frequency gain amplification factor can be realized by linear function, nonlinear function or table mapping, etc.

[0126] In step S65, the real-time high-frequency gain coefficient is the gain value finally applied to the high-frequency signal. The calculation of the real-time high-frequency gain coefficient comprehensively considers the attenuation detection dynamic threshold, the high-frequency gain amplification factor, and the current high-frequency band energy value. If the current high-frequency band energy value is lower than the attenuation detection dynamic threshold, it indicates that there may be tissue attenuation, and at this time, the real-time high-frequency gain coefficient needs to be calculated according to the high-frequency gain amplification factor and the gap between the current energy value and the threshold, to compensate for the attenuation. The calculation formula can be designed as the real-time high-frequency gain coefficient being equal to a basic high-frequency gain plus a term related to the high-frequency gain amplification factor and the energy value gap.

[0127] In step S66, the real-time low-frequency gain coefficient and the real-time high-frequency gain coefficient are applied to the low-frequency component and the high-frequency component of the frequency domain ultrasonic echo signal, respectively. The gain-adjusted frequency domain signal is converted back to a time domain signal through inverse Fourier transform and the like. The time domain signal undergoes subsequent image reconstruction processing, such as envelope detection, logarithmic compression, scan conversion, and image interpolation, to finally generate an ultrasonic endoscopic image. By simultaneously adjusting the low-frequency and high-frequency gains, the overall quality of the image can be improved while suppressing artifacts and compensating for tissue attenuation. Thus, a clearer and more accurate ultrasonic endoscopic image can be obtained to assist doctors in diagnosis and surgical operations.

[0128] Further, step S66 includes:

[0129] S661: multiplying the real-time low-frequency gain coefficient with the low-frequency component of the ultrasonic echo signal to realize gain adjustment of the low-frequency component; and multiplying the real-time high-frequency gain coefficient with the high-frequency component of the ultrasonic echo signal to realize gain adjustment of the high-frequency component;

[0130] S662: performing Gaussian filtering on the high-frequency component and the low-frequency component to smooth noise, to obtain a gain-adjusted frequency domain signal;

[0131] S663: performing inverse Fourier transform on the gain-adjusted frequency domain signal to convert the frequency domain signal into a time domain signal;

[0132] S664: performing envelope detection on the time domain signal, and performing logarithmic compression on the envelope-detected time domain signal to enhance image contrast;

[0133] S665: performing scan conversion on the logarithmically compressed time domain signal to convert the polar coordinate form of the ultrasonic echo signal into a rectangular coordinate form, performing image interpolation, and generating an ultrasonic endoscopic image.

[0134] In step S661, the real-time low-frequency gain coefficient and the real-time high-frequency gain coefficient are applied to the low-frequency component and the high-frequency component of the ultrasonic echo signal, respectively. Specifically, in the frequency domain, the ultrasonic echo signal is decomposed into two components of low frequency and high frequency. The calculated real-time low-frequency gain coefficient is multiplied by the low-frequency component to adjust the low-frequency signal strength. Similarly, the calculated real-time high-frequency gain coefficient is multiplied by the high-frequency component to adjust the high-frequency signal strength. Thus, independent gain control can be performed for different frequency components to more finely compensate for signal attenuation and suppress artifacts.

[0135] In step S662, Gaussian filtering is used to smooth noise. Specifically, the gain-adjusted frequency domain signal, containing low-frequency and high-frequency components, will be subjected to Gaussian filtering processing respectively. The Gaussian filter is a linear smoothing filter that can effectively reduce random noise in the signal. The standard deviation of the Gaussian filter can be adaptively adjusted according to the preset high-frequency noise level, and the size of the filter matches the resolution of the ultrasonic endoscopic image, ensuring the filtering effect while avoiding excessive smoothing of image details. Through Gaussian filtering, the signal-to-noise ratio of the image can be improved, making the image clearer.

[0136] In step S663, inverse Fourier transform is used to convert the frequency domain signal back to the time domain signal. Specifically, the gain-adjusted and filtered frequency domain signal needs to be converted back to the time domain through inverse Fourier transform for subsequent image reconstruction operations. The inverse Fourier transform is the inverse process of the Fourier transform, which can convert the signal from the frequency domain representation to the time domain representation. Thus, the frequency domain processed signal can be restored to the time domain ultrasonic echo signal.

[0137] In step S664, envelope detection and logarithmic compression are used to enhance image contrast. Specifically, the time domain signal is first subjected to envelope detection to extract the amplitude information of the ultrasonic echo signal, obtaining the envelope of the signal. Then, the envelope-detected signal is subjected to logarithmic compression. Logarithmic compression can expand the dynamic range of the image, mapping the amplitude value to a gray scale range more suitable for human eye observation. Thus, the contrast of the image can be enhanced, making the tissue structure and lesion details in the image more clearly visible.

[0138] In step S665, scan conversion and image interpolation are used to generate ultrasonic endoscopic images. Specifically, the time domain signal after logarithmic compression is usually ultrasonic echo data represented in polar coordinates, which needs to be converted to rectangular coordinates through scan conversion to present the image on the display device. The scan conversion process usually accompanies image interpolation operations, such as linear interpolation or bilinear interpolation, to fill in pixel gaps and generate continuous and smooth ultrasonic endoscopic images. Thus, the final ultrasonic endoscopic image suitable for physician observation and diagnosis is generated.

[0139] Further, step S662 comprises:

[0140] S6621: decompose the high-frequency component and the low-frequency component into real part signals and imaginary part signals respectively;

[0141] S6622: respectively perform Gaussian filtering on the real part signals and the imaginary part signals to obtain filtered real part signals and filtered imaginary part signals, wherein a standard deviation of the Gaussian filter is adaptively adjusted according to a preset high-frequency noise level, and a size of the Gaussian filter matches a resolution of the endoscopic ultrasound image;

[0142] S6623: combine the filtered real part signals and the filtered imaginary part signals into filtered high-frequency components and filtered low-frequency components, and combine the filtered high-frequency components and the filtered low-frequency components as the gain-adjusted frequency domain signals.

[0143] In step S6621, after the ultrasonic echo signal is decomposed into high-frequency components and low-frequency components in the frequency domain, each frequency component exists in the form of a complex number, containing a real part and an imaginary part. The real part signals and the imaginary part signals are decomposed in order to subsequently perform more refined processing on these two components of the signal.

[0144] In step S6622, the decomposed real part signals and the imaginary part signals are respectively subjected to Gaussian filtering. The standard deviation of the Gaussian filter is adaptively adjusted, and the adjustment is based on the preset high-frequency noise level. When the high-frequency noise level is high, a Gaussian filter with a larger standard deviation is used, and the filtering strength is thereby enhanced, thus achieving more effective noise smoothing. Conversely, when the noise level is low, a Gaussian filter with a smaller standard deviation is used to avoid over-smoothing of image details, and the resolution of the image is thereby guaranteed. In addition, the size of the Gaussian filter matches the resolution of the endoscopic ultrasound image, ensuring that the filtering operation effectively reduces noise while maintaining the clarity of the image. For example, the size of the Gaussian filter can be set to a size corresponding to one or several pixels in the image.

[0145] In step S6623, the filtered real part signals and the filtered imaginary part signals are recombined to form filtered high-frequency components and filtered low-frequency components. Finally, the filtered high-frequency components and the filtered low-frequency components are combined to form gain-adjusted frequency domain signals, providing higher-quality frequency domain data for subsequent image reconstruction steps. Through the above refined Gaussian filtering process, the noise in the endoscopic ultrasound image can be more effectively removed, and the signal-to-noise ratio and visual quality of the image are improved to assist doctors in making more accurate diagnoses.

[0146] In a second aspect, an endoscopic ultrasound image processing device is provided, which is applied in the steps of any of the above endoscopic ultrasound image processing methods. The device comprises:

[0147] The first acquisition module 201 is configured to acquire an ultrasonic echo signal, perform frequency domain analysis, and extract low-frequency energy information, which at least includes a low-frequency energy value, a low-frequency energy fluctuation range, and a low-frequency energy change rate.

[0148] The second acquisition module 202 is configured to acquire a pre-set low-frequency energy baseline value.

[0149] The dynamic threshold adjustment module 203 is configured to calculate an artifact detection dynamic threshold based on the low-frequency energy baseline value and the low-frequency energy fluctuation range.

[0150] The dynamic gain adjustment module 204 is configured to calculate a low-frequency gain suppression factor based on the low-frequency energy change rate.

[0151] The gain coefficient adjustment module 205 is configured to calculate a real-time low-frequency gain coefficient based on the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value.

[0152] The image reconstruction module 206 is configured to apply the real-time low-frequency gain coefficient to the ultrasonic echo signal in the frequency domain, perform gain adjustment, convert the gain-adjusted frequency domain signal into a time domain signal, perform image reconstruction, and generate an ultrasonic endoscopic image.

[0153] The first acquisition module 201 can receive an ultrasonic echo signal from an ultrasonic endoscope probe. The received signal is then used for frequency domain analysis, which can be achieved by fast Fourier transform. After frequency domain analysis, the low-frequency energy information of the ultrasonic echo signal is extracted. The low-frequency energy information includes a low-frequency energy value, which reflects the intensity of the low-frequency component; a low-frequency energy fluctuation range, which indicates the change amplitude of the low-frequency energy over time; and a low-frequency energy change rate, which represents the speed of the low-frequency energy change.

[0154] The second acquisition module 202 is configured to provide a pre-set low-frequency energy baseline value. The low-frequency energy baseline value represents the normal low-frequency energy level without artifacts or interference, which can be stored in the memory of the device or pre-set by the user.

[0155] The dynamic threshold adjustment module 203 uses the low-frequency energy baseline value provided by the second acquisition module 202 and the low-frequency energy fluctuation range extracted by the first acquisition module 201 for artifact detection dynamic threshold calculation. The calculation of the artifact detection dynamic threshold takes into account the normal fluctuation of the low-frequency energy, making the artifact detection more flexible and accurate.

[0156] The dynamic gain adjustment module 204 calculates a low-frequency gain suppression factor based on the low-frequency energy change rate provided by the first acquisition module 201. The low-frequency gain suppression factor is used to control the degree of low-frequency gain adjustment. The higher the change rate, the larger the suppression factor may be, thereby achieving suppression of artifacts.

[0157] The gain coefficient adjustment module 205 is the core component of the device, which comprehensively considers the artifact detection dynamic threshold calculated by the dynamic threshold adjustment module 203, the low-frequency gain suppression factor calculated by the dynamic gain adjustment module 204, and the current low-frequency energy value extracted by the first acquisition module 201, and calculates the real-time low-frequency gain coefficient. The real-time low-frequency gain coefficient will be used in the subsequent gain adjustment step.

[0158] The image reconstruction module 206 applies the real-time low-frequency gain coefficient calculated by the gain coefficient adjustment module to the frequency domain ultrasonic echo signal, realizes the gain adjustment of the ultrasonic echo signal. The gain-adjusted frequency domain signal is then converted back to a time domain signal, for example, by inverse Fourier transform to realize the conversion from frequency domain to time domain. Finally, the time domain signal is used for image reconstruction to generate an ultrasonic endoscopic image for the doctor to observe and diagnose.

[0159] In some embodiments, the first acquisition module 201 can be realized by a high-speed analog-to-digital converter and a digital signal processor. The analog-to-digital converter is responsible for converting the received analog ultrasonic echo signal into a digital signal, and the digital signal processor executes the frequency domain analysis and energy information extraction algorithm. The frequency domain analysis algorithm can use fast Fourier transform, and the low-frequency energy information extraction can be realized by a band-pass filter and an energy calculation unit.

[0160] The second acquisition module 202 can be composed of a memory, in which a low-frequency energy baseline value is pre-stored. The low-frequency energy baseline value can be pre-set according to experimental data or clinical experience.

[0161] The dynamic threshold adjustment module 203, the dynamic gain adjustment module 204 and the gain coefficient adjustment module 205 can be realized by programmable logic devices or microcontrollers, which run pre-programmed algorithms to calculate the artifact detection dynamic threshold, the low-frequency gain suppression factor and the real-time low-frequency gain coefficient in real time according to the input low-frequency energy information, the low-frequency energy baseline value and the low-frequency energy fluctuation range.

[0162] The image reconstruction module 206 can be composed of a graphics processor and image processing software. The graphics processor is responsible for high-speed signal processing and image calculation, and the image processing software executes the signal-to-image conversion and reconstruction algorithm, including the steps of inverse Fourier transform, envelope detection, logarithmic compression and scan conversion, etc.

[0163] For example, when the ultrasound endoscope probe scans a region with rich blood vessels, the first acquisition module 201 detects that the low-frequency energy value and the energy change rate increase, and the dynamic threshold adjustment module 203 and the dynamic gain adjustment module 204 adjust the artifact detection dynamic threshold and the low-frequency gain suppression factor accordingly. The gain coefficient adjustment module 205 calculates a reduced real-time low-frequency gain coefficient according to these parameters, and the image reconstruction module 206 applies this gain coefficient to process the ultrasound signal, so that the bleeding artifact is effectively suppressed and the image quality is improved in the finally generated ultrasound endoscope image.

[0164] In this document, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.

[0165] The above description is merely illustrative of the application, and not intended to limit the scope of the application. Various modifications and changes can be made by those of ordinary skill in the art without departing from the spirit and scope of the application. Any modification, equivalent replacement, improvement, and the like made within the spirit and principle of the application shall fall within the scope of the application.

Claims

1. An endoscopic ultrasound image processing method for improving ultrasound image quality in minimally invasive endoscopic ultrasound surgery, characterized in that, The method includes the following steps: S1: Acquire ultrasonic echo signals, perform frequency domain analysis, and extract low-frequency energy information. The low-frequency energy information includes at least: low-frequency energy value, low-frequency energy fluctuation range, and low-frequency energy change rate. S2: Obtain the preset low-frequency energy baseline value; S3: Calculate the artifact detection dynamic threshold based on the low-frequency energy baseline value and the low-frequency energy fluctuation range; S4: Calculate the low-frequency gain suppression factor based on the low-frequency energy change rate; S5: Calculate the real-time low-frequency gain coefficient based on the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value; S6: Apply the real-time low-frequency gain coefficient to the ultrasonic echo signal in the frequency domain to adjust the gain, convert the frequency domain signal after gain adjustment into a time domain signal, perform image reconstruction, and generate an ultrasonic endoscopic image. Step S6 includes: S61: Based on the ultrasonic echo signal, extract high-frequency energy information, wherein the high-frequency energy information includes at least: high-frequency energy value, high-frequency energy fluctuation range, and high-frequency energy change rate; S62: Obtain the preset high-frequency energy baseline value; S63: Calculate the attenuation detection dynamic threshold based on the high-frequency energy baseline value and the high-frequency energy fluctuation range; S64: Calculate the high-frequency gain amplification factor based on the high-frequency energy change rate; S65: Calculate the real-time high-frequency gain coefficient based on the attenuation detection dynamic threshold, the high-frequency gain amplification factor, and the high-frequency energy value; S66: The real-time low-frequency gain coefficient and the real-time high-frequency gain coefficient are applied to the ultrasound echo signal in the frequency domain to adjust the gain. The frequency domain signal after gain adjustment is converted into a time domain signal, and image reconstruction is performed to generate an endoscopic ultrasound image. In step S3, the formula for calculating the artifact detection dynamic threshold is: DT = BLV (1 + α FR / BLV) exp(-β (1 - CCF)); where DT is the artifact detection dynamic threshold, BLV is the low-frequency energy baseline value, FR is the low-frequency energy fluctuation range, and CCF is the coupling confidence factor. α is the influence factor of the low-frequency energy fluctuation range, representing the degree of positive influence of the low-frequency energy fluctuation range on the dynamic threshold of artifact detection; β represents the degree of nonlinear influence of the coupling confidence factor, indicating the degree of nonlinear negative influence of the coupling confidence factor on the dynamic threshold of artifact detection.

2. The method for processing endoscopic ultrasound images according to claim 1, characterized in that, Step S3 includes: S31: Obtain the coupling degree parameter between the endoscopic ultrasound probe and the lesion area, wherein the coupling degree parameter includes at least the ultrasound echo signal intensity; S32: Calculate the coupling confidence factor based on the intensity of the ultrasonic echo signal; S33: Calculate the artifact detection dynamic threshold based on the coupling confidence factor, the low-frequency energy baseline value, and the low-frequency energy fluctuation range.

3. The method for processing endoscopic ultrasound images according to claim 1, characterized in that, Step S4 includes: S41: Normalize the low-frequency energy change rate to obtain the normalized low-frequency energy change rate; S42: The low-frequency gain suppression factor is calculated using a nonlinear mapping function based on the normalized low-frequency energy change rate; wherein the nonlinear mapping function is the Sigmoid function, and the calculation formula is: Lf_inhibition = 1 / (1 + exp(-k Normalized_rate)), where Lf_inhibition is the low-frequency gain suppression factor, k is a preset sigmoid function steepness adjustment parameter; Normalized_rate is the normalized low-frequency energy change rate.

4. The method for processing endoscopic ultrasound images according to claim 1, characterized in that, Step S5 includes: S51: Determine whether the low-frequency energy value exceeds the artifact detection dynamic threshold. If it does, then determine that a bleeding artifact exists. S52: If the low-frequency energy value does not exceed the artifact detection dynamic threshold, then obtain the preset low-frequency reference gain and use the low-frequency reference gain as the real-time low-frequency gain coefficient. 53: If the low-frequency energy value exceeds the artifact detection dynamic threshold, then calculate the real-time low-frequency gain coefficient based on the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value.

5. The method for processing endoscopic ultrasound images according to claim 4, characterized in that, In step S53, the formula for calculating the real-time low-frequency gain coefficient based on the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value is as follows: Gain_lf = Base_gain - Lf_inhibition (Current_lf - DT), where Gain_lf is the real-time low-frequency gain coefficient, Base_gain is the preset low-frequency reference gain, Lf_inhibition is the low-frequency gain suppression factor, Current_lf is the low-frequency energy value, and DT is the artifact detection dynamic threshold.

6. The method for processing endoscopic ultrasound images according to claim 1, characterized in that, Step S66 includes: S661: Multiply the real-time low-frequency gain coefficient by the low-frequency component of the ultrasonic echo signal to adjust the gain of the low-frequency component; multiply the real-time high-frequency gain coefficient by the high-frequency component of the ultrasonic echo signal to adjust the gain of the high-frequency component. S662: Perform Gaussian filtering on the high-frequency component and the low-frequency component to smooth the noise and obtain the frequency domain signal with gain adjustment. S663: Perform an inverse Fourier transform on the frequency domain signal after gain adjustment to convert the frequency domain signal into a time domain signal; S664: Perform envelope detection on the time-domain signal, and perform logarithmic compression on the time-domain signal after envelope detection to enhance image contrast; S665: Perform a scanning transformation on the time-domain signal after logarithmic compression to convert the polar coordinate form of the ultrasound echo signal into rectangular coordinate form, perform image interpolation, and generate an endoscopic ultrasound image.

7. The method for processing endoscopic ultrasound images according to claim 1, characterized in that, Step S662 includes: S6621: Decompose the high-frequency component and the low-frequency component into real part signals and imaginary part signals, respectively; S6622: Gaussian filtering is performed on the real part signal and the imaginary part signal respectively to obtain the filtered real part signal and the filtered imaginary part signal, wherein the standard deviation of the Gaussian filter is adaptively adjusted according to the preset high-frequency noise level, and the size of the Gaussian filter is matched with the resolution of the ultrasound endoscope image. S6623: Combine the filtered real part signal and the filtered imaginary part signal to form the filtered high-frequency component and the low-frequency component, and combine the filtered high-frequency component and the low-frequency component to form the frequency domain signal after gain adjustment.

8. An endoscopic ultrasound image processing device, characterized in that, The device is used in the method according to any one of claims 1-7, the device comprising: The first acquisition module is used to acquire ultrasonic echo signals, perform frequency domain analysis, and extract low-frequency energy information. The low-frequency energy information includes at least: low-frequency energy value, low-frequency energy fluctuation range, and low-frequency energy change rate. Second acquisition module: used to acquire a pre-set low-frequency energy baseline value; Dynamic threshold adjustment module: used to calculate the dynamic threshold for artifact detection based on the low-frequency energy baseline value and the low-frequency energy fluctuation range; Dynamic gain adjustment module: used to calculate the low-frequency gain suppression factor based on the energy change rate of the low-frequency band; Gain coefficient adjustment module: used to calculate the real-time low-frequency gain coefficient based on the artifact detection dynamic threshold, the low-frequency gain suppression factor, and the low-frequency energy value; Image reconstruction module: The real-time low-frequency gain coefficient is applied to the ultrasonic echo signal in the frequency domain to adjust the gain. The frequency domain signal after gain adjustment is converted into a time domain signal to perform image reconstruction and generate an endoscopic ultrasound image.

Images