A method and system for processing artifact interference of musculoskeletal ultrasound

Abstract

The present invention relates to the field of image processing, and in particular to a method and system for processing artifact interference in musculoskeletal ultrasound, comprising: recording an acoustic shadow artifact area as a target area; reducing the ultrasonic frequency to obtain a first area; performing contrast enhancement on the target area and matching it with pixels in the first area to obtain an interference area; increasing the ultrasonic frequency to obtain a second area; recording an area outside the target area in the second area as an acoustic shadow disappearance area; increasing the grayscale values ​​of all pixels outside the interference area in the second area to obtain a brightness-enhanced area and performing contrast enhancement to obtain a third area; obtaining a non-interference area based on the acoustic shadow disappearance area; obtaining a noise interference degree of pixels in the non-interference area; adjusting the grayscale values ​​of pixels in the interference area using the noise interference degree to obtain a grayscale correction area; and obtaining a musculoskeletal ultrasound image with artifacts eliminated using the grayscale correction area. The present invention eliminates artifact interference.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a method and system for processing artifact interference of musculoskeletal ultrasound. Background Art

[0002] Musculoskeletal ultrasound (MSU) is a technique that uses high-frequency ultrasound waves to image muscles, bones, and related soft tissues. When sound waves encounter tissues of varying densities, they are reflected, refracted, and scattered. The reflected sound waves are received by the probe and converted into ultrasound images. Acoustic shadowing is a common ultrasound image artifact in MSU testing, typically appearing behind highly reflective tissue interfaces, such as bone or calcified areas. Acoustic shadowing can obscure or distort the true characteristics of the diseased tissue, affecting diagnostic accuracy.

[0003] The general method to remove acoustic shadow artifacts is to directly enhance the image of the acoustic shadow artifact area, such as enhancing the contrast. However, this solution is easily interfered by the noise in the artifact area, resulting in the inability to restore or display accurate muscle and bone texture. Summary of the Invention

[0004] To solve the above problems, the present invention provides a method and system for processing artifact interference of musculoskeletal ultrasound.

[0005] The present invention provides a method and system for processing artifact interference in musculoskeletal ultrasound using the following technical solutions:

[0006] An embodiment of the present invention provides a method for processing artifact interference of musculoskeletal ultrasound, the method comprising the following steps:

[0007] A musculoskeletal ultrasound image containing an acoustic shadow artifact region is recorded as a target image, and the acoustic shadow artifact region in the target image is recorded as a target region; the ultrasonic frequency is reduced to obtain a first image; the region in the first image that is identical to the target region is recorded as a first region; contrast enhancement is performed on the target region so that the enhanced target region and the first region have a grayscale distribution with minimal difference, and the texture of the pixel points in the enhanced target region is matched with the texture of the pixel points in the first region to obtain an interference region in the target region;

[0008] Increasing the ultrasonic frequency and obtaining a second image; recording the acoustic shadow artifact region in the second image as a second region; recording the region outside the target region within the second region as an acoustic shadow subsidence region; increasing the grayscale values ​​of all pixels outside the interference region within the second region to obtain a brightness-enhanced region, wherein the difference in grayscale values ​​between all pixels in the brightness-enhanced region and the target region is minimal;

[0009] performing contrast enhancement on the brightness-enhanced region to obtain a third region, wherein the third region has a grayscale distribution with the smallest difference from the first region;

[0010] The texture of the pixel points in the acoustic shadow subsidence area in the target image is matched with the texture of the pixel points in the acoustic shadow subsidence area in the second image to obtain the non-interference area in the acoustic shadow subsidence area, and the texture of the pixel points in the non-interference area in the third area is matched with the texture of the pixel points in the non-interference area in the first image to obtain the noise interference degree of each pixel point in the non-interference area in the third area. The grayscale value of the pixel point in the interference area in the target area is adjusted using the noise interference degree to obtain a grayscale correction area. The grayscale correction area is contrast enhanced and then fused with the first area to obtain a musculoskeletal ultrasound image with artifacts eliminated.

[0011] Preferably, the contrast enhancement of the target region such that the enhanced target region has a grayscale distribution with the smallest difference from the first region comprises the following specific steps:

[0012] Obtaining a grayscale histogram composed of the grayscale values ​​of all pixels in the first area, recorded as a first histogram;

[0013] According to the first histogram, the grayscale values ​​of all pixels in the target area are histogram-normalized to obtain the enhanced target area.

[0014] Preferably, the step of matching the texture of the pixels in the enhanced target area with the texture of the pixels in the first area to obtain the interference area in the target area comprises the following specific steps:

[0015] Matching the texture of the pixels in the enhanced target area with the texture of the pixels in the first area to obtain the noise interference degree of each pixel in the enhanced target area;

[0016] In the enhanced target area, the area formed by all pixels whose noise interference degree is greater than the first preset threshold is recorded as the interference area;

[0017] Matching the texture of the pixel points in the acoustic shadow subsidence area in the target image with the texture of the pixel points in the acoustic shadow subsidence area in the second image to obtain a non-interference area in the acoustic shadow subsidence area includes the following specific steps:

[0018] Matching the texture of the pixel points in the acoustic shadow subsidence area in the target image with the texture of the pixel points in the acoustic shadow subsidence area in the second image to obtain the noise interference degree of each pixel point in the acoustic shadow subsidence area in the target image;

[0019] In the sound-shadow subsidence area of ​​the target image, the area formed by all pixels whose noise interference level is less than a second preset threshold is recorded as a non-interference area;

[0020] The noise interference degree is calculated by the distribution texture similarity between pixel points.

[0021] Preferably, the specific steps for obtaining the noise interference degree are as follows:

[0022] Record the pixels in the enhanced target area, the pixels in the sound shadow subsidence area in the target image, or the pixels in the non-interference area in the third area as a first pixel point set, and record the pixels in the first area, the pixels in the sound shadow subsidence area in the second image, or all the pixels in the non-interference area on the first image as a second pixel point set;

[0023] The pixels in the first pixel set are recorded as target pixels, and the pixels in the second pixel set are recorded as reference pixels.

[0024] For any target pixel, the reference pixel with the greatest distribution texture similarity to the target pixel is recorded as the matching reference pixel of the target pixel;

[0025] For any target pixel A, the matching reference pixel corresponding to the target pixel A is recorded as B, and the number of target pixels corresponding to the matching reference pixel B is recorded as N. The distribution texture similarity between the i-th target pixel and the matching reference pixel B is recorded as ,Will Denoted as the noise interference degree of the target pixel point A, where a represents the distribution texture similarity between the target pixel point A and the matching reference target pixel point B; exp() represents an exponential function with a natural constant as the base.

[0026] Preferably, the specific steps of obtaining the distribution texture similarity between the pixel points are as follows:

[0027] For any pixel point, in the neighborhood of the pixel point, the Sobel operator is used to obtain the gradient directions of all pixels in the neighborhood. The gradient directions of all pixels constitute a gradient direction histogram, which is recorded as the neighborhood texture distribution feature of the pixel point; the cosine similarity of the neighborhood texture distribution features of any two pixels is recorded as the distribution texture similarity between the pixels.

[0028] Preferably, the step of increasing the grayscale values ​​of all pixels outside the interference area in the second area to obtain the brightness-enhanced area includes the following specific steps:

[0029] In the second area, the mean grayscale value A1 of all pixels outside the interference area is obtained; in the target area, the mean grayscale value A2 of all pixels outside the interference area is obtained; the difference between A2 and A1 is recorded as B, and the grayscale value of all pixels outside the interference area in the second area is added with B to obtain the brightness improvement area.

[0030] Preferably, the grayscale values ​​of pixels in the interference area of ​​the target area are adjusted using the noise interference degree to obtain a grayscale correction area, which includes the following specific steps:

[0031] Obtain a grayscale histogram of all pixels in the target area. The grayscale histogram represents a histogram curve, where the horizontal axis is the grayscale value and the vertical axis is the frequency of the grayscale value in the target area.

[0032] Obtain any pixel point b within the interference area of ​​the target area. The grayscale value and the frequency of occurrence of the grayscale value of pixel point b correspond to a point Qb on the histogram curve. Filter point Qb on the histogram curve to obtain the filtered point Qb. The vertical coordinate of the filtered point Qb is marked as Fb. Obtain the pixel point within the target area that has the smallest grayscale difference with pixel point b and a grayscale value occurrence frequency equal to Fb. Use the grayscale value of this pixel point as the adjusted grayscale value of pixel point b.

[0033] Adjust the grayscale values ​​of all pixels in the interference area of ​​the target area, and the target area after the grayscale value adjustment is recorded as the grayscale correction area;

[0034] The size of the filter kernel used in the filtering is determined by the contrast enhancement error amount, and the contrast enhancement error amount is positively correlated with the noise interference degree of all pixels in the non-interference area of ​​the third area.

[0035] Preferably, the filter kernel size is , w represents the contrast enhancement error, L0 represents the preset basic filter kernel size, m represents the noise interference degree of pixel b in the target area, Indicates the rounding symbol.

[0036] Preferably, the contrast enhancement error amount is equal to the average value of the noise interference levels of all pixels in the non-interference area of ​​the third area.

[0037] In another embodiment of the present invention, a system for processing artifact interference of musculoskeletal ultrasound is provided. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for processing artifact interference of musculoskeletal ultrasound are implemented.

[0038] The beneficial effects of the technical solution of the present invention are:

[0039] The present invention performs contrast enhancement on the target area so that the enhanced target area has a grayscale distribution with the smallest difference from the first area, matches the texture of the pixel points in the enhanced target area with the texture of the pixel points in the first area, and obtains an interference area in the target area; the obtained interference area can preliminarily describe the noise interference situation or noise distribution situation in the enhanced target area.

[0040] Increase the ultrasonic frequency and obtain a second image; the acoustic shadow artifact area in the second image is recorded as the second area; the area outside the target area in the second area is recorded as the acoustic shadow subsidence area; increase the grayscale values ​​of all pixels outside the interference area in the second area to obtain a brightness enhancement area, and the difference in the grayscale values ​​of all pixels in the brightness enhancement area and the target area is minimized. In this process, the brightness of the brightness enhancement area is aligned with that of the target area. When contrast enhancement is subsequently performed on the brightness enhancement area, the problem of large differences in contrast enhancement effects caused by the different ultrasonic frequencies used during acquisition of the second area and the target area is avoided. The reason why the pixels in the interference area are not taken into account is that the pixels in the interference area may be pixels introduced by ultrasonic image noise, and the grayscale values ​​of these pixels interfere with the brightness enhancement process and the subsequent contrast enhancement process.

[0041] The brightness-enhanced region is then contrast-enhanced to obtain a third region, which has a grayscale distribution that is minimally different from the first region. The texture of the pixels within the acoustic shadow subsidence region in the target image is matched with the texture of the pixels within the acoustic shadow subsidence region in the second image, resulting in a non-interference region within the acoustic shadow subsidence region. Compared to the target region, the brightness-enhanced region is acquired at a higher ultrasonic frequency and then brightness-adjusted, providing greater detail but also exhibiting more noise interference. The third region eliminates the influence of the interference region, resulting in the presence of noise-affected regions in the third region being identical or similar to those in the enhanced target region. Furthermore, the subsidence-enhanced region represents an artifact-induced change due to the increased frequency. The muscle-bone texture within the subsidence-enhanced region in the target image is not affected by artifacts, resulting in a relatively less noise-affected non-interference region. However, the subsidence-enhanced region in the second image is affected by artifacts, allowing the non-interference region to be used to analyze and describe the artifact interference. Therefore, the third and non-interference regions can be used to modify the contrast enhancement method, reducing the impact of noise on the contrast enhancement process.

[0042] Furthermore, the texture of the pixels within the non-interference area of ​​the third region is matched with the texture of the pixels within the non-interference area of ​​the first image to obtain the noise interference level for each pixel within the non-interference area of ​​the third region. This noise interference level is then used to adjust the grayscale values ​​of the pixels within the interference area of ​​the target region to obtain a grayscale correction region. This process adjusts the grayscale values ​​of the pixels within the interference area of ​​the target region to remove pixel features (such as slope, maximum, and minimum values) displayed on the histogram curve. These features can introduce interference during contrast enhancement, ultimately restoring and displaying a more accurate musculoskeletal texture. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 This is a flowchart of a method for processing artifact interference of musculoskeletal ultrasound provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0045] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a method and system for processing artifact interference in musculoskeletal ultrasound according to the present invention. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0047] The following describes in detail a method and system for processing artifact interference in musculoskeletal ultrasound provided by the present invention with reference to the accompanying drawings.

[0048] See also Figure 1 , which shows a flowchart of a method for processing artifact interference of musculoskeletal ultrasound provided by one embodiment of the present invention, the method comprising the following steps:

[0049] Step S001: a musculoskeletal ultrasound image with an acoustic shadow artifact region is recorded as a target image, and the acoustic shadow artifact region in the target image is recorded as a target region.

[0050] An ultrasonic probe is used to collect musculoskeletal ultrasound images. The ultrasound frequency is P0. In this embodiment, P0 is 10 MHz. In other embodiments, P0 may be set to other values, which are not specifically limited in this embodiment. Acoustic shadow artifact areas in the musculoskeletal ultrasound images are detected.

[0051] Acoustic shadow artifacts are common in musculoskeletal ultrasound testing. They are caused by the fact that behind some strongly reflective objects (such as bones or calcifications), the sound waves are reflected or absorbed in large quantities, making it impossible to form an obvious image, thus forming a dark shadow area.

[0052] As a preferred example, detecting an acoustic shadow artifact area in a musculoskeletal ultrasound image includes the following method:

[0053] The acoustic shadow artifact region in the musculoskeletal ultrasound image is obtained using a semantic segmentation network, for example, the DeeplabV3 semantic segmentation network is used. This network is a well-known technology, and the specific principles are not described in detail in this embodiment.

[0054] As another example, detecting an acoustic shadow artifact region in a musculoskeletal ultrasound image includes the following method:

[0055] The musculoskeletal ultrasound image is binary segmented using a threshold segmentation algorithm to obtain all shadow areas in the musculoskeletal ultrasound image. In this embodiment, the first segmentation threshold used in the threshold segmentation is 50, and the connected domain composed of pixel points in the musculoskeletal ultrasound image whose grayscale value is less than the first segmentation threshold is recorded as the shadow area.

[0056] Furthermore, the threshold segmentation algorithm is used again to perform binary segmentation on the musculoskeletal ultrasound image to obtain all the highlighted areas in the musculoskeletal ultrasound image. In this embodiment, the second segmentation threshold used in the threshold segmentation is 180, and the connected domain composed of pixels in the musculoskeletal ultrasound image whose grayscale values ​​are greater than the second segmentation threshold is recorded as the highlighted area.

[0057] In this embodiment, the shadow region below and closest to the highlighted region is recorded as a candidate region. The maximum height of each candidate region (that is, the distance between two pixels on the candidate region's outline that are the farthest apart in the vertical direction) is obtained. Candidate regions whose maximum height is less than a preset height threshold are recorded as acoustic shadow artifact regions.

[0058] In this embodiment, the preset height threshold is 20% of the maximum height of the highlight area above the candidate area.

[0059] When an acoustic shadow artifact region exists in the musculoskeletal ultrasound image, the acquired musculoskeletal ultrasound image is recorded as a target image, and the acoustic shadow artifact region on the target image is recorded as a target region.

[0060] Step S002, reduce the ultrasonic frequency and obtain a first image; the area where the target area is located in the first image is recorded as the first area; enhance the contrast of the target area, match the texture of the pixel points in the enhanced target area with the texture of the pixel points in the first area, and obtain the interference area in the target area.

[0061] When there is an acoustic shadow artifact area in the musculoskeletal ultrasound image, the ultrasonic frequency is further reduced to P1. In this embodiment, P1=(1-p1×P0), where p1 represents the frequency reduction coefficient. This embodiment is described using p1=10% as an example. In other embodiments, it can be set to other values, and this embodiment does not make specific limitations.

[0062] After reducing the ultrasound frequency, a muscle-bone ultrasound image at the same position and with the same probe direction is collected again and recorded as the first image.

[0063] On the first image, the area encompassing the target region is designated the first region. Compared to the target image, the first image has lower resolution and fewer detailed musculoskeletal textures, as low-frequency ultrasound has greater penetration but weaker ability to distinguish subtle musculoskeletal textures. However, the (non-detailed) musculoskeletal textures within the first region are relatively distinct and clear.

[0064] This embodiment performs image enhancement (e.g., contrast enhancement) on the target area, and then matches and fuses it with the first area, so that the unclear or lost muscle-bone texture in the target area is revealed, while also including certain muscle-bone texture details, thereby reducing artifact interference.

[0065] First, the contrast of the target region is enhanced so that the enhanced target region has a grayscale distribution with the smallest difference from the first region.

[0066] It should be noted that the target image is acquired at a higher ultrasonic frequency than the first image, and the resolution of the target image is higher than that of the first image, but the noise in the target image is relatively more, especially these noises cannot be ignored in the contrast enhancement process.

[0067] As an example, contrast enhancement is performed on the target region so that the enhanced target region has a grayscale distribution with the smallest difference from the first region, including the following methods:

[0068] Obtaining a grayscale histogram composed of the grayscale values ​​of all pixels in the first area, recorded as a first histogram;

[0069] According to the first histogram, grayscale values ​​of all pixels in the target area are histogram-normalized to obtain an enhanced target area. The enhanced target area has a similar histogram to the first area.

[0070] The above-mentioned histogram normalization based on the first histogram is a well-known technique, and the specific principle thereof will not be described in detail in this embodiment.

[0071] Furthermore, the enhanced target region is matched with the first region to obtain an interference region within the target region. The interference region describes the region that is interfered with by noise during the contrast enhancement process.

[0072] As an example, matching the texture of the pixel points in the enhanced target area with the texture of the pixel points in the first area to obtain the interference area in the target area includes:

[0073] For any pixel in the enhanced target area or the first area, the Sobel operator is used to obtain the gradient directions of all pixels in the neighborhood of the pixel (for example, a 9×9 neighborhood centered on the pixel). The gradient directions of all pixels constitute a gradient direction histogram, which is used to describe the probability of occurrence of each gradient direction in the neighborhood. The gradient direction histogram is recorded as the neighborhood texture distribution feature of the pixel.

[0074] Two pixels are randomly obtained from the enhanced target area and the first area respectively, and the distribution texture similarity of the two pixels is calculated according to the neighborhood texture distribution characteristics of the two pixels. The pixel points in the enhanced target area are recorded as target pixels, and the pixel points in the first area are recorded as reference pixels.

[0075] For any target pixel, the reference pixel with the greatest distribution texture similarity to the target pixel is recorded as the matching reference pixel of the target pixel.

[0076] It should be noted that there may be multiple target pixels that correspond to the same matching reference pixel. For any target pixel A, the matching reference pixel corresponding to target pixel A is denoted as B, and the number of target pixels (including target pixel A) corresponding to matching reference pixel B is denoted as N. The distribution texture similarity between the i-th target pixel and the matching reference pixel B is denoted as ,Will It is denoted as the noise interference degree of the target pixel A after the enhanced contrast (abbreviated as the noise interference degree of the target pixel A), where a represents the distribution texture similarity between the target pixel A and the matching reference target pixel B; exp() represents an exponential function with a natural constant as the base.

[0077] Thus, the texture of the pixels in the enhanced target area is matched with the texture of the pixels in the first area, and the noise interference degree of each pixel in the enhanced target area is obtained.

[0078] The smaller the noise interference level is, the The larger it is, the more the musculoskeletal texture represented by the target pixel point A can match the musculoskeletal texture represented by the matching reference target pixel point B (or the two are the same musculoskeletal textures but at different resolutions). At the same time, the musculoskeletal texture represented by the matching reference target pixel point B cannot be well matched with the musculoskeletal texture represented by other target pixels (that is, target pixels other than A). At this time, it means that the musculoskeletal texture represented by the target pixel point A is less affected by the noise in the target area during the contrast enhancement process.

[0079] On the contrary, the greater the noise interference, the The smaller it is, the less the musculoskeletal texture represented by the target pixel A can match the musculoskeletal texture represented by the matching reference pixel B. At the same time, the musculoskeletal texture represented by the matching reference pixel B can be better matched with the musculoskeletal texture represented by other target pixels. This means that the musculoskeletal texture represented by the target pixel A is more disturbed by the noise in the target area during the contrast enhancement process, or even the target pixel A is formed after the noise is contrast enhanced.

[0080] As an example, the distribution texture similarity of two pixels is calculated based on the neighborhood texture distribution features of the two pixels, including a method of obtaining the cosine similarity of the neighborhood texture distribution features of the two pixels, and using the cosine similarity as the distribution texture similarity.

[0081] Furthermore, in the enhanced target area, all pixels with noise interference greater than a first preset threshold th1 are obtained, and the area formed by these pixels is recorded as an interference area. The interference area is within the enhanced target area and also within the target area.

[0082] This embodiment is described by taking th1=0.57 as an example. In other embodiments, th1 may be set to other values, which is not specifically limited in this embodiment.

[0083] Step S003, increase the ultrasonic frequency and obtain a second image; record the acoustic shadow artifact area in the second image as the second area; record the area outside the target area in the second area as the acoustic shadow subsidence area; increase the grayscale values ​​of all pixels outside the interference area in the second area to obtain a brightness-enhanced area; and perform contrast enhancement on the brightness-enhanced area to obtain a third area.

[0084] Further, the ultrasonic frequency is increased to P2. In this embodiment, P2=(1+p2×P0), where p2 represents the frequency increase coefficient. This embodiment is described using p2=0.2×p1 as an example. In other embodiments, it can be set to other values, which is not specifically limited in this embodiment.

[0085] After increasing the ultrasound frequency, a muscle-bone ultrasound image is acquired again at the same location and with the same probe orientation, which is recorded as a second image. An acoustic shadow artifact region in the second image is detected and recorded as a second region.

[0086] Compared to the target image, the second image has higher resolution and more detailed musculoskeletal texture, due to the high-frequency ultrasound's ability to distinguish subtle musculoskeletal textures but its weaker penetration. However, the overall grayscale within the second region is darker, making the musculoskeletal texture less prominent. Furthermore, the second region includes the target area. The area within the second region but outside the target area in the second image is designated as the acoustic shadow subsidence region. This region describes the artifacts created by the increased ultrasound frequency and reduced penetration.

[0087] The grayscale values ​​of all pixels outside the interference area in the second area are increased to obtain a brightness-enhanced area, where the difference between the grayscale values ​​of all pixels in the brightness-enhanced area and the target area is minimal. The brightness-enhanced area does not include the interference area.

[0088] As an example, increasing the grayscale values ​​of all pixels outside the interference area in the second area to obtain a brightness-enhanced area includes the following methods:

[0089] In the second area, the mean grayscale value A1 of all pixels outside the interference area is obtained; in the target area, the mean grayscale value A2 of all pixels outside the interference area is obtained; the difference between A2 and A1 is recorded as B, and the grayscale value of all pixels outside the interference area in the second area is added with B to obtain the brightness improvement area.

[0090] The overall brightness of the boosted region (i.e., the mean of the overall grayscale values) is minimized compared to the overall brightness of the target region. This ensures that the brightness of the boosted region is aligned with the target region. This allows for subsequent contrast enhancement of the boosted region to avoid significant discrepancies in contrast enhancement due to the different ultrasonic frequencies used during acquisition. Pixels within the interference region are not considered because they may be caused by noise introduced by the ultrasound image, and their grayscale values ​​interfere with the brightness boost and subsequent contrast enhancement processes.

[0091] The brightness-enhanced region is contrast-enhanced so that the enhanced brightness-enhanced region has a grayscale distribution with the smallest difference from the first region. The enhanced brightness-enhanced region is recorded as the third region. The specific process is the same as step S002.

[0092] Furthermore, the texture of the pixels in the acoustic shadow subsidence area in the target image is matched with the texture of the pixels in the acoustic shadow subsidence area in the second image to obtain the noise interference level of each pixel in the acoustic shadow subsidence area in the target image. This process is similar to step S002.

[0093] Furthermore, in the sound and shadow subsidence area of ​​the target image, all pixels whose noise interference level is less than a second preset threshold th2 are obtained, and the area formed by these pixels is recorded as a non-interference area.

[0094] The purpose of obtaining the third region and the non-interference region in the above process is that after the target region is enhanced in the above process, the enhanced target region obtained has a higher resolution than the first region, but also has more noise-affected areas. The reason for the existence of the noise-affected areas is that the target region is acquired at a higher ultrasonic frequency and has a stronger perception of noise and muscle-bone texture. In addition, due to the interference of artifacts, the noise in the target region is relatively large. When contrast is enhanced, the noise in the target region has a greater impact on the contrast enhancement process, resulting in the enhanced target region having a non-negligible noise-affected area. Compared with the target region, the brightness-enhanced region is acquired at a higher ultrasonic frequency and then brightness adjusted. It has further detailed information and also has more noise interference. However, since the difference between P2 and P0 is smaller than that between P1 and P0, and the third region removes the influence of the interference area, the existence of the noise-affected areas in the third region is the same or similar to the existence of the noise-affected areas in the enhanced target region. Furthermore, the faded enhancement region is an artifact-induced change region due to frequency increase. The muscle-bone texture within the faded enhancement region in the target image is not affected by artifacts, so the resulting non-interference region is relatively less affected by noise. However, the faded enhancement region in the second image is affected by artifacts, so the non-interference region can be used to analyze and describe the artifact interference. Therefore, this embodiment utilizes the third region and the non-interference region to modify the contrast enhancement method, reducing the impact of noise during the contrast enhancement process.

[0095] Step S004: using the third area and the non-interference area to correct the contrast enhancement method to obtain a grayscale correction area.

[0096] As an example, the contrast enhancement method is modified using the third region and the non-interference region, including the following methods:

[0097] All pixel points in the non-interference area of ​​the third region are recorded as the first pixel distribution, and all pixel points in the non-interference area on the first image are recorded as the second pixel distribution. The texture of the pixel points in the first pixel distribution is matched with the texture of the pixel points in the second pixel distribution to obtain the noise interference degree of each pixel point in the first pixel distribution. This process is the same as step S002.

[0098] When the noise interference level of each pixel point in the first pixel distribution is relatively high, it indicates that when contrast enhancement is performed on the target region or the second region, noise interference in the contrast enhancement process is relatively high. Specifically, when contrast enhancement is performed on the target region, the interference region is also involved in the process, resulting in noise interference. The interference region participates in (or participates excessively in) the contrast enhancement process of the target region, causing the interference region actually obtained in step S002 to be unreliable. This in turn causes noise interference to still exist when the brightness of the second region is enhanced (i.e., when obtaining the brightness-enhanced region). (i.e., although the interference region is excluded when obtaining the brightness-enhanced region, the interference region may be unreliable because the interference region participates in the contrast enhancement process of the target region.) Ultimately, the noise interference level of each pixel point in the first pixel distribution is relatively high.

[0099] The grayscale values ​​of the pixels in the interference area of ​​the target area are adjusted using the noise interference degree of each pixel in the first pixel distribution to obtain a grayscale correction area; note that the grayscale values ​​of the pixels outside the interference area of ​​the target area are not adjusted.

[0100] As an example, the grayscale values ​​of the pixels in the interference area of ​​the target area are adjusted using the noise interference degree of all pixels in the first pixel distribution, including the following methods:

[0101] Obtain a grayscale histogram of all pixels in the target area. The grayscale histogram represents a histogram curve, where the horizontal axis is the grayscale value and the vertical axis is the frequency of occurrence of the grayscale value in the target area.

[0102] Obtain any pixel b within the interference region of the target image. Obtain the grayscale value and frequency of occurrence of pixel b. This grayscale value and frequency correspond to a point Qb on the histogram curve. Filter point Qb on the histogram curve to obtain a filtered point Qb. The vertical coordinate of filtered point Qb is labeled Fb. Determine the pixel within the target region that has the smallest grayscale difference from pixel b and a grayscale value frequency equal to Fb. This pixel's grayscale value is used as the grayscale value of pixel b. The grayscale difference here refers to the absolute value of the grayscale difference.

[0103] This embodiment uses the mean filtering method for filtering, and the filter kernel size is , w represents the contrast enhancement error. In this embodiment, w is set to be equal to the mean value of the noise interference degree of all pixels in the first pixel distribution. The larger the mean value of the noise interference degree is, the more difficult it is to restore the muscle-bone texture included in the non-interference area in the third region during contrast enhancement, indicating that the contrast enhancement process has a larger error due to the interference of noise.

[0104] L0 represents the size of the basic filter kernel, and this embodiment is described by taking L0=3 as an example. Indicates the rounding up symbol. It should also be noted that when the filter kernel size is an even number, the filter kernel size will be automatically increased by one so that the filter kernel size is always an odd number.

[0105] The purpose of filtering and resetting the grayscale value of pixel b in the above process is to remove the features (such as slope, maximum and minimum values) exhibited by pixel b on the histogram curve. This is because contrast enhancement in this embodiment is primarily based on the distribution characteristics of the grayscale histogram. By removing these features from the histogram curve, interference introduced by pixel b during the contrast enhancement process can be avoided. The greater the average noise interference level for all pixels within the first pixel distribution, the more necessary it is to remove the features exhibited by pixel b on the histogram curve.

[0106] Similarly, the grayscale values ​​of all pixels in the interference area are updated.

[0107] As another example, the grayscale value of the pixel in the interference area of ​​the target area is adjusted using the noise interference degree of each pixel in the first pixel distribution, including the following method:

[0108] When the mean value of the noise interference degree of all pixels in the first pixel distribution is greater than the preset interference threshold (for example, greater than 0.35), the grayscale values ​​of all pixels in all interference areas are mean filtered, for example, using a 7×7 filter kernel for filtering, and the grayscale value after filtering is the updated grayscale value of all pixels in the interference area.

[0109] This example calculates quickly but has low accuracy. This is because different pixels within the interference area have different interference effects during contrast enhancement. Directly performing mean filtering on the grayscale values ​​of all pixels within the interference area results in incomplete removal of the effects of some pixels, which still introduces interference during contrast enhancement. Alternatively, some pixels contain non-noise muscle-bone information, which, after removal, makes it impossible to retain more detailed muscle-bone texture.

[0110] In some other embodiments, the grayscale value of the pixel in the interference area of ​​the target area is adjusted using the noise interference level of each pixel in the first pixel distribution, including the following methods:

[0111] The filter kernel size is , where m represents the noise interference degree of pixel b obtained in step S002. The larger m is, the more likely pixel b is to contain more noise. In this case, the more noise pixel b contains, the more serious the noise interference in the contrast enhancement process (that is, the greater the error in the contrast enhancement process), and the more necessary it is to remove the noise from pixel b.

[0112] The target area after the grayscale value is updated and adjusted is recorded as a grayscale correction area.

[0113] Step S005: performing contrast enhancement on the grayscale correction region and fusing the image with the first region to obtain a musculoskeletal ultrasound image with artifacts eliminated.

[0114] The grayscale correction region is contrast-enhanced so that the enhanced grayscale correction region has a grayscale distribution with the smallest difference from the first region (similar to step S002 ).

[0115] The enhanced grayscale correction area is further fused with the first area to obtain a pseudo-image interference elimination area.

[0116] As an example, the enhanced grayscale-corrected region is fused with the first region, including a method of fusing the enhanced grayscale-corrected region with the first region using a Gaussian pyramid fusion algorithm to obtain a fused region. The Gaussian pyramid fusion algorithm is well known in the art, and its principles are not further described in this embodiment.

[0117] The fused region encompasses both the high-resolution, artifact-free musculoskeletal texture details in the target image acquired at the high-frequency ultrasound frequency, and the distinct, artifact-free, lower-resolution musculoskeletal texture in the first image acquired at the low-frequency ultrasound frequency. The target region is replaced with the fused region on the target image to produce an artifact-free musculoskeletal ultrasound image, which is then displayed on a monitor.

[0118] In another embodiment, when there is no texture detail in the target area obtained in step S001, for example, when the average grayscale value of all pixels in the target area is less than 50, it indicates that there is extremely strong artifact interference when performing muscle-bone detection at the ultrasonic frequency of P0.

[0119] In this case, one method is to directly replace the target region with the first region to obtain an artifact-free musculoskeletal ultrasound image, which is then displayed on a monitor. This method is computationally fast, but does not preserve or restore the musculoskeletal texture details affected by artifacts.

[0120] Another method is to reduce the ultrasonic frequency by 1% (other embodiments may reduce it to other values), re-record the reduced ultrasonic frequency as P0, record the resulting musculoskeletal ultrasound image as the target image, and record the acoustic shadow artifact region within the target image as the target region. Then, the method described in steps S002 to S005 is followed to obtain an artifact-free musculoskeletal ultrasound image, which is then displayed on a monitor. This method preserves or restores the musculoskeletal texture details despite artifact interference as much as possible.

[0121] In another embodiment, if the second region obtained in step S003 lacks any texture detail, for example, if the mean grayscale value of all pixels in the second region is less than 30, this indicates that strong artifacts are present when performing musculoskeletal testing at the ultrasonic frequency P2. In this case, P2 is increased by 1% (other values ​​may be used in other embodiments), and the increased ultrasonic frequency is re-denoted as P2. The artifact-free musculoskeletal ultrasound image is obtained using the methods of steps S003 to S005, and the image is displayed on a monitor.

[0122] Similarly, when there is no obvious texture detail in the first area obtained in step S002, for example, when the mean grayscale value of all pixels in the first area is less than 80, P1 is reduced by 1%, and the reduced ultrasonic frequency is renamed as P1. Then, the musculoskeletal ultrasound image with artifacts eliminated is obtained according to the method of steps S002 to S005 above, and the musculoskeletal ultrasound image is displayed on the display.

[0123] In another embodiment of the present invention, a system for processing artifact interference of musculoskeletal ultrasound is provided. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above steps S001 to S005 are implemented.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for processing artifact interference of musculoskeletal ultrasound, characterized in that: The method comprises the following steps: A musculoskeletal ultrasound image containing an acoustic shadow artifact region is recorded as a target image, and the acoustic shadow artifact region in the target image is recorded as a target region; the ultrasonic frequency is reduced to obtain a first image; the region in the first image that is identical to the target region is recorded as a first region; contrast enhancement is performed on the target region so that the enhanced target region and the first region have a grayscale distribution with minimal difference, and the texture of the pixel points in the enhanced target region is matched with the texture of the pixel points in the first region to obtain an interference region in the target region; Increasing the ultrasonic frequency and obtaining a second image; recording the acoustic shadow artifact region in the second image as a second region; recording the region outside the target region within the second region as an acoustic shadow subsidence region; increasing the grayscale values ​​of all pixels outside the interference region within the second region to obtain a brightness-enhanced region, wherein the difference in grayscale values ​​between all pixels in the brightness-enhanced region and the target region is minimal; performing contrast enhancement on the brightness-enhanced region to obtain a third region, wherein the third region has a grayscale distribution with the smallest difference from the first region; The texture of the pixel points in the same area corresponding to the acoustic shadow subsidence area in the target image is matched with the texture of the pixel points in the acoustic shadow subsidence area in the second image to obtain the non-interference area in the same area corresponding to the acoustic shadow subsidence area in the target image, and the texture of the pixel points in the same area corresponding to the non-interference area in the third area is matched with the texture of the pixel points in the same area corresponding to the non-interference area in the first image to obtain the noise interference degree of each pixel point in the same area corresponding to the non-interference area in the third area, and the grayscale value of the pixel points in the interference area in the target area is adjusted according to the noise interference degree to obtain a grayscale correction area, and the grayscale correction area is contrast enhanced and then fused with the first area to obtain a musculoskeletal ultrasound image with artifacts eliminated.

2. The method for processing artifact interference of musculoskeletal ultrasound according to claim 1, characterized in that: The contrast enhancement of the target region so that the enhanced target region has a grayscale distribution with the smallest difference from the first region includes the following specific steps: Obtaining a grayscale histogram composed of the grayscale values ​​of all pixels in the first area, recorded as a first histogram; According to the first histogram, the grayscale values ​​of all pixels in the target area are histogram-normalized to obtain the enhanced target area.

3. The method for processing artifact interference of musculoskeletal ultrasound according to claim 1, characterized in that: The step of matching the texture of the pixels in the enhanced target area with the texture of the pixels in the first area to obtain the interference area in the target area includes the following specific steps: Matching the texture of the pixels in the enhanced target area with the texture of the pixels in the first area to obtain the noise interference degree of each pixel in the enhanced target area; In the enhanced target area, the area formed by all pixels whose noise interference degree is greater than the first preset threshold is recorded as the interference area; The texture of the pixel points in the same area corresponding to the acoustic shadow subsidence area in the target image is matched with the texture of the pixel points in the acoustic shadow subsidence area in the second image to obtain the non-interference area in the same area corresponding to the acoustic shadow subsidence area in the target image, which includes the following specific steps: Matching the texture of the pixel points in the same area corresponding to the acoustic shadow subsidence region in the target image with the texture of the pixel points in the acoustic shadow subsidence region in the second image to obtain the noise interference degree of each pixel point in the same area corresponding to the acoustic shadow subsidence region in the target image; In the target image, the area corresponding to the acoustic shadow disappearance area and the area formed by all pixels whose noise interference level is less than a second preset threshold is recorded as a non-interference area; The noise interference degree is calculated by the distribution texture similarity between pixel points.

4. A method for processing artifact interference of musculoskeletal ultrasound according to claim 1 or 3, characterized in that: The specific steps for obtaining the noise interference degree are as follows: The pixels in the enhanced target area, the pixels in the same area corresponding to the sound shadow subsidence area in the target image, or the pixels in the same area corresponding to the non-interference area in the third area are recorded as a first pixel point set, and the pixels in the first area, the pixels in the sound shadow subsidence area in the second image, or all the pixels in the same area corresponding to the non-interference area on the first image are recorded as a second pixel point set; The pixels in the first pixel set are recorded as target pixels, and the pixels in the second pixel set are recorded as reference pixels. For any target pixel, the reference pixel with the greatest distribution texture similarity to the target pixel is recorded as the matching reference pixel of the target pixel; For any target pixel A, the matching reference pixel corresponding to the target pixel A is recorded as B, and the number of target pixels corresponding to the matching reference pixel B is recorded as N. The distribution texture similarity between the i-th target pixel and the matching reference pixel B is recorded as ,Will Denoted as the noise interference degree of the target pixel point A, where a represents the distribution texture similarity between the target pixel point A and the matching reference target pixel point B; exp() represents an exponential function with a natural constant as the base.

5. The method for processing artifact interference of musculoskeletal ultrasound according to claim 3, characterized in that: The specific steps for obtaining the distribution texture similarity between the pixels are as follows: For any pixel point, in the neighborhood of the pixel point, the Sobel operator is used to obtain the gradient directions of all pixels in the neighborhood. The gradient directions of all pixels constitute a gradient direction histogram, which is recorded as the neighborhood texture distribution feature of the pixel point. The cosine similarity of the neighborhood texture distribution features of any two pixels is recorded as the distribution texture similarity between the pixels.

6. The method for processing artifact interference of musculoskeletal ultrasound according to claim 1, characterized in that: The method of increasing the grayscale values ​​of all pixels outside the corresponding interference area in the second area to obtain a brightness-enhanced area includes the following specific steps: In the second area, obtain the mean grayscale value A1 of all pixels outside the interference area; in the target area, obtain the mean grayscale value A2 of all pixels outside the interference area; the difference between A2 and A1 is recorded as B, and the grayscale value of all pixels outside the interference area in the second area corresponding to the interference area is added with B to obtain the brightness improvement area.

7. The method for processing artifact interference of musculoskeletal ultrasound according to claim 1, characterized in that: The grayscale values ​​of the pixels in the interference area of ​​the target area are adjusted using the noise interference degree to obtain a grayscale correction area, which includes the following specific steps: Obtain a grayscale histogram of all pixels in the target area. The grayscale histogram represents a histogram curve, where the horizontal axis is the grayscale value and the vertical axis is the frequency of the grayscale value in the target area. Obtain any pixel point b within the interference area of ​​the target area. The grayscale value and the frequency of occurrence of the grayscale value of pixel point b correspond to a point Qb on the histogram curve. Filter point Qb on the histogram curve to obtain the filtered point Qb. The vertical coordinate of the filtered point Qb is marked as Fb. Obtain the pixel point within the target area that has the smallest grayscale difference with pixel point b and a grayscale value occurrence frequency equal to Fb. Use the grayscale value of this pixel point as the adjusted grayscale value of pixel point b. Adjust the grayscale values ​​of all pixels in the interference area of ​​the target area, and the target area after the grayscale value adjustment is recorded as the grayscale correction area; The size of the filter kernel used in the filtering is determined by the contrast enhancement error amount, and the contrast enhancement error amount is positively correlated with the noise interference degree of all pixels in the non-interference area of ​​the third area corresponding to the same area.

8. The method for processing artifact interference of musculoskeletal ultrasound according to claim 7, characterized in that: The filter kernel size is , w represents the contrast enhancement error, L0 represents the preset basic filter kernel size, m represents the noise interference degree of pixel b in the target area, Indicates the round-up symbol.

9. The method for processing artifact interference of musculoskeletal ultrasound according to claim 7, characterized in that: The contrast enhancement error amount is equal to the average value of the noise interference degree of all pixels in the non-interference area of ​​the third area corresponding to the same area.

10. A system for processing artifact interference of musculoskeletal ultrasound, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for processing artifact interference of musculoskeletal ultrasound according to any one of claims 1 to 9 are implemented.

Images