Transcranial ultrasound image adaptive optimization method and system based on noise form

Abstract

The invention relates to a self-adaptive optimization method and system, in particular to a transcranial ultrasound image self-adaptive optimization method and system based on a noise form. According to the technical scheme provided by the invention, the transcranial ultrasound image adaptive optimization method based on the noise form comprises the following steps: acquiring a transcranial ultrasound image generated through craniocerebral ultrasound, and at least extracting the noise form of noise in the transcranial ultrasound image; selecting a target skull model corresponding to the extracted noise form based on the extracted noise form; and correcting the obtained transcranial ultrasound image by using a target skull model so as to de-noise the transcranial ultrasound image and generate a target transcranial ultrasound image. According to the method, the imaging quality of the transcranial ultrasound image can be adaptively optimized, and the adaptability of craniocerebral ultrasound is improved.

Description

Technical Field

[0001] This invention relates to an adaptive optimization method and system, and more particularly to an adaptive optimization method and system for transcranial ultrasound images based on noise morphology. Background Technology

[0002] Transcranial ultrasound (TCE) plays an irreplaceable role in neonatal intensive care, intraoperative navigation, and cerebrovascular assessment due to its unique advantages such as being radiation-free, portable, and real-time. TCE involves scanning the skull with ultrasound and generating transcranial images, which provide a non-invasive view of the brain's anatomical structure.

[0003] Due to the presence of the skull as a strong reflective layer, ultrasound waves are partially reflected by the skull. These reflected sound waves bounce back and forth between the ultrasound probe and the skull, producing multiple reflections. In transcranial ultrasound images, this manifests as multiple layers of cloud-like noise below the near-field skull layer. For ultrasound waves penetrating the skull, the significant difference between the sound velocity and acoustic impedance of the skull and that of conventional tissues leads to signal distortion, resulting in poor focusing of the transcranial ultrasound image and affecting diagnosis.

[0004] For the reasons mentioned above, transcranial ultrasound has not yet been as widely used as ultrasound of organs such as the abdomen and superficial organs. To improve the effectiveness of transcranial ultrasound, a pre-set skull model is typically used. This model has fixed parameters for skull thickness, curvature, and sound velocity. This model is then used to compensate for and correct the ultrasound signal delay, thereby enhancing the quality of transcranial ultrasound images. However, a skull model with fixed parameters cannot consistently yield optimal results when dealing with different patients and different acoustic windows of the skull, affecting the accuracy of the doctor's diagnosis. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method and system for adaptive optimization of transcranial ultrasound images based on noise morphology, which can adaptively optimize the imaging quality of transcranial ultrasound images and improve the adaptability of cranial ultrasound.

[0006] According to the technical solution provided by the present invention, a transcranial ultrasound image adaptive optimization method based on noise morphology is provided, the transcranial ultrasound image adaptive optimization method comprising: Acquire transcranial ultrasound images generated by transcranial ultrasound, and at least extract the noise morphology of noise present in the transcranial ultrasound images; Based on the extracted noise pattern, select the target skull model corresponding to the extracted noise pattern; The transcranial ultrasound images obtained above are corrected using a target skull model to denoise the transcranial ultrasound images and generate the target transcranial ultrasound image.

[0007] The extracted noise morphology is at least the noise morphology of near-field cloud-like noise in transcranial ultrasound images, wherein, The noise pattern includes at least one of the following: noise region depth, noise region gray-scale mean, and noise region gray-scale variance. When the noise pattern includes the depth of the noise region, select the target skull thickness corresponding to the depth of the noise region. When the noise pattern includes the average gray value of the noise region, select the target skull sound velocity corresponding to the average gray value of the noise region; When the noise pattern includes the gray-level variance of the noise region, select the target skull curvature corresponding to the gray-level variance of the noise region; Based on the selected target skull thickness, target skull sound velocity, and / or target skull curvature, a target skull model corresponding to the noise morphology is formed.

[0008] When extracting the noise morphology of near-field cloud-like noise in transcranial ultrasound images, it includes: Transcranial ultrasound images are preprocessed to remove texture information in order to generate preprocessed transcranial images; The near-field cloud-like noise region in the transcranial preprocessed image is segmented and extracted to obtain the target near-field cloud-like noise region; The target near-field cloud-like noise region obtained by the above segmentation shall be subjected to at least a noise pattern calculation to generate a noise pattern corresponding to the target near-field cloud-like noise region.

[0009] Preprocessing transcranial ultrasound images to remove texture information includes at least low-pass filtering of the transcranial ultrasound images, wherein... When performing low-pass filtering, the low-pass filtering method includes one of two-dimensional Gaussian filtering, median filtering, or Butterworth low-pass filtering.

[0010] When segmenting and extracting near-field cloud-like noise regions within a transcranially processed image, the segmentation and extraction method includes a two-dimensional OTU segmentation method, wherein... When using the two-dimensional OTU segmentation method for segmentation and extraction, the grayscale threshold of the target segmentation pixel is determined. The transcranial processed image is segmented using the determined target segmentation pixel grayscale threshold to obtain the target near-field cloud-like noise region.

[0011] Determining the grayscale threshold of the target segmented pixels includes: Construction and reference grayscale threshold u The corresponding grayscale threshold filtering expression is:

[0012] in, To filter reference values, For pixels with gray levels less than a reference gray level threshold u The probability, For pixel grayscale values ​​not less than the reference grayscale threshold u The probability, For pixels with gray levels less than a reference gray level threshold u The average gray level, For pixel grayscale not less than the reference threshold u The average gray level; Iterate through the reference grayscale thresholds, and when the maximum reference value is found, set the corresponding reference grayscale threshold. u Configure the grayscale threshold as the target segmentation pixel.

[0013] Before calculating the noise morphology of the target's near-field cloud-like noise region, noise reduction optimization processing is also performed on the target's near-field cloud-like noise region, including... When performing noise reduction optimization, at least the morphological opening operation is performed on the target's near-field cloud-like noise region.

[0014] After selecting the target skull model, imaging correction processing is performed, including... Imaging correction processing includes: The beamforming delay used in acquiring the transcranial ultrasound images was corrected using the target skull model to update the generated target transcranial ultrasound images.

[0015] After extracting the noise morphology of noise present in transcranial ultrasound images, the following is also included: The excitation voltage of the ultrasonic probe unit is modulated based on noise morphology to generate the desired excitation voltage. Transcranial ultrasound scanning is performed by driving an ultrasound probe unit under a desired excitation voltage to optimize the noise morphology of the generated transcranial ultrasound images.

[0016] A noise morphology-based adaptive optimization system for transcranial ultrasound images includes a transcranial ultrasound scanning device, wherein the transcranial ultrasound scanning device adaptively optimizes the generated transcranial ultrasound images using the method described above.

[0017] The advantages of this invention are as follows: During adaptive optimization, the noise morphology present in the transcranial ultrasound image should be extracted first, and a target skull model should be selected based on the extracted noise morphology. Subsequently, the selected target skull model is used to correct the transcranial ultrasound image. Compared with the existing correction using a skull model with fixed parameters, since the selected target skull model corresponds to the extracted noise morphology, the transcranial ultrasound image can be effectively corrected using the target skull model, thereby improving the quality of the generated transcranial ultrasound image and enhancing the adaptability of cranial ultrasound. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of an embodiment of the present invention for adaptive optimization of transcranial ultrasound images.

[0019] Figure 2 This is a schematic diagram of one embodiment of transcranial ultrasound imaging.

[0020] Figure 3 This is a schematic diagram of an embodiment of the present invention obtained through segmentation and extraction.

[0021] Figure labeling: 100 - Transcranial ultrasound image, 110 - Near-field cloud-like noise, 1210 - Far-field distortion and poor focusing area. Detailed Implementation

[0022] The present invention will be further described below with reference to specific accompanying drawings and embodiments.

[0023] To adaptively optimize the imaging quality of transcranial ultrasound images 100, this invention provides an adaptive optimization method for transcranial ultrasound images 100 based on noise morphology. Specifically, the adaptive optimization method for transcranial ultrasound images 100 includes: Acquire a transcranial ultrasound image 100 generated by transcranial ultrasound, and at least extract the noise morphology of the noise present in the transcranial ultrasound image 100. Based on the extracted noise pattern, select the target skull model corresponding to the extracted noise pattern; The transcranial ultrasound image 100 obtained above is corrected using the target skull model in order to denoise the transcranial ultrasound image 100 and generate the target transcranial ultrasound image.

[0024] Figure 2 An embodiment of a transcranial ultrasound image 100 is illustrated. As can be seen from the background art and the illustrations, the transcranial ultrasound image 100 generally contains near-field cloud-like noise 110 and far-field distortion and poor focusing regions 120. The mechanism of the formation of near-field cloud-like noise 110 and far-field distortion and poor focusing regions 120 can be referred to the corresponding descriptions above. It is understood that the adaptive optimization of the present invention mainly aims to improve the imaging quality of the transcranial ultrasound image 100.

[0025] Figure 1The diagram illustrates an embodiment of the adaptive optimization of the present invention. As shown in the diagram, during adaptive optimization, the noise patterns present in the transcranial ultrasound image 100 are first extracted, and a target skull model is selected based on the extracted noise patterns. Subsequently, the selected target skull model is used to correct the transcranial ultrasound image 100. It should be understood that the method and process of correcting the transcranial ultrasound image 100 using the target skull model after selection can be consistent with existing technologies. After correcting the transcranial ultrasound image 100, denoising of the transcranial ultrasound image 100 can be achieved, thereby obtaining a target transcranial ultrasound image with better image quality.

[0026] It should be understood that the adaptive optimization method of the present invention is mainly based on the selection of the target skull model based on the extracted noise morphology, rather than the use of a skull model with fixed parameters as in the prior art. Since the selected target skull model corresponds to the extracted noise morphology, the target skull model can be used to effectively correct the transcranial ultrasound image 100.

[0027] In practice, after selecting the target skull model, in order to achieve 100% correction of the transcranial ultrasound images, imaging correction processing should generally be performed. This imaging correction processing includes: The beamforming delay used to acquire the transcranial ultrasound image 100 was corrected using the target skull model to update the generated target transcranial ultrasound image.

[0028] Specifically, the imaging correction process mainly corrects the beamforming delay used to generate the transcranial ultrasound image 100. The correction method for model-related changes can employ existing methods, such as using a phase screen combined with Snell's law. The method and process of model-related change correction can be consistent with existing techniques. Furthermore, other methods for correcting model-related changes can be selected as needed, and will not be elaborated here. As can be seen from the above adaptive optimization description, the adaptive optimization method of this invention can be applied to different transcranial ultrasound imaging modalities, such as grayscale imaging, Doppler flow imaging, and super-resolution microscopy, thereby improving the adaptability of cranial ultrasound.

[0029] In one embodiment of the present invention, the extracted noise morphology is at least the noise morphology of near-field cloud-like noise 110 within the transcranial ultrasound image 100, wherein, The noise pattern includes at least one of the following: noise region depth, noise region gray-scale mean, and noise region gray-scale variance. When the noise pattern includes the depth of the noise region, select the target skull thickness corresponding to the depth of the noise region; When the noise pattern includes the average gray value of the noise region, select the target skull sound velocity corresponding to the average gray value of the noise region; When the noise pattern includes the gray-level variance of the noise region, select the target skull curvature corresponding to the gray-level variance of the noise region; Based on the selected target skull thickness, target skull sound velocity, and / or target skull curvature, a target skull model corresponding to the noise morphology is formed.

[0030] To effectively select the target skull model, this invention primarily extracts the noise morphology of near-field cloud-like noise 100 within the transcranial ultrasound image 100. Specifically, the extracted noise morphology is at least one of the following: noise region depth, noise region mean gray value, and noise region gray value variance. For example, the extracted noise morphology can be at least the noise region depth. Alternatively, it can simultaneously include noise region depth, noise region mean gray value, and noise region gray value variance. The content of the noise morphology can be selected as needed, prioritizing the selection of the desired target skull model. Generally, if the noise morphology has only one noise feature, noise region depth should be preferred.

[0031] Figure 3 The figure shows one embodiment of the noise region depth, where d represents the noise region depth. Figure 3 The upper-middle fan-shaped area is the region where the near-field cloud-like noise 110 is located. The depth of the noise region is the depth of the near-field cloud-like noise 100 within the transcranial ultrasound image 100. The gray-scale mean of the noise region specifically refers to the mean of the pixel gray-scale within the region where the near-field cloud-like noise 100 is located. The gray-scale variance of the noise region specifically refers to the variance of the pixel gray-scale within the region where the near-field cloud-like noise 100 is located. The method for extracting the noise morphology will be explained in detail below.

[0032] When the noise morphology includes a noise region depth, the corresponding target skull thickness should be selected based on the noise region depth. This means selecting the skull thickness within the target skull model and setting that thickness as the target skull thickness. In practice, skull models with different skull thicknesses can be constructed, and a reference skull thickness can be determined. The reference skull thickness can be an existing default skull thickness, such as a skull thickness with fixed parameters. In this case, the reference noise region depth can be determined. It should be understood that, given the reference noise region depth and reference skull thickness, the quality of the generated skull ultrasound image 100 should at least meet the requirements for ultrasound image diagnostic analysis.

[0033] During adaptive optimization, the depth of the noise region is compared with the depth of a reference noise region, and the corresponding target skull thickness is selected based on the comparison result. Generally, when the noise region depth is greater than the reference noise region depth, the selected target skull thickness should be greater than the reference skull thickness; conversely, when the noise region depth is less than the reference noise region depth, the selected target skull thickness should be less than the reference skull thickness. In practice, data fitting can be performed using multiple skull depths, noise region depths, and the image quality of the transcranial ultrasound image 100 to determine the relationship between the noise region depth and the selected target skull thickness. The specific method for fitting and selecting the target skull thickness should meet the requirements described here. The specific data fitting method can be selected as needed and will not be elaborated further here.

[0034] In ultrasound scanning, the greater the difference in ultrasonic impedance, the stronger the reflection, the brighter the signal, and the higher the grayscale value. In one embodiment of the present invention, when the noise pattern includes the average grayscale value of the noise region, the corresponding target skull sound velocity can be selected based on the average grayscale value of the noise region. The method of selecting the target skull sound velocity based on the average grayscale value of the noise region can refer to the above description of selecting the target skull thickness based on the depth of the noise region. Specifically, when the average grayscale value of the noise region is greater than the average grayscale value of the reference noise region, the selected target skull sound velocity should be greater than the reference skull sound velocity; conversely, when the average grayscale value of the noise region is less than the average grayscale value of the reference noise region, the selected target skull sound velocity should be less than the reference skull sound velocity. The meaning of the reference noise region average grayscale value and the reference skull sound velocity, as well as the specific selection method, can be found in the above description and will not be repeated here.

[0035] Skull curvature affects the ultrasound propagation path, including the ultrasound incident and exit directions and path length, ultimately leading to poor ultrasound focusing. This results in reduced in-phase enhancement, increased phase difference signal cancellation positions, and increased gray-scale variance. In specific implementation, when the noise pattern includes the gray-scale variance of the noise region, the target skull curvature is selected based on the gray-scale variance of the noise region. The method for selecting the target skull curvature based on the gray-scale variance of the noise region can be referenced in the above explanation of selecting the target skull thickness based on the noise region depth. Specifically, when the gray-scale variance of the noise region is greater than the gray-scale variance of the reference noise region, the selected target skull curvature should be greater than the reference skull curvature; conversely, when the gray-scale variance of the noise region is less than the gray-scale variance of the reference noise region, the selected target skull curvature should be less than the reference skull curvature. The meaning of the reference noise region gray-scale variance and the reference skull curvature, as well as their specific selection methods, can be found in the above explanation and will not be repeated here.

[0036] After selecting the target skull thickness, target skull sound velocity, and / or target skull curvature, a corresponding target skull model can be selected. In practice, a skull model library can be constructed, containing skull models with different skull thicknesses, skull sound velocities, and combinations of skull regions. After determining the target skull thickness, target skull sound velocity, and / or target skull curvature, a corresponding skull model can be selected from the skull model library in the manner described above, and used as the target skull model. As can be seen from the above description of the method for selecting the target skull model, using the target skull model can effectively correct the transcranial ultrasound image 100, improving the imaging quality of the obtained target skull image.

[0037] In one embodiment of the present invention, extracting the noise morphology of near-field cloud-like noise 110 within a transcranial ultrasound image 100 includes: The transcranial ultrasound image 100 is preprocessed to remove texture information in order to generate a preprocessed transcranial image; The near-field cloud-like noise region in the transcranial preprocessed image is segmented and extracted to obtain the target near-field cloud-like noise region; The target near-field cloud-like noise region obtained by the above segmentation shall be subjected to at least a noise pattern calculation to generate a noise pattern corresponding to the target near-field cloud-like noise region.

[0038] To improve the accuracy of noise morphology extraction, the transcranial ultrasound image 100 should undergo preprocessing to remove texture information. In one embodiment of the present invention, the preprocessing to remove texture information from the transcranial ultrasound image 100 includes at least low-pass filtering of the transcranial ultrasound image 100, wherein... When performing low-pass filtering, the low-pass filtering method includes one of two-dimensional Gaussian filtering, median filtering, or Butterworth low-pass filtering. Specific low-pass filtering methods will not be listed here.

[0039] In one embodiment of the present invention, when segmenting and extracting near-field cloud-like noise regions in a transcranial processed image, the segmentation and extraction method includes a two-dimensional OTU segmentation method, wherein... When using the two-dimensional OTU segmentation method for segmentation and extraction, the grayscale threshold of the target segmentation pixel is determined. The transcranial processed image is segmented using the determined target segmentation pixel grayscale threshold to obtain the target near-field cloud-like noise region.

[0040] It is understandable that other gray-level connected component clustering segmentation methods can also be used during segmentation and extraction. When using the two-dimensional OTU segmentation method, the gray-level threshold of the target segmentation pixel should be determined first. In one embodiment of the present invention, determining the gray-level threshold of the target segmentation pixel includes: Construction and reference grayscale threshold uThe corresponding grayscale threshold filtering expression is:

[0041] in, To filter reference values, For pixels with gray levels less than a reference gray level threshold u The probability, For pixel grayscale values ​​not less than the reference grayscale threshold u The probability, For pixels with gray levels less than a reference gray level threshold u The average gray level, For pixel grayscale not less than the reference threshold u The average gray level; Iterate through the reference grayscale thresholds, and when the maximum reference value is found, set the corresponding reference grayscale threshold. u Configure the grayscale threshold as the target segmentation pixel.

[0042] It should be understood that for images preprocessed via the transcranial camera, the grayscale value of each pixel in the preprocessed image can be determined. After setting a reference grayscale threshold, the number of pixels whose grayscale value is less than the reference grayscale threshold can be counted. Then, dividing the counted number of pixels by the total number of pixels gives the probability that the pixel grayscale value is less than the reference grayscale threshold. Similarly, the number of pixels whose grayscale value is not less than the reference grayscale threshold can be counted. Then, dividing the counted number of pixels by the total number of pixels gives the probability that the pixel grayscale value is not less than the reference grayscale threshold.

[0043] In practice, the average gray level can be calculated from the gray levels of all pixels whose gray levels are less than the reference gray level threshold. Similarly, given that there are pixels whose grayscale values ​​are not less than a reference grayscale threshold, the average grayscale value can be calculated. Based on the grayscale distribution of pixels in the transcranial preprocessed image, multiple reference grayscale thresholds can be traversed, and the corresponding filtering reference values ​​can be calculated. When the filtering reference value is the largest, the corresponding reference grayscale threshold is configured as the grayscale threshold of the target segmentation pixel.

[0044] Once the target segmentation pixel grayscale threshold is determined, it can be used to segment the transcranially processed image to obtain the target near-field cloud-like noise region. Understandably, the method of segmenting the transcranially processed image using the target segmentation pixel grayscale threshold is consistent with existing technologies. Figure 3 In the image, the upper fan-shaped region is the segmented near-field cloud-like noise region of the target.

[0045] To improve the reliability of subsequent noise pattern extraction, in one embodiment of the present invention, before calculating the noise pattern of the target near-field cloud-like noise region, noise reduction optimization processing is performed on the target near-field cloud-like noise region. When performing noise reduction optimization, at least the morphological opening operation is performed on the target's near-field cloud-like noise region.

[0046] In practice, the morphological opening operation includes sequential erosion and dilation operations. This operation eliminates internal noise in the near-field cloud-like noise region of the target. Other noise reduction optimization methods can also be used, depending on the specific requirements.

[0047] In one embodiment of the present invention, after extracting the noise morphology of noise present in the transcranial ultrasound image 100, the method further includes: The excitation voltage of the ultrasonic probe unit is modulated based on noise morphology to generate the desired excitation voltage. Transcranial ultrasound scanning is performed by driving an ultrasound probe unit under a desired excitation voltage to optimize the noise morphology within the generated transcranial ultrasound image 100.

[0048] It should be noted that the noise pattern present in a transcranial ultrasound image of 100 pixels is generally related to the ultrasonic signal emitted by the ultrasound probe unit. Therefore, after extracting the noise pattern, the excitation voltage of the ultrasound probe unit can be adjusted according to the noise pattern. The ultrasound probe unit is the probe used to perform the transcranial ultrasound scan, and it can adopt commonly used existing forms. Generally, an excitation voltage is preset during a transcranial ultrasound scan. Based on the current noise pattern, the current excitation voltage can be adjusted to form the desired excitation voltage. Adjusting the current excitation voltage specifically refers to increasing or decreasing the excitation voltage, thereby forming the corresponding desired excitation voltage.

[0049] After generating the desired excitation voltage, the ultrasound probe unit can be driven to perform transcranial ultrasound scanning under the desired excitation voltage. It is understood that the generated desired excitation voltage should be based on optimizing the noise morphology within the transcranial ultrasound image 100, such as reducing the noise region depth, reducing the noise region gray mean, and / or reducing the noise region gray variance. Therefore, those skilled in the art can select the method of adjusting the excitation voltage according to the current noise morphology. Specific adjustment methods will not be illustrated here.

[0050] Based on the above description, a transcranial ultrasound image adaptive optimization system based on noise morphology can be obtained. In one embodiment of the present invention, a transcranial ultrasound scanning device is included, which uses the above method to adaptively optimize the generated transcranial ultrasound image.

[0051] It should be understood that the transcranial ultrasound scanning device can adopt existing common forms, such as including an ultrasound probe unit and a terminal device adapted and connected to the ultrasound probe unit. The terminal device can be a computer device, specifically designed to meet the requirements of generating transcranial ultrasound images 100 and performing the adaptive optimization processing described above.

Claims

1. A noise pattern based transcranial ultrasound image adaptive optimization method, characterized in that, The transcranial ultrasound image adaptive optimization method comprises: acquiring a transcranial ultrasound image generated by a transcranial brain ultrasound, and extracting at least a noise form of noise existing in the transcranial ultrasound image; based on the extracted noise form, selecting a target skull model corresponding to the extracted noise form; using the target skull model to correct the acquired transcranial ultrasound image to denoise the transcranial ultrasound image and generate a target transcranial ultrasound image.

2. The noise morphology based transcranial ultrasound image adaptive optimization method of claim 1, wherein: The noise form of the extracted noise is at least the noise form of the near-field cloud-like noise in the transcranial ultrasound image, wherein The noise form at least includes one of the noise region depth, the noise region gray mean value and the noise region gray variance; When the noise form includes the noise region depth, a target skull thickness corresponding to the noise region depth is selected; When the noise form includes the noise region gray mean value, a target skull sound speed corresponding to the noise region gray mean value is selected; When the noise form includes the noise region gray variance, a target skull curvature corresponding to the noise region gray variance is selected; Based on the selected target skull thickness, target skull sound speed and / or target skull curvature, a target skull model corresponding to the noise form is formed.

3. The noise morphology based transcranial ultrasound image adaptive optimization method of claim 2, wherein, When extracting the noise form of the near-field cloud-like noise in the transcranial ultrasound image, it comprises: performing preprocessing on the transcranial ultrasound image to remove texture information to generate a transcranial preprocessed image; segmenting and extracting the near-field cloud-like noise region in the transcranial preprocessed image to obtain a target near-field cloud-like noise region; performing at least noise form calculation on the segmented target near-field cloud-like noise region to generate a noise form corresponding to the target near-field cloud-like noise region.

4. The noise morphology based transcranial ultrasound image adaptive optimization method of claim 3, wherein, When preprocessing the transcranial ultrasound image to remove texture information, at least includes low-pass filtering the transcranial ultrasound image, wherein When performing low-pass filtering, the low-pass filtering method includes one of two-dimensional Gaussian filtering, median filtering or Butterworth low-pass filtering.

5. The noise morphology based transcranial ultrasound image adaptive optimization method of claim 3, wherein, When segmenting and extracting the near-field cloud-like noise region in the transcranial processed image, the segmentation and extraction method comprises a two-dimensional ostu segmentation method, wherein When using the two-dimensional ostu segmentation method for segmentation and extraction, a target segmentation pixel gray threshold is determined; segmenting the transcranial processed image using the determined target segmentation pixel gray threshold to obtain a target near-field cloud-like noise region.

6. The noise morphology based transcranial ultrasound image adaptive optimization method of claim 5, wherein, When determining the target segmentation pixel gray threshold, it comprises: Constructing a reference gray scale threshold u The corresponding gray scale threshold screening expression is then: wherein, is a screening reference value, is a probability that a pixel gray scale is less than a reference gray scale threshold u , is a probability that a pixel gray scale is not less than a reference gray scale threshold u , is a gray scale mean value that a pixel gray scale is less than a reference gray scale threshold u , is a gray scale mean value that a pixel gray scale is not less than a reference threshold u . When the reference value is the largest, the corresponding reference gray threshold value is selected as the target segmentation pixel gray threshold value u is configured as the target segmentation pixel gray threshold value.

7. The noise morphology based transcranial ultrasound image adaptive optimization method of claim 3, wherein, Before performing noise form calculation on the target near-field cloud-like noise region, the target near-field cloud-like noise region is also subjected to denoising optimization processing, wherein When performing denoising optimization processing, at least performing morphological opening operation on the target near-field cloud-like noise region.

8. The noise morphology based transcranial ultrasound image adaptive optimization method according to any one of claims 1 to 7, characterized in that, After selecting the target skull model, imaging correction processing is performed, wherein When performing imaging correction processing, it comprises: using the target skull model to correct the model-related change amount of the beamforming delay used when acquiring the transcranial ultrasound image to update and generate a target transcranial ultrasound image.

9. The noise morphology based transcranial ultrasound image adaptive optimization method according to any one of claims 1 to 7, characterized in that, After extracting the noise form of the noise existing in the transcranial ultrasound image, it further comprises: based on the noise form, adjusting the excitation voltage of the ultrasonic probe unit to generate a desired excitation voltage; Driving the ultrasound probe elements at the desired excitation voltage to perform a transcranial ultrasound scan to optimize the noise pattern of the generated transcranial ultrasound image.

10. A noise pattern based transcranial ultrasound image adaptive optimization system, characterized in that, The application comprises a transcranial ultrasound scanning device, which is adapted to optimize the generated transcranial ultrasound image by using the method according to any one of claims 1 to 9.

Images