Ultrasonic reverberation suppression method based on stepped energy difference

Abstract

The invention relates to an ultrasonic reverberation suppression method based on step energy difference. The method comprises the steps of providing a target ultrasonic image to be subjected to reverberation suppression; based on the step energy difference caused by the reverberation noise in the depth direction of the target ultrasonic image, the reverberation noise area in the target ultrasonic image is identified and detected; on the basis of the region position of the recognized reverberation noise region, reverberation suppression is carried out on the reverberation noise region in the target ultrasonic image, so that the target ultrasonic image is updated after reverberation suppression, during reverberation suppression, a plurality of reverberation suppression target segments are extracted in the reverberation noise region in the depth direction of the target ultrasonic image, and then reverberation suppression is carried out on the reverberation suppression target segments; and smoothing the amplitude of each reverberation suppression target section so as to denoise a reverberation noise area in the target ultrasonic image. According to the invention, reverberation suppression can be effectively realized, and the reverberation suppression precision and reliability are improved.

Description

Technical Field

[0001] This invention relates to a method, and more particularly to an ultrasonic reverberation suppression method based on step energy difference. Background Technology

[0002] In medical ultrasound imaging, the main method utilizes the differences in acoustic impedance between human tissues to create varying degrees of reflection of ultrasound waves. A larger acoustic impedance difference results in stronger reflections, which are represented as high brightness in the ultrasound image. For superficial and abdominal ultrasound scans, the significant acoustic impedance differences between the skin, fascia, peritoneum, cavity walls, and surrounding tissues often create strong reflection interfaces. Sound waves bounce back and forth between the probe surface and these strong reflection interfaces, ultimately appearing as a cloud-like artifact beneath the strong reflection interface on the ultrasound image, generally referred to as reverberation noise / multiple reflection noise. This cloud-like artifact shrouds the tissue or cavity beneath the interface, affecting the doctor's identification and diagnosis of nearby tissues and lesions.

[0003] Reverberation noise originates from ultrasound physics phenomena, where reflected echoes are mixed with normal tissue echoes, making them difficult to distinguish using signal processing techniques. Currently, the commonly used reverberation suppression method in ultrasound systems is time gain compensation (TGC), which involves attenuating / reducing the signal energy at a certain depth (the depth where reverberation noise may occur) to reduce the impact of reverberation noise.

[0004] The TGC method for suppressing reverberation noise has a fixed range of suppression and lacks adaptability. The required depth of reverberation suppression may vary for different patients, and a single preset may result in suppression occurring in areas of normal tissue in some patients, or the suppression range may be insufficient. Furthermore, a single preset is also inadequate for different scanning sections and for different organs.

[0005] Furthermore, TGC only exists in depth direction control, and can only uniformly suppress the lateral direction. The width of reverberation noise is often limited, and uniform suppression may damage the normal tissue on both sides of the reverberation, affecting the doctor's diagnosis and making it difficult to meet the actual needs of ultrasound diagnosis. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology and provide an ultrasonic reverberation suppression method based on step energy difference, which can effectively achieve reverberation suppression and improve the accuracy and reliability of reverberation suppression.

[0007] According to the technical solution provided by the present invention, an ultrasonic reverberation suppression method based on step energy difference is provided, the ultrasonic reverberation suppression method comprising: Provide the target ultrasound image for reverberation suppression; Based on the step energy difference caused by reverberation noise in the depth direction of the target ultrasound image, the reverberation noise region in the target ultrasound image is identified and detected. When the reverberation noise region is identified in the target ultrasound region, the regional location of the reverberation noise region in the target ultrasound image is determined. Based on the location of the identified reverberant noise region, reverberation suppression is performed on the reverberant noise region within the target ultrasound image. The target ultrasound image is then updated after reverberation suppression. During reverberation suppression, several reverberation suppression target segments are extracted within the reverberation noise region along the depth direction of the target ultrasound image. Subsequently, the amplitude of each reverberation suppression target segment is smoothed to denoise the reverberation noise region within the target ultrasound image.

[0008] When identifying and detecting reverberant noise regions within a target ultrasound image, the process includes first performing depth feature recognition and detection on the target ultrasound image, wherein... When reverberant noise is determined to exist through deep feature recognition detection, lateral feature recognition detection is then performed on the target ultrasound image. After lateral feature recognition and detection, the location of the reverberant noise region within the target ultrasound image is identified and determined.

[0009] When performing deep feature recognition and detection, the following are included: Several reference scan lines are selected on the target ultrasound image, and each reference scan line extends along the depth direction of the target ultrasound image. For each reference scan line, the scan line depth feature of the reference scan line is extracted, and the reverberation noise feature state of each reference scan line is determined. When at least two reference scan lines exhibit reverberation noise characteristics, the presence of reverberation noise is confirmed after depth feature recognition detection.

[0010] The scan line depth feature includes several scan line depth sub-features, wherein the scan line depth sub-features include at least the scan line amplitude feature; When extracting the scan line depth features for each reference scan line, the following is included: Differentiate the amplitude characteristics of each reference scan line to obtain the amplitude abrupt change point of the current reference scan line; Based on all acquired amplitude abrupt change points, a depth step model is fitted to generate a step-like amplitude-depth fitting line for the current reference scan line, where... The stepped amplitude-depth fitting line includes several candidate line segments connected in sequence. Within the stepped amplitude-depth fitting line, the horizontal axis represents the depth of each candidate line segment within the target ultrasound image, and the vertical axis represents the amplitude of each candidate line segment within the target ultrasound image. Based on the stepped amplitude-depth fitting line, the scan line amplitude characteristics of the current reference scan line are determined; Along the depth direction of the target ultrasound image, when the scan line amplitude feature shows a gradual decreasing trend, the scan line depth sub-feature corresponding to the scan line amplitude feature is a valid feature. At this time, the reverberation noise feature state of the current reference scan line is configured to have reverberation noise feature.

[0011] The scan line depth feature also includes, as sub-features of scan line depth, the length feature between abrupt change points, the step morphology similarity feature, and / or the depth feature of abrupt change point index, wherein... When there are multiple scan line depth sub-features within the scan line depth feature, the reverberation noise feature state of the current reference scan line will be configured to have reverberation noise features only when all scan line depth sub-features are valid features. The length feature between abrupt change points characterizes the length between two adjacent amplitude abrupt change points. When the length feature between abrupt change points is not greater than the length of the candidate line segment connected to the amplitude abrupt change point at the previous depth position, the scan line depth sub-feature corresponding to the length feature between abrupt change points is a valid feature. The step shape similarity feature characterizes the shape similarity between the step amplitude-depth fitting line and different preset reverberation step models. When there is a shape similarity greater than the shape similarity threshold, the scan line depth sub-feature corresponding to the step shape similarity feature is a valid feature. The mutation point index depth feature represents the index depth of each mutation point in the depth direction of the target ultrasound image. When the index depth of all mutation points matches the preset index depth, the scan line depth sub-feature corresponding to the mutation point index depth feature is a valid feature.

[0012] When calculating morphological similarity, the following are included: Calculate the dot product ratio between the stepped amplitude-depth fitting line and each preset reverberation step model, and use the calculated dot product ratio as the morphological similarity between the stepped amplitude-depth fitting line and the current preset reverberation step model. The preset reverberation ladder model is a preset amplitude-depth fitting line that characterizes the distribution of reverberation noise in ultrasound images.

[0013] When the reverberation noise is first-order reverberation, the index depth of the first amplitude abrupt change point along the depth direction of the target ultrasound image is less than 1 / 3 of the total depth index of the target ultrasound image, and the index depth of the second amplitude abrupt change point is less than 1 / 2 of the total depth index of the target ultrasound image.

[0014] Lateral feature recognition and detection includes: Along the arrangement order of the target scan lines on the target ultrasound image, the lateral continuity characteristics of two adjacent target scan lines are determined sequentially. The target scan line is a reference scan line with reverberant noise characteristics. The lateral continuity characteristics characterize the coherence of the reverberant noise characteristics of the two target scan lines. Based on the lateral continuity characteristics of two adjacent target scan lines, all target scan lines that satisfy the lateral continuity distribution trend are identified, and the region where the identified target scan lines are located is configured as the region location of the reverberation noise region.

[0015] Determining the lateral continuity characteristics of two adjacent target scan lines includes: Based on the stepped amplitude-depth fitting line corresponding to each target scan line, at least the line segment overlap features and line segment similarity features of each candidate line segment within two target scan lines are determined. When both the line segment overlap feature and the line segment similarity feature are valid features, the lateral continuity feature of the two adjacent target scan lines is lateral continuity. Along the arrangement direction of the target scan line on the target ultrasound image, when the determined lateral continuity features are successively lateral continuity, or the lateral continuity features of lateral continuity dominate, then it is determined that the lateral continuity distribution trend is satisfied.

[0016] When smoothing the amplitude of each reverberation suppression target segment, the methods include smoothing the amplitude of the reverberation suppression target segment by setting it to zero, smoothing the amplitude of the reverberation suppression target segment by linearly decreasing it, or smoothing the amplitude of the reverberation suppression target segment by curve-decreasing it.

[0017] The advantages of this invention are as follows: Based on the step energy difference caused by reverberation noise in the depth direction of the target ultrasound image, the reverberation noise region in the target ultrasound image is identified and detected. When a reverberation noise region is identified in the target ultrasound image, its location within the target ultrasound image is determined. Based on the location of the identified reverberation noise region, reverberation suppression is performed on the reverberation noise region in the target ultrasound image. After reverberation suppression, the target ultrasound image is updated, thereby effectively achieving reverberation suppression and improving the accuracy and reliability of reverberation suppression. Attached Figure Description

[0018] Figure 1 This is a schematic flowchart illustrating one embodiment of ultrasonic reverberation suppression according to the present invention.

[0019] Figure 2 This is a schematic diagram of one embodiment of the present invention for selecting a reference scan line on a target ultrasound image.

[0020] Figure 3 This is a schematic diagram of one embodiment of the present invention for determining the amplitude abrupt change point of each reference scan line.

[0021] Figure 4 This is an embodiment of the present invention in which the ultrasonic reverberation noise is continuously distributed in the transverse direction.

[0022] Figure 5 This is a schematic diagram of the first embodiment of the preset reverberation ladder model of the present invention.

[0023] Figure 6 This is a schematic diagram of a second embodiment of the pre-set reverberation ladder model of the present invention.

[0024] Figure 7 This is a schematic diagram of the third embodiment of the pre-set reverberation ladder model of the present invention.

[0025] Figure 8 This is a schematic diagram of the fourth embodiment of the pre-set reverberation ladder model of the present invention.

[0026] Figure 9 This is a schematic diagram of the first embodiment of the smoothing process of the present invention.

[0027] Figure 10 This is a schematic diagram of a second embodiment of the smoothing process of the present invention.

[0028] Figure 11 This is a schematic diagram of a third embodiment of the smoothing process of the present invention. Detailed Implementation

[0029] The present invention will be further described below with reference to specific accompanying drawings and embodiments.

[0030] To effectively suppress reverberation and improve its accuracy and reliability, this invention provides an ultrasonic reverberation suppression method based on step energy difference. Specifically, the ultrasonic reverberation suppression method includes: Provide the target ultrasound image for reverberation suppression; Based on the step energy difference caused by reverberation noise in the depth direction of the target ultrasound image, the reverberation noise region in the target ultrasound image is identified and detected. When the reverberation noise region is identified in the target ultrasound region, the regional location of the reverberation noise region in the target ultrasound image is determined. Based on the location of the identified reverberant noise region, reverberation suppression is performed on the reverberant noise region within the target ultrasound image. The target ultrasound image is then updated after reverberation suppression. During reverberation suppression, several reverberation suppression target segments are extracted within the reverberation noise region along the depth direction of the target ultrasound image. Subsequently, the amplitude of each reverberation suppression target segment is smoothed to denoise the reverberation noise region within the target ultrasound image.

[0031] Depend on Figure 1It is understood that when performing ultrasonic reverberation suppression in this invention, a target ultrasonic image should be provided. That is, the object of ultrasonic reverberation suppression in this invention is consistent with the object of existing reverberation suppression. Therefore, commonly used methods can be used to acquire / provide the target ultrasonic image. Figure 2 An embodiment of the target ultrasound image is shown in the figure; the specific methods of acquiring / providing the target ultrasound image will not be described here.

[0032] To overcome the problems caused by performing reverberation suppression on the entire target ultrasound image in existing technologies, when performing ultrasound reverberation suppression, the reverberant noise region within the target ultrasound image should be identified first. In specific implementation, when identifying the reverberant noise region, this invention is mainly based on the step energy difference caused by the reverberant noise in the depth direction of the target ultrasound image. When a reverberant noise region is identified within the target ultrasound image, the location of the reverberant noise region within the target ultrasound image should be determined simultaneously. The specific method and process for identifying the reverberant noise region will be explained in detail below.

[0033] Once the location of the reverberant noise region is identified, this invention should suppress reverberation within that region, while leaving the region outside the reverberant noise region unsuppressed to avoid damaging the display of normal tissue on either side of the reverberation. After reverberation updating, the target ultrasound image should be updated to obtain a new target ultrasound image. Displaying and outputting the updated target ultrasound image can effectively improve the diagnostic accuracy of medical personnel.

[0034] In one embodiment of the present invention, during reverberation suppression, a reverberation suppression target segment should be determined. The reverberation suppression target segment is also the object of reverberation suppression. The length direction of the reverberation suppression target segment generally coincides with the depth direction of the target ultrasound image. The method for generating the reverberation suppression target segment will be described below. After determining the reverberation suppression target segment, the amplitude of each reverberation suppression target segment can be smoothed to denoise the reverberation noise region within the target ultrasound image. That is, when performing reverberation suppression in this invention, the amplitude of the reverberation suppression target segment is mainly smoothed. The smoothing method will also be described below.

[0035] In one embodiment of the present invention, when identifying and detecting reverberant noise regions within a target ultrasound image, the process includes first performing depth feature identification and detection on the target ultrasound image, wherein... When reverberant noise is determined to exist through deep feature recognition detection, lateral feature recognition detection is then performed on the target ultrasound image. After lateral feature recognition and detection, the location of the reverberant noise region within the target ultrasound image is identified and determined.

[0036] It should be noted that when identifying and detecting reverberant noise regions based on the step energy difference caused by reverberant noise in the depth direction of the target ultrasound image, depth feature identification should be performed first. After depth feature identification, the main purpose is to determine whether reverberant noise exists in the target ultrasound image. Once the presence of reverberant noise is confirmed, lateral feature identification should be performed to determine the regional location of the reverberant noise region within the target ultrasound image.

[0037] In one embodiment of the present invention, deep feature recognition and detection includes: Several reference scan lines are selected on the target ultrasound image, and each reference scan line extends along the depth direction of the target ultrasound image. For each reference scan line, the scan line depth feature of the reference scan line is extracted, and the reverberation noise feature state of each reference scan line is determined. When at least two reference scan lines exhibit reverberation noise characteristics, the presence of reverberation noise is confirmed after depth feature recognition detection.

[0038] Figure 2 The figure illustrates one embodiment of selecting a reference scan line on the target ultrasound image; the blue dashed line in the figure represents the reference scan line. When performing depth feature recognition detection, multiple reference scan lines should be selected, such as multiple reference scan lines uniformly selected along the width direction of the target ultrasound image. Generally, multiple reference scan lines are parallel to each other (linear array) or radiate along the same center (phased array, arc array), and the reference scan lines extend along the depth direction of the target ultrasound image. That is, the reference scan lines are generally related to the ultrasound scanning method that forms the ultrasound image. The depth direction of the target ultrasound image is the depth direction of the ultrasound image in the usual sense; the meaning of the depth direction will not be elaborated here.

[0039] After selecting a reference scan line, the scan line depth features of each reference scan line should be extracted. Based on the extracted scan line depth features, the reverberation noise feature state corresponding to each reference scan line can be further determined. The reverberation noise feature state is either having reverberation noise features or not having reverberation noise features. When reverberation noise features are present, it can be considered that there may be reverberation noise at the position corresponding to the current reference scan line. Similarly, when the reverberation noise feature state is not having reverberation noise features, it can be considered that there is no reverberation noise at the position corresponding to the current reference scan line.

[0040] To avoid misidentification, reverberation noise is generally determined to exist only when at least two reference scan lines exhibit reverberation noise characteristics, thus allowing subsequent lateral feature recognition and detection to proceed.

[0041] In one embodiment of the present invention, the scan line depth feature includes a plurality of scan line depth sub-features, wherein the scan line depth sub-features include at least scan line amplitude features; When extracting the scan line depth features for each reference scan line, the following is included: Differentiate the amplitude characteristics of each reference scan line to obtain the amplitude abrupt change point of the current reference scan line; Based on all acquired amplitude abrupt change points, a depth step model is fitted to generate a step-like amplitude-depth fitting line for the current reference scan line, where... The stepped amplitude-depth fitting line includes several candidate line segments connected in sequence. Within the stepped amplitude-depth fitting line, the horizontal axis represents the depth of each candidate line segment within the target ultrasound image, and the vertical axis represents the amplitude of each candidate line segment within the target ultrasound image. Based on the stepped amplitude-depth fitting line, the scan line amplitude characteristics of the current reference scan line are determined; Along the depth direction of the target ultrasound image, when the scan line amplitude feature shows a gradual decreasing trend, the scan line depth sub-feature corresponding to the scan line amplitude feature is a valid feature. At this time, the reverberation noise feature state of the current reference scan line is configured to have reverberation noise feature.

[0042] In practice, the scan line depth feature can generally include one or more scan line depth sub-features. When identifying and detecting reverberant noise regions based on step energy differences, the scan line depth sub-features should generally include at least the scan line amplitude feature. That is, the scan line amplitude feature is a necessary scan line depth sub-feature within the scan line depth feature. Other scan line depth sub-features within the scan line depth feature can be selected as needed. Examples of specific selections will be given below.

[0043] When extracting the scan line depth features of each reference scan line, a stepped amplitude-depth fitting line should generally be constructed for the current reference scan line. Subsequently, based on the stepped amplitude-depth fitting line, the corresponding scan line amplitude features within the scan line depth features can be determined. One feasible implementation for constructing the stepped amplitude-depth fitting line is as follows: Based on the pixels corresponding to the reference scan lines and the target ultrasound image, the amplitude characteristics of each reference scan line can be determined. These amplitude characteristics can be the envelope or grayscale values ​​formed by the pixels corresponding to each current scan line. Figures 5-8 As shown in the diagram, the reverberation depth stepwise nature manifests as a gradient from bright to dark, from the strong interface to the multiple reflection artifacts and then to the tissue or cavity tissue. Different stepwise layers may contain descents or boundary bulges. Therefore, differentiating the amplitude characteristics of the reference scan line is necessary to determine the abrupt amplitude change points of the current reference scan line. Figure 3The image above shows the amplitude-depth correspondence of a reference scan line. Figure 3 The image below is a comparison of... Figure 3 The diagram obtained by differentiating the above figure is derived from... Figure 3 It can be seen that when there are three amplitude features with different amplitudes along the depth direction, after differentiation, two amplitude abrupt change points can be obtained. These amplitude abrupt change points can be understood as the extreme points of the derivative, and also the points where the amplitude changes at the edges of the stepped amplitude features. When the amplitude features of each reference scan line are other cases, refer to... Figure 3 And the corresponding explanations here, the specifics are based on obtaining the amplitude change point.

[0044] After obtaining the amplitude abrupt change point of the current reference scan line, a depth step model should be fitted. The purpose of the depth step model fitting is to construct a corresponding step-shaped amplitude-depth fitting line. Figures 5-8 An embodiment of a stepped amplitude-depth fitting line is shown. As illustrated, the formed stepped amplitude-depth fitting line is a combination of multiple line segments relating to depth and amplitude. The details of depth and amplitude can be found in the corresponding descriptions above. As shown in the illustration, the stepped amplitude-depth fitting line includes multiple candidate line segments, such as... Figures 5-8 The example shown includes three candidate line segments, the position and amplitude of which are related to the position of the reference scan line on the target ultrasound image.

[0045] A feasible depth ladder model fitting is as follows: After obtaining each amplitude abrupt change point, the length between two amplitude abrupt change points can be determined, and the depth-amplitude corresponding to the candidate line segment formed by connecting the two amplitude abrupt change points can be obtained. When the two amplitude abrupt change points belong to the abrupt change points corresponding to the first reverberation noise, a candidate line segment is configured in the direction of the vertical axis representing amplitude, with the first amplitude abrupt change point as the reference, and a candidate line segment is configured in the direction of the horizontal axis representing depth, with the last amplitude abrupt change point as the center. The length of the configured candidate line segment is: the smaller of 1.1 times the length of the candidate line segment formed by connecting the first and last amplitude abrupt change points and the corresponding lengths of the corresponding boundaries of the first and last abrupt change points and the target ultrasound image. If the length corresponding to 1.1 times the length of the line segment connecting the first and last amplitude abrupt change points exceeds the upper boundary of the target ultrasound image, that is, it exceeds the range of the target ultrasound image, then the length between the first amplitude abrupt change point and the upper boundary of the target ultrasound image should be determined. Otherwise, the length of the candidate line segment formed by connecting the first and last amplitude abrupt change points should be 1.1 times the length of the candidate line segment. The situation of the candidate line segment connected to the last amplitude abrupt change point can be referred to here.

[0046] As explained above, the first and last amplitude abrupt change points are negative extreme abrupt change points. When positive extreme abrupt change points also exist, the first and last amplitude abrupt change points should be connected to the depth-amplitude corresponding to the corresponding positive extreme abrupt change points to form the corresponding candidate line segments. It should be noted that of the two amplitude abrupt change points, the first amplitude abrupt change point is the amplitude abrupt change point near the near-field region within the target ultrasound image, and the last amplitude abrupt change point is the amplitude abrupt change point near the far-field region within the target ultrasound image. First-order reverberation noise specifically refers to the reverberation noise that appears first along the depth direction of the target ultrasound image.

[0047] The above illustrates one embodiment of determining candidate line segments and their corresponding connections when fitting a depth ladder model. In specific implementation, after determining each candidate line segment, the depth index on the scan line corresponding to each candidate line segment and the amplitude feature corresponding to each depth index can be obtained, which will not be elaborated here.

[0048] As explained above, the constructed stepped amplitude-depth fitting line should be consistent with... Figures 5-8 Similar to the example shown, the scan line amplitude characteristics of the current reference scan line can be determined based on the stepped amplitude-depth fitting line. For example, along the depth direction of the target ultrasound image, it can be determined whether the scan line amplitude characteristics show a gradually decreasing trend. In specific implementation, the amplitude state corresponding to the candidate line segment can be extracted to determine whether the amplitude shows a gradually decreasing trend. It should be understood that when the scan line amplitude characteristics show a gradually decreasing trend, the scan line depth sub-feature corresponding to the scan line amplitude characteristics is a valid feature. At this time, the reverberation noise feature state of the current reference scan line is configured to have reverberation noise characteristics. Specifically, the valid feature is the scan line depth sub-feature that can be used to determine whether it has reverberation noise characteristics.

[0049] In one embodiment of the present invention, the scan line depth feature further includes, respectively, a length feature between abrupt change points, a step morphology similarity feature, and / or abrupt change point index depth feature, which are respectively considered as sub-features of scan line depth. When there are multiple scan line depth sub-features within the scan line depth feature, the reverberation noise feature state of the current reference scan line will be configured to have reverberation noise features only when all scan line depth sub-features are valid features. The length feature between abrupt change points characterizes the length between two adjacent amplitude abrupt change points. When the length feature between abrupt change points is not greater than the length of the candidate line segment connected to the amplitude abrupt change point at the previous depth position, the scan line depth sub-feature corresponding to the length feature between abrupt change points is a valid feature. The step shape similarity feature characterizes the shape similarity between the step amplitude-depth fitting line and different preset reverberation step models. When there is a shape similarity greater than the shape similarity threshold, the scan line depth sub-feature corresponding to the step shape similarity feature is a valid feature. The mutation point index depth feature represents the index depth of each mutation point in the depth direction of the target ultrasound image. When the index depth of all mutation points matches the preset index depth, the scan line depth sub-feature corresponding to the mutation point index depth feature is a valid feature.

[0050] To improve the accuracy of reverberation noise feature identification, the scan line depth feature can include multiple scan line depth sub-features. When multiple scan line depth sub-features exist, the current reference scan line's reverberation noise feature state can only be configured to have reverberation noise characteristics when all scan line depth sub-features are valid. The meaning of valid features can be referred to the above explanation regarding scan line amplitude features. In specific implementation, based on the characteristics of the stepped amplitude-depth fitting line, when the scan line depth feature includes multiple scan line depth sub-features, the corresponding scan line depth sub-features can be the length feature between abrupt change points, the step shape similarity feature, and / or the abrupt change point index depth feature. When higher reverberation noise feature identification accuracy is required, the scan line depth feature preferably includes scan line depth sub-features, the length feature between abrupt change points, the step shape similarity feature, and the abrupt change point index depth feature.

[0051] Specifically, the length feature between abrupt change points characterizes the length between two adjacent amplitude abrupt change points, which is the length of the candidate line segment formed by connecting the first and last amplitude abrupt change points mentioned above. The method for determining the corresponding length can be found in the above explanation. The previous depth position specifically refers to the depth position corresponding to the first amplitude abrupt change point. Generally, the candidate line segment formed by connecting the first and last amplitude abrupt change points represents the depth position where reverberation noise may exist. For example, when a stepped amplitude-depth fitting line can be formed... Figure 5 In the diagram, the line segments corresponding to the reverberation noise are the candidate line segments formed by connecting the first and last amplitude abrupt change points. Therefore, the candidate line segment connecting the amplitude abrupt change point at the previous depth position is... Figure 5 The candidate line segments corresponding to strong boundaries are marked in the text; other cases can be referred to here for explanation. As can be seen from the above description of the configuration method for each candidate line segment, the length of the corresponding candidate line segment can be determined. Subsequently, the line segment lengths can be directly compared to determine whether the scan line depth sub-feature corresponding to the length feature between abrupt change points is a valid feature.

[0052] Depend on Figures 5-8It is known that the reverberation noise in different target ultrasound images varies. Therefore, the shape of the stepped amplitude-depth fitting line obtained by fitting the depth step model will differ from the actual image. In practice, a preset reverberation step model library can be constructed. This library includes several preset reverberation step models, which can be constructed in accordance with the above method. Preset reverberation step models should generally be based on ultrasound images of the same type as the target ultrasound image. Specifically, "same type of ultrasound image" refers to images generated by ultrasound scanning the same area using an ultrasound probe in the same working state. Alternatively, computer simulation can be used to generate corresponding preset reverberation step models. The generation method of preset reverberation step models can be selected as needed, and will not be illustrated here.

[0053] In practice, the morphological similarity between the stepped amplitude-depth fitting line and different preset reverberation stepped models is calculated, and the calculated morphological similarity is compared with the morphological similarity threshold. When there is a morphological similarity greater than the morphological similarity threshold, the scan line depth sub-feature corresponding to the stepped morphological similarity feature is a valid feature. Specifically, the morphological similarity threshold can be selected as 0.8. Of course, the corresponding morphological similarity threshold can also be selected according to the recognition accuracy of the reverberation noise region.

[0054] In one embodiment of the present invention, calculating morphological similarity includes: Calculate the dot product ratio between the stepped amplitude-depth fitting line and each preset reverberation step model, and use the calculated dot product ratio as the morphological similarity between the stepped amplitude-depth fitting line and the current preset reverberation step model. The preset reverberation ladder model is a preset amplitude-depth fitting line that characterizes the distribution of reverberation noise in ultrasound images.

[0055] One feasible method for calculating the dot product ratio is as follows:

[0056] in, To predetermine the dot product ratio between the reverberation step model and the step-shaped amplitude-depth fitting line, For the preset reverberation ladder model, The amplitude-depth fitting line is stepped. For the first reverberation step model within the preset reverberation step model The magnitude feature corresponding to each depth index. The step-like amplitude-depth fitting line is the first The magnitude feature corresponding to each depth index The number of depth indices within the stepped amplitude-depth fitting line.

[0057] In practice, when calculating the dot product ratio, the number of corresponding depth indices and the amplitude features corresponding to each depth index can be determined by referring to the above method. After calculating the dot product ratio using the above method, the morphological similarity between the stepped amplitude-depth fitting line and each preset reverberation step model can be obtained.

[0058] In one embodiment of the present invention, when the reverberation noise is first-order reverberation, the index depth of the first amplitude abrupt change point along the depth direction of the target ultrasound image is less than 1 / 3 of the total depth index of the target ultrasound image, and at the same time, the index depth of the second amplitude abrupt change point is less than 1 / 2 of the total depth index of the target ultrasound image.

[0059] Specifically, the first amplitude mutation point is the aforementioned initial amplitude mutation point, and the second amplitude mutation point is the aforementioned tail amplitude mutation point. The total depth index of the target ultrasound image can be determined using techniques commonly used in this technical field, and will not be elaborated here.

[0060] In one embodiment of the present invention, the lateral feature recognition and detection includes: Along the arrangement order of the target scan lines on the target ultrasound image, the lateral continuity characteristics of two adjacent target scan lines are determined sequentially. The target scan line is a reference scan line with reverberant noise characteristics. The lateral continuity characteristics characterize the coherence of the reverberant noise characteristics of the two target scan lines. Based on the lateral continuity characteristics of two adjacent target scan lines, all target scan lines that satisfy the lateral continuity distribution trend are identified, and the region where the identified target scan lines are located is configured as the region location of the reverberation noise region.

[0061] As explained above, after depth feature recognition detection, regions with reverberant noise characteristics within the target ultrasound image can be identified. However, the exact location of these reverberant noise regions cannot be determined at this stage. To improve the accuracy of identifying the location of reverberant noise regions, lateral feature recognition detection should be performed. Therefore, lateral feature recognition detection can specifically determine the location of the reverberant noise regions.

[0062] It should be noted that the reverberation in the horizontal direction usually exhibits a similar stepped pattern between two adjacent scan lines. Figure 4 The image illustrates one embodiment of the lateral reverberation behavior between two adjacent scan lines, where color and width represent the corresponding amplitude. Therefore, during lateral feature recognition and detection, the lateral continuity characteristics of two adjacent target scan lines are primarily determined, with the target scan line representing the reverberant noise characteristic state and the reference scan line exhibiting reverberant features. The arrangement order of the target scan lines in the target ultrasound image should be consistent with the method of selecting the reference scan line, such as... Figure 2In the diagram, you can follow either the left-to-right or right-to-left order, depending on your needs.

[0063] In practice, after determining the lateral continuity characteristics of two adjacent target scan lines, it can be determined whether the lateral reverberation performance is satisfied. Subsequently, by analyzing and determining the lateral continuity distribution trend of all target scan lines, the regional location of the reverberation noise area can be determined.

[0064] In one embodiment of the present invention, determining the lateral continuity feature of two adjacent target scan lines includes: Based on the stepped amplitude-depth fitting line corresponding to each target scan line, at least the line segment overlap features and line segment similarity features of each candidate line segment within two target scan lines are determined. When both the line segment overlap feature and the line segment similarity feature are valid features, the lateral continuity feature of the two adjacent target scan lines is lateral continuity. Along the arrangement direction of the target scan line on the target ultrasound image, when the determined lateral continuity features are successively lateral continuity, or the lateral continuity features of lateral continuity dominate, then it is determined that the lateral continuity distribution trend is satisfied.

[0065] As explained above, for each target scan line, a corresponding stepped amplitude-depth fitting line can be determined. Subsequently, in the order of the depth direction, the candidate line segments corresponding to two adjacent target scan lines can be determined. Then, the line segment overlap of each corresponding candidate line segment can be calculated. For example, the line segment overlap of the two candidate line segments ranked first is calculated. After that, the line segment overlap of the two corresponding candidate line segments ranked second is calculated. When there are multiple candidate line segments, they are sorted in sequence and the corresponding line segment overlap is calculated.

[0066] When calculating line segment overlap, the main calculation involves dividing the overlap depth index of two candidate line segments by the total number of sorted line segment indices. In practice, the overlap depth index can be determined based on the depth index of each candidate line segment. The total number of sorted line segment indices is generally taken as the number of depth indices present on the two candidate line segments. After calculating the line segment overlap for each candidate line segment, the calculated overlap should be compared with a line segment overlap threshold. If the overlap for all candidate line segments corresponding to two adjacent target scan lines is not less than the line segment overlap threshold, then the line segment overlap feature is considered a valid feature; otherwise, it is considered an invalid feature. Generally, the line segment overlap threshold can be 0.8. Of course, other values ​​can be selected as needed, based on the requirement of effectively performing lateral feature recognition and detection.

[0067] When determining line segment similarity features, the main focus is on determining the line segment similarity of each candidate line segment. The line segment similarity of each candidate line segment can be calculated using the corresponding dot product ratio mentioned above. The difference is that the dot product ratio of the corresponding ranked candidate line segments is directly calculated, thus obtaining the corresponding line segment similarity. Here, "ranking" specifically refers to the arrangement order of candidate line segments along the same or opposite depth direction within the stepped amplitude-depth fitting line. Refer to the illustrations and corresponding explanations for details. After calculating the dot product ratios of two candidate line segments with the same ranking, these ratios should be compared with the line segment similarity threshold. If the dot product ratios of all ranked candidate line segments are not less than the corresponding line segment similarity threshold, then the line segment similarity feature is a valid feature; otherwise, the line segment similarity feature is an invalid feature. The line segment similarity threshold can be set to 0.8. Of course, other values ​​can be selected as needed, based on the requirement of effectively performing lateral feature recognition and detection.

[0068] In addition, when determining whether the lateral continuity feature of two adjacent target scan lines is laterally continuous, the number of candidate line segments can be counted. If the total number of candidate line segments that simultaneously meet the line segment overlap feature and line segment similarity feature are not less than 5, then the lateral continuity feature of two adjacent target scan lines can be considered to be laterally continuous, so as to further improve the accuracy of lateral feature recognition and detection.

[0069] When the lateral continuity feature of two adjacent target scan lines is laterally continuous, but the lateral continuity feature of two subsequent adjacent target scan lines is not laterally continuous, the distribution position of the previously laterally continuous target scan lines can be configured as the regional position of the reverberation noise region. To avoid inaccurate identification of the regional position of the reverberation noise region, the region position of the reverberation noise region can be determined when the lateral continuity feature is dominant. Specifically, the lateral continuity feature being dominant means that there are two adjacent target scan lines whose lateral continuity feature is not laterally continuous, but there are still multiple target scan lines with lateral continuity features. Of course, other dominant situations are also possible, and the specific choice can be made according to the needs, based on the ability to effectively determine the regional position of the reverberation noise region.

[0070] In one embodiment of the present invention, when smoothing the amplitude of each reverberation suppression target segment, the method includes smoothing the amplitude of the reverberation suppression target segment by setting it to zero, smoothing the amplitude of the reverberation suppression target segment by linearly decreasing it, or smoothing the amplitude of the reverberation suppression target segment by curve decreasing it.

[0071] It should be noted that the reverberation suppression target segment can generally be the candidate line segment between two adjacent amplitude abrupt change points mentioned above. Of course, after determining the location of the reverberation noise region, suppression lines corresponding to or partially corresponding to the scan lines can be selected within the region where the reverberation noise region is located. Based on the depth of the reverberation noise region, the reverberation suppression target segment of the suppression line can be determined. The choice of the reverberation suppression target segment can be made as needed, and will not be listed here. It is understood that each suppression line can be generated from one scan line, and the number of suppression lines can be the same as or less than the number of scan lines. When the number of suppression lines is less than the number of scan lines, partial reverberation suppression can be achieved within the reverberation noise region. The number of suppression lines can be selected and determined according to the reverberation suppression requirements, and will not be listed here.

[0072] The smoothing method can be selected as needed. Figure 9 An embodiment is shown in which the amplitude of the reverberation suppression target segment is smoothed by setting it to zero. Figure 10 The image shows an embodiment in which the amplitude at the reverberation suppression target end is smoothed in a linear manner. Figure 11 The diagram shows how the amplitude of the reverberation suppression target segment is smoothed by decreasing the curve. It is understandable that... Figures 9-11 This example only illustrates smoothing within the stepped amplitude-depth fitting line. In practice, smoothing can be performed according to the corresponding depth index and amplitude of the reverberation suppression target segment, following the aforementioned smoothing method. After smoothing, the target ultrasound image can be updated. Furthermore, the smoothed reverberation suppression target segment area can also undergo two-dimensional smoothing filtering, such as using a two-dimensional Gaussian filter kernel or other common low-pass filtering methods like median filtering, to naturally smooth the image in the depth and lateral directions.

Claims

1. An ultrasonic reverberation suppression method based on step energy difference, characterized in that, The ultrasonic reverberation suppression method includes: Provide the target ultrasound image for reverberation suppression; Based on the step energy difference caused by reverberation noise in the depth direction of the target ultrasound image, the reverberation noise region in the target ultrasound image is identified and detected. When the reverberation noise region is identified in the target ultrasound region, the regional location of the reverberation noise region in the target ultrasound image is determined. Based on the location of the identified reverberant noise region, reverberation suppression is performed on the reverberant noise region within the target ultrasound image. The target ultrasound image is then updated after reverberation suppression. During reverberation suppression, several reverberation suppression target segments are extracted within the reverberation noise region along the depth direction of the target ultrasound image. Subsequently, the amplitude of each reverberation suppression target segment is smoothed to denoise the reverberation noise region within the target ultrasound image.

2. The ultrasonic reverberation suppression method based on step energy difference according to claim 1, characterized in that, When identifying and detecting reverberant noise regions within a target ultrasound image, the process includes first performing depth feature recognition and detection on the target ultrasound image, wherein... When reverberant noise is determined to exist through deep feature recognition detection, lateral feature recognition detection is then performed on the target ultrasound image. After lateral feature recognition and detection, the location of the reverberant noise region within the target ultrasound image is identified and determined.

3. The ultrasonic reverberation suppression method based on step energy difference according to claim 2, characterized in that, When performing deep feature recognition and detection, the following are included: Several reference scan lines are selected on the target ultrasound image, and each reference scan line extends along the depth direction of the target ultrasound image. For each reference scan line, the scan line depth feature of the reference scan line is extracted, and the reverberation noise feature state of each reference scan line is determined. When at least two reference scan lines exhibit reverberation noise characteristics, the presence of reverberation noise is confirmed after depth feature recognition detection.

4. The ultrasonic reverberation suppression method based on step energy difference according to claim 3, characterized in that, The scan line depth feature includes several scan line depth sub-features, wherein the scan line depth sub-features include at least the scan line amplitude feature; When extracting the scan line depth features for each reference scan line, the following is included: Differentiate the amplitude characteristics of each reference scan line to obtain the amplitude abrupt change point of the current reference scan line; Based on all acquired amplitude abrupt change points, a depth step model is fitted to generate a step-like amplitude-depth fitting line for the current reference scan line, where... The stepped amplitude-depth fitting line includes several candidate line segments connected in sequence. Within the stepped amplitude-depth fitting line, the horizontal axis represents the depth of each candidate line segment within the target ultrasound image, and the vertical axis represents the amplitude of each candidate line segment within the target ultrasound image. Based on the stepped amplitude-depth fitting line, the scan line amplitude characteristics of the current reference scan line are determined; Along the depth direction of the target ultrasound image, when the amplitude feature of the scan line gradually decreases, the scan line depth sub-feature corresponding to the scan line amplitude feature is a valid feature. At this time, the reverberation noise feature state of the current reference scan line is configured to have reverberation noise feature.

5. The ultrasonic reverberation suppression method based on step energy difference according to claim 4, characterized in that, The scan line depth feature also includes, as sub-features of scan line depth, the length feature between abrupt change points, the step morphology similarity feature, and / or the depth feature of abrupt change point index, wherein... When there are multiple scan line depth sub-features within the scan line depth feature, the reverberation noise feature state of the current reference scan line will be configured to have reverberation noise features only when all scan line depth sub-features are valid features. The length feature between abrupt change points characterizes the length between two adjacent amplitude abrupt change points. When the length feature between abrupt change points is not greater than the length of the candidate line segment connected to the amplitude abrupt change point at the previous depth position, the scan line depth sub-feature corresponding to the length feature between abrupt change points is a valid feature. The step shape similarity feature characterizes the shape similarity between the step amplitude-depth fitting line and different preset reverberation step models. When there is a shape similarity greater than the shape similarity threshold, the scan line depth sub-feature corresponding to the step shape similarity feature is a valid feature. The mutation point index depth feature represents the index depth of each mutation point in the depth direction of the target ultrasound image. When the index depth of all mutation points matches the preset index depth, the scan line depth sub-feature corresponding to the mutation point index depth feature is a valid feature.

6. The ultrasonic reverberation suppression method based on step energy difference according to claim 4, characterized in that, When calculating morphological similarity, the following are included: Calculate the dot product ratio between the stepped amplitude-depth fitting line and each preset reverberation step model, and use the calculated dot product ratio as the morphological similarity between the stepped amplitude-depth fitting line and the current preset reverberation step model. The preset reverberation ladder model is a preset amplitude-depth fitting line that characterizes the distribution of reverberation noise in ultrasound images.

7. The ultrasonic reverberation suppression method based on step energy difference according to claim 5, characterized in that, When the reverberation noise is first-order reverberation, the index depth of the first amplitude abrupt change point along the depth direction of the target ultrasound image is less than 1 / 3 of the total depth index of the target ultrasound image, and the index depth of the second amplitude abrupt change point is less than 1 / 2 of the total depth index of the target ultrasound image.

8. The ultrasonic reverberation suppression method based on step energy difference according to any one of claims 2 to 7, characterized in that, Lateral feature recognition and detection includes: Along the arrangement order of the target scan lines on the target ultrasound image, the lateral continuity characteristics of two adjacent target scan lines are determined sequentially. The target scan line is a reference scan line with reverberant noise characteristics. The lateral continuity characteristics characterize the coherence of the reverberant noise characteristics of the two target scan lines. Based on the lateral continuity characteristics of two adjacent target scan lines, all target scan lines that satisfy the lateral continuity distribution trend are identified, and the region where the identified target scan lines are located is configured as the region location of the reverberation noise region.

9. The ultrasonic reverberation suppression method based on step energy difference according to claim 8, characterized in that, Determining the lateral continuity characteristics of two adjacent target scan lines includes: Based on the stepped amplitude-depth fitting line corresponding to each target scan line, at least the line segment overlap features and line segment similarity features of each candidate line segment within two target scan lines are determined. When both the line segment overlap feature and the line segment similarity feature are valid features, the lateral continuity feature of the two adjacent target scan lines is lateral continuity. Along the direction of the target scan line on the target ultrasound image, when the determined lateral continuity features are successively lateral continuity, or the lateral continuity features of lateral continuity dominate, then it is determined that the lateral continuity distribution trend is satisfied.

10. The ultrasonic reverberation suppression method based on step energy difference according to any one of claims 1 to 7, characterized in that, When smoothing the amplitude of each reverberation suppression target segment, the methods include smoothing the amplitude of the reverberation suppression target segment by setting it to zero, smoothing the amplitude of the reverberation suppression target segment by linearly decreasing it, or smoothing the amplitude of the reverberation suppression target segment by curve-like decreasing it.

Images