Quality assessment method for three-dimensional breast ultrasound image

Abstract

The present invention relates to the field of three-dimensional ultrasound image processing, and particularly relates to a quality assessment method for a three-dimensional breast ultrasound image. In the present invention, the following technical solution is used: parsing a three-dimensional breast ultrasound image, so as to obtain several cross-sectional image frames; identifying all artifact regions from the cross-sectional image frames, and marking the sizes thereof; on the basis of the aspect ratios and average pixel values of the artifact regions, classifying the artifact regions; on the basis of frame numbers of cross-sectional images where the artifact regions are located, grouping the artifact regions; and compiling statistics on artifact results, and then outputting a quality assessment result. The beneficial effects of the present invention lie in: the sizes and categories of artifact regions can be obtained, and on the basis of frame numbers corresponding to the artifact regions, the categories, quantities and widths of artifact regions present in cross-sectional image frames can be determined, such that a quality assessment result of a three-dimensional breast ultrasound image can be accurately output on this basis, thereby helping a doctor to quickly and accurately determine the quality of an ultrasound examination image, and thus improving the assessment efficiency.

Description

A method for quality inspection of three-dimensional ultrasound images of breast tissue Technical Field

[0001] This invention relates to the field of three-dimensional ultrasound image processing, and more particularly to a method for quality inspection of three-dimensional ultrasound images of breast tissue. Background Technology

[0002] To accurately detect and determine the presence of lesions in the breast, most current methods employ three-dimensional ultrasound imaging. However, in these images, artifacts such as empty areas, poor coupling, or multilayer reflections frequently appear due to improper detection (e.g., insufficient probe-to-scan area contact or the presence of air bubbles or thin layers of air between the probe and the scan site). These artifacts affect the detection of breast lesions to varying degrees. Quality assessment of three-dimensional ultrasound images often requires physicians to identify and judge artifacts based on the cross-sectional images resolved from the image. Furthermore, determining the quality of the three-dimensional ultrasound image relies heavily on the physician's experience, leading to inaccurate judgments and low efficiency. While some pre-trained artifact detection models identify artifact regions, they can only identify the extent of the artifact, not its type. Different types of artifacts have varying impacts on the quality of the three-dimensional ultrasound image, thus these models also cannot accurately output quality assessment results. Summary of the Invention

[0003] The purpose of this invention is to provide a quality inspection method for three-dimensional ultrasound images of breast tissue. Specifically, it provides a quality inspection method that can automatically and quickly detect the size, classification, and number of artifact regions in each frame of cross-sectional images of three-dimensional ultrasound images of breast tissue and output the quality inspection results.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a method for quality detection of three-dimensional ultrasound images of breast tissue, characterized by comprising the following steps:

[0005] S01. First, perform three-dimensional ultrasound detection on the breast to obtain a three-dimensional ultrasound breast image, and then extract several cross-sectional images from the three-dimensional ultrasound breast image.

[0006] S02. Using the trained artifact detection model, perform artifact region detection on several frames of cross-sectional image sequence, detect and segment all I artifact regions present in the several frames of cross-sectional images, and record the widths W1, W2, W3, ... W of each artifact region. I and heights H1, H2, H3, ... H I , and the frame number of the cross-sectional image where the I artifact regions are located.

[0007] S03. Classify the I artifact regions, specifically including:

[0008] S31. First, remove the artifact regions whose width is less than the set threshold W0 from the I artifact regions.

[0009] S32. Calculate the aspect ratios W1 / H1, W2 / H2, W3 / H3, ..., W for I artifact regions. I / H I Then, the aspect ratios of the artifact regions removed in step S31 are arranged in ascending order. The point where the aspect ratio increases significantly is found. The artifact regions corresponding to the aspect ratio increase point and the subsequent aspect ratios are classified as "multilayer reflection" artifacts. The remaining artifact regions are classified as "poor coupling" or "empty area" artifacts.

[0010] S33. In the artifact regions classified as "poorly coupled" or "empty area" artifacts in step S32, calculate the average pixel value of each artifact region and arrange the average pixel values ​​in ascending order. Find the point where the average pixel value increases significantly by a jump, and classify the artifact region corresponding to the average pixel value increase point and the subsequent average pixel values ​​as "poorly coupled" artifacts. The remaining artifact regions are classified as "empty area" artifacts.

[0011] S04. The artifact regions classified in steps S32 and S33 are grouped according to the frame number of the cross-sectional image where the artifact region is located. Artifact regions with the same frame number in the cross-sectional image are grouped into the same group, and the classification, number and width of artifact regions in the same group are counted respectively.

[0012] S05. Output the quality detection results of the three-dimensional ultrasound image of the breast based on the statistical results obtained in step S04.

[0013] Specifically, in step S05, when outputting the quality detection result of the three-dimensional ultrasound image of the breast based on the statistical results obtained in step S04, the following steps are included:

[0014] S51. If there is an artifact region classified as "multi-layer reflection" artifact in any group of artifact regions, the quality detection result of the three-dimensional ultrasound image of the breast is directly determined to be unqualified, and the result "quality detection unqualified, multi-layer reflection exists" is output. If there is no artifact region classified as "multi-layer reflection" artifact in any group of artifact regions, proceed to S52.

[0015] S52. If the width of the artifact region is greater than the set width threshold W' or the number of artifact regions in the same group is greater than the set quantity threshold N, then the cross-sectional image quality of the frame number corresponding to the artifact region is unqualified, and the number of frames of the unqualified cross-sectional image is recorded.

[0016] S53. If the number of frames of the cross-sectional image with substandard quality recorded in step S52 is greater than the set frame number threshold M, the quality detection result of the three-dimensional ultrasound image of the breast is determined to be substandard. If it is not greater than the set frame number threshold M, the quality detection result of the three-dimensional ultrasound image of the breast is determined to be qualified.

[0017] Specifically, in step S53, when the number of frames of the cross-sectional image with substandard quality recorded in step S52 is greater than the set frame number threshold M, not only is the quality detection result of the three-dimensional ultrasound image of the breast determined to be substandard, but also the quality problem of the three-dimensional ultrasound image of the breast is output according to the number of artifact regions classified as "poor coupling" artifacts and the number of artifact regions classified as "empty region" artifacts in step S33.

[0018] The beneficial effects of this invention are as follows: by filtering the aspect ratio and average pixel value of the artifact regions in each frame of the three-dimensional ultrasound image of the breast, the size and classification of each artifact region can be obtained. Based on the frame number corresponding to the artifact region, the classification, number and width of the artifact regions in each frame of the cross-sectional image can be determined. Based on this, the quality detection results of the three-dimensional ultrasound image of the breast can be accurately output, which facilitates doctors to make a quick and accurate judgment on the quality of the ultrasound detection image and improves the detection efficiency. Detailed Implementation

[0019] Example 1: A method for quality inspection of three-dimensional ultrasound images of breast tissue, characterized by comprising the following steps:

[0020] S01. First, perform three-dimensional ultrasound detection on the breast to obtain a three-dimensional ultrasound breast image, and then extract several cross-sectional images from the three-dimensional ultrasound breast image.

[0021] S02. Using the trained artifact detection model, perform artifact region detection on several frames of cross-sectional image sequence, detect and segment all I artifact regions present in the several frames of cross-sectional images, and record the widths W1, W2, W3, ... W of each artifact region. I and heights H1, H2, H3, ... H I , and the frame number of the cross-sectional image where the I artifact regions are located.

[0022] S03. Classify the I artifact regions, specifically including:

[0023] S31. First, remove the artifact regions whose width is less than the set threshold W0 from the I artifact regions.

[0024] S32. Calculate the aspect ratios W1 / H1, W2 / H2, W3 / H3, ..., W for I artifact regions. I / HI Then, the aspect ratios of the artifact regions removed in step S31 are arranged in ascending order. The point where the aspect ratio increases significantly is found. The artifact regions corresponding to the aspect ratio increase point and the subsequent aspect ratios are classified as "multilayer reflection" artifacts. The remaining artifact regions are classified as "poor coupling" or "empty area" artifacts.

[0025] S33. In the artifact regions classified as "poorly coupled" or "empty area" artifacts in step S32, calculate the average pixel value of each artifact region and arrange the average pixel values ​​in ascending order. Find the point where the average pixel value increases significantly by a jump, and classify the artifact region corresponding to the average pixel value increase point and the subsequent average pixel values ​​as "poorly coupled" artifacts. The remaining artifact regions are classified as "empty area" artifacts.

[0026] S04. The artifact regions classified in steps S32 and S33 are grouped according to the frame number of the cross-sectional image where the artifact region is located. Artifact regions with the same frame number in the cross-sectional image are grouped into the same group, and the classification, number and width of artifact regions in the same group are counted respectively.

[0027] S05. Output the quality detection results of the three-dimensional ultrasound image of the breast based on the statistical results obtained in step S04.

[0028] Specifically, in step S05, when outputting the quality detection result of the three-dimensional ultrasound image of the breast based on the statistical results obtained in step S04, the following steps are included:

[0029] S51. If there is an artifact region classified as "multi-layer reflection" artifact in any group of artifact regions, the quality detection result of the three-dimensional ultrasound image of the breast is directly determined to be unqualified, and the result "quality detection unqualified, multi-layer reflection exists" is output. If there is no artifact region classified as "multi-layer reflection" artifact in any group of artifact regions, proceed to S52.

[0030] S52. If the width of the artifact region is greater than the set width threshold W' or the number of artifact regions in the same group is greater than the set quantity threshold N, then the cross-sectional image quality of the frame number corresponding to the artifact region is unqualified, and the number of frames of the unqualified cross-sectional image is recorded.

[0031] S53. If the number of frames of the cross-sectional image with substandard quality recorded in step S52 is greater than the set frame number threshold M, the quality detection result of the three-dimensional ultrasound image of the breast is determined to be substandard. If it is not greater than the set frame number threshold M, the quality detection result of the three-dimensional ultrasound image of the breast is determined to be qualified.

[0032] Specifically, in step S53, when the number of frames of the cross-sectional image with substandard quality recorded in step S52 exceeds the set frame number threshold M, not only is the quality detection result of the three-dimensional ultrasound image of the breast determined to be substandard, but also, based on the number of artifact regions classified as "poor coupling" artifacts and the number of artifact regions classified as "empty regions" artifacts in step S33, the quality problem of the three-dimensional ultrasound image of the breast is output. Specifically, the quality problem of the three-dimensional ultrasound image of the breast can be output directly by outputting the number of artifact regions classified as "poor coupling" artifacts and the number of artifact regions classified as "empty regions" artifacts, or it can be output based on the number of "poor coupling" and "empty regions" artifacts. For example, when the number of "poor coupling" artifacts exceeds the set value, the problem of "poor coupling" artifacts in the three-dimensional ultrasound image of the breast is output; when the number of "empty regions" artifacts exceeds the set value, the problem of "empty regions" artifacts in the three-dimensional ultrasound image of the breast is output.

[0033] Of course, the above are only preferred embodiments of the present invention and are not intended to limit the scope of application of the present invention. Therefore, any equivalent changes made to the principle of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for quality inspection of three-dimensional ultrasound images of breast tissue, characterized in that, Includes the following steps: S01. First, perform three-dimensional ultrasound detection on the breast to obtain a three-dimensional ultrasound breast image, and then extract several cross-sectional images from the three-dimensional ultrasound breast image. S02. Using the trained artifact detection model, perform artifact region detection on several frames of cross-sectional image sequence, detect and segment all I artifact regions present in the several frames of cross-sectional images, and record the widths W1, W2, W3, ... W of each artifact region. I and heights H1, H2, H3, ... H I , and the frame number of the cross-sectional image where the I artifact regions are located; S03. Classify the I artifact regions, specifically including: S31. First, remove the artifact regions whose width is less than the set threshold W0 from the I artifact regions. S32. Calculate the aspect ratios W1 / H1, W2 / H2, W3 / H3, ..., W for I artifact regions. I / H I Then, the aspect ratios of the artifact regions removed in step S31 are arranged in ascending order. The aspect ratio growth point with obvious jump is found. The artifact regions corresponding to the aspect ratio growth point and the subsequent aspect ratio are classified as "multilayer reflection" artifacts. The remaining artifact regions are classified as "poor coupling" or "empty area" artifacts. S33. In the artifact regions classified as "poorly coupled" or "empty region" artifacts in step S32, calculate the average pixel value of each artifact region and arrange the average pixel values ​​in ascending order. Find the point where the average pixel value increases significantly by a jump, and classify the artifact region corresponding to the average pixel value increase point and the average pixel value after it as "poorly coupled" artifact. The remaining artifact regions are classified as "empty region" artifacts. S04. The artifact regions that have been classified in steps S32 and S33 are grouped according to the frame number of the cross-sectional image where the artifact region is located. The artifact regions with the same frame number in the cross-sectional image are grouped into the same group, and the classification, number and width of the artifact regions in the same group are counted respectively. S05. Output the quality detection results of the three-dimensional ultrasound image of the breast based on the statistical results obtained in step S04.

2. The method for quality inspection of three-dimensional ultrasound images of breast tissue according to claim 1, characterized in that: In step S05, when outputting the quality detection result of the three-dimensional ultrasound image of the breast based on the statistical results obtained in step S04, the following steps are specifically included: S51. If there is an artifact region classified as "multi-layer reflection" artifact in any group of artifact regions, the quality detection result of the three-dimensional ultrasound image of the breast is directly determined to be unqualified, and the result "quality detection unqualified, multi-layer reflection exists" is output. If there is no artifact region classified as "multi-layer reflection" artifact in any group of artifact regions, proceed to S52. S52. If the width of the artifact region is greater than the set width threshold W' or the number of artifact regions in the same group is greater than the set quantity threshold N, then the cross-sectional image quality of the frame number corresponding to the artifact region is unqualified, and the number of frames of the unqualified cross-sectional image is recorded. S53. If the number of frames of the cross-sectional image with substandard quality recorded in step S52 is greater than the set frame number threshold M, the quality detection result of the three-dimensional ultrasound image of the breast is determined to be substandard. If it is not greater than the set frame number threshold M, the quality detection result of the three-dimensional ultrasound image of the breast is determined to be qualified.

3. The method for quality inspection of three-dimensional ultrasound images of breast tissue according to claim 2, characterized in that: In step S53, when the number of frames of the cross-sectional image with substandard quality recorded in step S52 is greater than the set frame number threshold M, not only is the quality detection result of the three-dimensional ultrasound image of the breast determined to be substandard, but also the quality problem of the three-dimensional ultrasound image of the breast is output according to the number of artifact regions classified as "poor coupling" artifacts and the number of artifact regions classified as "empty region" artifacts in step S33.