A CT image recognition and annotation method based on neurosurgery

Abstract

This invention relates to the field of CT image recognition and annotation technology, and particularly to a CT image recognition and annotation method based on neurosurgery. The method includes: segmenting a CT image into several grids; measuring the grayscale value of each grid; annotating abnormal grids; denotating several abnormal grids; connecting the boundaries of the remaining abnormal grids with their corresponding adjacent grids; further refining the annotated grids; and outputting the annotated CT image. This invention, through precise measurement and analysis of grayscale values, can identify abnormal regions in the image. By filtering and annotating abnormal grids, closed abnormal regions are formed, which helps to quickly locate potential lesions or injuries. By dynamically segmenting adjacent grids and continuing annotation, this method can adapt to abnormal regions of different shapes and sizes, ensuring the continuity and integrity of the annotation. It effectively solves the problem of poor image recognition and annotation capabilities for unfamiliar new types of images due to reliance on specific datasets.

Description

Technical Field

[0001] The present invention relates to the technical field of CT image recognition and annotation, and in particular to a CT image recognition and annotation method based on neurosurgery. Background Art

[0002] With the rapid development of medical imaging technology and the increasing popularity of computer-aided diagnosis systems in clinical applications, the demand for accurate CT image recognition and annotation methods is growing, especially in the field of neurosurgery, where accurate image segmentation is crucial for surgical planning, navigation, and ultimately treatment outcomes.

[0003] Patent document with publication number CN115311249A discloses a method for segmenting blood vessels and nerves in neurosurgery microsurgery images. The method converts surgical videos into high-quality color images, creates and enhances annotation datasets, uses an improved FastFCN deep learning model, and goes through a training and testing process to segment blood vessels and nerve structures in surgical images.

[0004] It can be seen that the following problems exist: deep learning models usually require powerful computing resources for training and inference, which is a challenge for some resource-constrained medical institutions, and the performance of the model is highly dependent on the quality and diversity of the training data. If the data set is not comprehensive or biased, it will affect the generalization ability of the model. The model may perform well on a specific data set, but may perform poorly on new types of microscopic images that have never been seen. Summary of the invention

[0005] To this end, the present invention provides a CT image recognition and annotation method based on neurosurgery to overcome the problem in the prior art that the method relies on a specific data set, resulting in poor recognition and annotation capabilities for new types of images that have not been seen before.

[0006] To achieve the above object, the present invention provides a CT image recognition and annotation method based on neurosurgery, comprising:

[0007] The CT image to be annotated is divided into a plurality of square grids at a preset grid spacing;

[0008] Measure the grayscale value of each grid, calculate the average grayscale according to the measurement results, determine the abnormal grid according to the calculation results and the preset standard grayscale, and mark the abnormal grid within the preset standard abnormal range to form a marked abnormal grid;

[0009] Calculate the average center point based on the center point coordinates of all annotated anomaly grids;

[0010] The standard deviation is calculated based on the distance between the center point of each anomaly grid and the average center point, and the annotations of several anomaly grids are cancelled based on the calculation result and a preset standard deviation threshold;

[0011] Connect the remaining several anomaly-marked grids and the boundaries of the corresponding adjacent grids to form an anomaly range, and mark the grids within the anomaly range to form a corrected anomaly range;

[0012] Calculate the average grayscale difference between the abnormal grid at the boundary of the corrected abnormal range in the horizontal direction and the adjacent grids, and end the horizontal marking according to the calculation result and the preset standard grayscale difference, or split the adjacent grids of the abnormal grid at the boundary of the corrected abnormal range in the horizontal direction and continue marking until the horizontal marking is ended;

[0013] Calculate the average grayscale difference between the abnormal grid at the boundary of the corrected abnormal range in the vertical direction and the adjacent grids, and end the vertical marking according to the calculation result and the preset standard grayscale difference, or split the adjacent grids of the abnormal grid at the boundary of the corrected abnormal range in the vertical direction and continue marking until the vertical marking is ended;

[0014] After completing the annotation in all directions, the CT image with annotations is output.

[0015] Furthermore, the calculating of the average center point according to the center point coordinates of all annotated abnormal grids includes:

[0016] Calculate the average value of the horizontal axis of the center point coordinates of all marked abnormal grids to obtain the horizontal coordinate of the average center point;

[0017] Calculate the average value of the vertical axis of the center point coordinates of all marked abnormal grids to obtain the vertical coordinate of the average center point.

[0018] Furthermore, the step of cancelling the annotations of several abnormally annotated grids according to the calculation results and a preset standard deviation threshold comprises:

[0019] Unlabel the anomaly grids whose standard deviation of the distance from the label center point to the mean center point is greater than the standard deviation threshold.

[0020] Furthermore, the step of calculating the average grayscale difference between the abnormal grid and the adjacent grids for correcting the boundary of the abnormal range in the horizontal direction comprises:

[0021] The average grayscale of the grids adjacent to the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction is calculated, and the difference between the average grayscale of the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction and the average grayscale of the adjacent grids is taken.

[0022] Furthermore, the marking in the horizontal direction is ended according to the calculation result and the preset standard grayscale difference value, including:

[0023] When the average grayscale difference between the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction and the adjacent grid is greater than the preset standard grayscale difference, the marking in the horizontal direction is ended.

[0024] Further, according to the calculation result and the preset standard grayscale difference, the adjacent grids of the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction are segmented and the marking is continued until the marking in the horizontal direction is finished, including:

[0025] The segmentation ratio is calculated based on the average grayscale difference between the abnormal grid and the adjacent grid at the boundary of the corrected abnormal range in the horizontal direction and the preset standard grayscale difference. The adjacent grids are divided into several sub-grids according to the segmentation ratio. The average grayscale difference between the abnormal grid and the adjacent sub-grid at the boundary of the corrected abnormal range in the horizontal direction is calculated. The labeling is ended or continued according to the calculation result and the preset standard grayscale difference.

[0026] Furthermore, the step of ending or continuing the marking according to the calculation result and the preset standard grayscale difference value includes:

[0027] When the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the horizontal direction is greater than the preset standard grayscale difference, the marking is terminated;

[0028] When the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the horizontal direction is less than or equal to the preset standard grayscale difference, the new segmentation ratio continues to be calculated until the marking is completed.

[0029] Further, according to the calculation result and the preset standard grayscale difference, the adjacent grids of the abnormal grid that corrects the boundary of the abnormal range in the vertical direction are segmented and the marking is continued until the marking in the vertical direction is finished, including:

[0030] The segmentation ratio is calculated based on the average grayscale difference between the abnormal grid and the adjacent grid at the boundary of the corrected abnormal range in the vertical direction and the preset standard grayscale difference. The adjacent grids are divided into several sub-grids according to the segmentation ratio. The average grayscale difference between the abnormal grid and the adjacent sub-grid at the boundary of the corrected abnormal range in the vertical direction is calculated. The labeling is ended or continued according to the calculation result and the preset standard grayscale difference.

[0031] Furthermore, ending the marking according to the calculation result and the preset standard grayscale difference value includes:

[0032] The marking is terminated when the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the vertical direction is greater than the preset standard grayscale difference;

[0033] Further, the marking according to the calculation result and the preset standard grayscale difference includes:

[0034] When the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the vertical direction is less than or equal to the preset standard grayscale difference, the new segmentation ratio continues to be calculated until the marking is completed.

[0035] Compared with the prior art, the beneficial effect of the present invention is that by accurately measuring and analyzing the grayscale value, the abnormal area in the image can be identified, and by screening and labeling the abnormal grid, a closed abnormal area is formed, which helps to quickly locate potential lesions or injuries. In addition, by dynamically segmenting adjacent grids and continuing to label, the method can adapt to abnormal areas of different shapes and sizes, ensuring the continuity and integrity of the labeling. The final output of the labeled CT image provides richer and more accurate visual information, thereby improving the accuracy of diagnosis, and effectively solving the problem of poor recognition and labeling capabilities for new types of images that have not been seen due to reliance on specific data sets.

[0036] Furthermore, firstly, by calculating the average center point, a reference benchmark is provided for abnormal grids, which helps to identify the spatial distribution pattern of abnormal grids. Secondly, by setting the standard deviation threshold to cancel the annotation of abnormal grids far away from the average center point, possible false positives are effectively removed, and the accuracy and credibility of the annotation results are enhanced. In addition, this method can adaptively adjust the annotation process to ensure that the final annotation results reflect the true situation of the image and reduce the influence of human subjective judgment.

[0037] Furthermore, the termination point of the annotation is automatically determined by calculating the grayscale difference. This decision-making mechanism based on grayscale contrast effectively avoids over-annotation in areas where the grayscale changes are not significant, thereby improving the accuracy and efficiency of annotation. In addition, through automated grayscale analysis and annotation termination, the interference of human factors is reduced, ensuring the consistency of annotation results between different operators.

[0038] Furthermore, by calculating the grayscale difference and comparing the threshold, the method can accurately identify the edge of the abnormal area, avoiding unnecessary labeling in areas where the grayscale change is not obvious, thereby improving the accuracy and efficiency of labeling. In addition, by ending the labeling when the grayscale change is large and continuing to subdivide and label when the grayscale change is small, the method can adapt to different image features and the shape of the abnormal area, ensuring the consistency and reliability of the labeling results. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of the process of the CT image recognition and annotation method based on neurosurgery in this embodiment;

[0040] Figure 2 The decision logic diagram that cancels some abnormal grids is used for this embodiment;

[0041] Figure 3 This is a decision logic diagram for ending the annotation in the horizontal direction of this embodiment;

[0042] Figure 4 This is the decision logic diagram for ending the annotation in the vertical direction for this embodiment. DETAILED DESCRIPTION

[0043] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0044] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0045] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0046] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] See also Figure 1 As shown, it is a flow chart of the CT image recognition and annotation method based on neurosurgery in this embodiment;

[0048] A CT image recognition and annotation method based on neurosurgery, comprising:

[0049] The CT image to be annotated is divided into a plurality of square grids at a preset grid spacing;

[0050] Measure the grayscale value of each grid, calculate the average grayscale according to the measurement results, determine the abnormal grid according to the calculation results and the preset standard grayscale, and mark the abnormal grid within the preset standard abnormal range to form a marked abnormal grid;

[0051] Calculate the average center point based on the center point coordinates of all annotated anomaly grids;

[0052] The standard deviation is calculated based on the distance between the center point of each anomaly grid and the average center point, and the annotations of several anomaly grids are cancelled based on the calculation result and a preset standard deviation threshold;

[0053] Connect the remaining several anomaly-marked grids and the boundaries of the corresponding adjacent grids to form an anomaly range, and mark the grids within the anomaly range to form a corrected anomaly range;

[0054] Calculate the average grayscale difference between the abnormal grid at the boundary of the corrected abnormal range in the horizontal direction and the adjacent grids, and end the horizontal marking according to the calculation result and the preset standard grayscale difference, or split the adjacent grids of the abnormal grid at the boundary of the corrected abnormal range in the horizontal direction and continue marking until the horizontal marking is ended;

[0055] Calculate the average grayscale difference between the abnormal grid at the boundary of the corrected abnormal range in the vertical direction and the adjacent grids, and end the vertical marking according to the calculation result and the preset standard grayscale difference, or split the adjacent grids of the abnormal grid at the boundary of the corrected abnormal range in the vertical direction and continue marking until the vertical marking is ended;

[0056] After completing the annotation in all directions, the CT image with annotations is output.

[0057] First, the CT image is divided into multiple small grids, and the grayscale value of each grid is measured and used to calculate the overall average grayscale. Compared with the average grayscale, grids with larger deviations are considered abnormal and are preliminarily annotated within the preset abnormal range. Next, the average position of the center points of these abnormal grids and the standard deviation to the average center point are calculated to remove those grids with abnormal distribution or outliers to improve the accuracy of annotation. Then, by connecting the remaining abnormal grids and their adjacent boundaries, a closed abnormal area is formed, and further annotation is performed within the area. In the horizontal and vertical directions, the grayscale difference between the abnormal grid and the adjacent grid is used to decide whether to continue subdividing and annotating until the preset grayscale difference threshold is reached. Finally, after completing the annotation in all directions, an annotated CT image is generated to provide doctors with clearer image information and assist in diagnosis and treatment planning.

[0058] By accurately measuring and analyzing the grayscale value, the abnormal area in the image can be identified, and by screening and labeling the abnormal grid, a closed abnormal area is formed, which helps to quickly locate potential lesions or injuries. In addition, by dynamically segmenting adjacent grids and continuing to label, this method can adapt to abnormal areas of different shapes and sizes, ensuring the continuity and integrity of the labeling. The final output CT image with annotations provides richer and more accurate visual information, which may improve the accuracy of diagnosis.

[0059] Specifically, the calculation of the average center point according to the center point coordinates of all annotated abnormal grids includes:

[0060] Calculate the average value of the horizontal axis of the center point coordinates of all marked abnormal grids to obtain the horizontal coordinate of the average center point;

[0061] Calculate the average value of the vertical axis of the center point coordinates of all marked abnormal grids to obtain the vertical coordinate of the average center point.

[0062] Please continue reading Figure 2 As shown, it is a decision logic diagram for canceling a number of anomaly-marked grids in this embodiment;

[0063] Specifically, the method of cancelling the annotations of several abnormally annotated grids according to the calculation results and the preset standard deviation threshold comprises:

[0064] Unlabel the anomaly grids whose standard deviation of the distance from the label center point to the mean center point is greater than the standard deviation threshold.

[0065] The first step involves gridding the CT images to facilitate the systematic analysis of each area in the image. After measuring the grayscale value of each grid and calculating the overall average grayscale, the abnormal grids that deviate significantly from the average grayscale are identified and preliminarily annotated within the preset standard abnormal range. Subsequently, the center points of all annotated abnormal grids are determined, and the horizontal and vertical axis averages of these center points are calculated to obtain the coordinates of the average center point. Using the average center point, the distance from the center point of each abnormal grid to it is further calculated, and the standard deviation of these distances is obtained. According to the preset standard deviation threshold, those abnormal grids whose distance exceeds the threshold are removed to ensure the accuracy of the annotation. This process helps to filter out isolated or statistically unrepresentative abnormal points, thereby improving the reliability of subsequent image analysis.

[0066] First, by calculating the average center point, a reference benchmark is provided for abnormal grids, which helps to identify the spatial distribution pattern of abnormal grids. Secondly, by setting the standard deviation threshold to cancel the annotation of abnormal grids far away from the average center point, possible false positives are effectively removed, enhancing the accuracy and credibility of the annotation results. In addition, this method can adaptively adjust the annotation process to ensure that the final annotation results reflect the true situation of the image and reduce the influence of human subjective judgment.

[0067] Specifically, the calculation of the average grayscale difference between the abnormal grid and the adjacent grids for correcting the boundary of the abnormal range in the horizontal direction includes:

[0068] The average grayscale of the grids adjacent to the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction is calculated, and the difference between the average grayscale of the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction and the average grayscale of the adjacent grids is taken.

[0069] Please continue reading Figure 3 As shown, it is a decision logic diagram for ending the marking in the horizontal direction in this embodiment;

[0070] Specifically, the marking in the horizontal direction according to the calculation result and the preset standard grayscale difference includes:

[0071] When the average grayscale difference between the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction and the adjacent grid is greater than the preset standard grayscale difference, the marking in the horizontal direction is ended.

[0072] When calculating the boundary of the corrected abnormal range in the horizontal direction, a detailed analysis of the grayscale values ​​of the abnormal grid and its adjacent grids is involved. First, determine the abnormal grids at the boundary of the abnormal range in the horizontal direction, measure the grayscale values ​​of all adjacent grids on the row where these abnormal grids are located, and calculate their average grayscale. Next, calculate the average grayscale of these abnormal grids themselves, and find the difference between the average grayscale of the adjacent grids. This difference reflects the grayscale contrast between the abnormal grid and its neighboring area. If this difference exceeds the preset standard grayscale difference, it indicates that the grayscale contrast between the abnormal grid and the surrounding area is obvious enough. At this time, further annotation in this horizontal direction will be terminated to ensure the accuracy and effectiveness of the annotation.

[0073] The end point of annotation is automatically determined by calculating the grayscale difference. This decision-making mechanism based on grayscale comparison effectively avoids over-annotation in areas where the grayscale change is not significant, thereby improving the accuracy and efficiency of annotation. In addition, through automated grayscale analysis and annotation termination, the interference of human factors is reduced, ensuring the consistency of annotation results between different operators.

[0074] Specifically, according to the calculation result and the preset standard grayscale difference, the adjacent grids of the abnormal grid that corrects the boundary of the abnormal range in the horizontal direction are segmented and the marking is continued until the marking in the horizontal direction is finished, including:

[0075] The segmentation ratio is calculated based on the average grayscale difference between the abnormal grid and the adjacent grid at the boundary of the corrected abnormal range in the horizontal direction and the preset standard grayscale difference. The adjacent grids are divided into several sub-grids according to the segmentation ratio. The average grayscale difference between the abnormal grid and the adjacent sub-grid at the boundary of the corrected abnormal range in the horizontal direction is calculated. The labeling is ended or continued according to the calculation result and the preset standard grayscale difference.

[0076] Specifically, the step of ending or continuing the marking according to the calculation result and the preset standard grayscale difference value includes:

[0077] When the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the horizontal direction is greater than the preset standard grayscale difference, the marking is terminated;

[0078] When the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the horizontal direction is less than or equal to the preset standard grayscale difference, the new segmentation ratio continues to be calculated until the marking is completed.

[0079] Please continue reading Figure 4 As shown, it is a decision logic diagram for ending the marking in the vertical direction in this embodiment;

[0080] Specifically, according to the calculation result and the preset standard grayscale difference, the adjacent grids of the abnormal grid that corrects the boundary of the abnormal range in the vertical direction are segmented and the marking is continued until the marking in the vertical direction is finished, including:

[0081] The segmentation ratio is calculated based on the average grayscale difference between the abnormal grid and the adjacent grid at the boundary of the corrected abnormal range in the vertical direction and the preset standard grayscale difference. The adjacent grids are divided into several sub-grids according to the segmentation ratio. The average grayscale difference between the abnormal grid and the adjacent sub-grid at the boundary of the corrected abnormal range in the vertical direction is calculated. The labeling is ended or continued according to the calculation result and the preset standard grayscale difference.

[0082] Specifically, the marking ends according to the calculation result and the preset standard grayscale difference value, including:

[0083] The marking is terminated when the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the vertical direction is greater than the preset standard grayscale difference;

[0084] Specifically, the following are further marked based on the calculation results and the preset standard grayscale difference:

[0085] When the average grayscale difference between the abnormal grid and the adjacent subgrid that corrects the boundary of the abnormal range in the vertical direction is less than or equal to the preset standard grayscale difference, the new segmentation ratio continues to be calculated until the marking is completed.

[0086] Finely mark the boundaries of the abnormal range in the horizontal and vertical directions. First, measure the grayscale values ​​of the abnormal grid and its directly adjacent grids, and calculate the average grayscale of these grids. Then, based on the comparison between the average grayscale difference between the abnormal grid and the adjacent grids and the preset standard grayscale difference, determine whether the adjacent grids need to be further segmented. If the average grayscale difference is greater than the standard grayscale difference, the annotation in this direction ends; if it is less than or equal to the standard grayscale difference, the segmentation ratio is calculated according to the difference, and the adjacent grids are divided into smaller sub-grids, and the above process is repeated until the end condition is met. This process ensures that the boundaries of the abnormal area are clearly distinguished from the surrounding tissues in terms of grayscale changes, while maintaining the continuity and integrity of the annotation.

[0087] By calculating the grayscale difference and comparing the threshold, this method can accurately identify the edge of the abnormal area and avoid unnecessary labeling in areas where the grayscale change is not obvious, thereby improving the accuracy and efficiency of labeling. In addition, by ending the labeling when the grayscale change is large and continuing to subdivide and label when the grayscale change is small, this method can adapt to different image features and the shape of the abnormal area, ensuring the consistency and reliability of the labeling results.

[0088] The preset standard grayscale refers to the grayscale value used as a reference during image segmentation and analysis, which is used to compare with the actual measured grid grayscale value to identify areas that deviate significantly from normal conditions.

[0089] It usually depends on the specific application scenario of the image, the quality and characteristics of the image, and the medical expert's prior knowledge of the normal and abnormal grayscale range.

[0090] The specific value needs to be determined based on the actual image data and the advice of domain experts. In some cases, statistics of the image grayscale values, such as the mean or median, may be used as a starting point.

[0091] The standard abnormal range refers to one or more preset areas in the image that are considered to be particularly important or need special attention, such as areas that may contain abnormal structures or features.

[0092] This scope is usually set based on medical knowledge, image content, and analysis purpose. For example, in neurosurgery, a specific area of ​​the brain may be of interest.

[0093] The size and location of the scope will be determined according to specific medical diagnostic requirements and image characteristics.

[0094] The standard deviation threshold is a statistical parameter used to evaluate the degree of dispersion of data point distribution. In this method, it is used to determine which abnormal grid labels should be cancelled.

[0095] It depends on the standard deviation of the distance from the center point of the abnormal grid to the average center point, as well as the requirement for annotation accuracy.

[0096] Setting the standard deviation threshold requires balancing precision and recall, and is usually determined based on previous research or experimental results.

[0097] The standard grayscale difference refers to a preset standard for the grayscale difference between an abnormal grid and an adjacent grid in the horizontal or vertical direction, which is used to decide whether to continue marking in that direction.

[0098] It depends on the grayscale variation characteristics of the image and the requirements for segmentation boundary clarity.

[0099] The setting of this value needs to take into account the level of detail of the image and the accuracy requirements of the segmentation, and may be adjusted according to the specific content of the image and the purpose of analysis.

[0100] In this embodiment, reasonably setting these criteria helps to improve the accuracy and efficiency of image segmentation:

[0101] The preset standard grayscale can help quickly identify abnormal areas in the image.

[0102] Standard anomaly ranges ensure that the analysis is focused on the most important areas of the image.

[0103] The standard deviation threshold helps to remove outliers and improve the reliability of annotations.

[0104] The standard grayscale difference ensures the clarity and accuracy of the boundaries by controlling the fineness of the segmentation process.

[0105] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A CT image recognition and annotation method based on neurosurgery, characterized in that: include: The CT image to be annotated is divided into a plurality of square grids at a preset grid spacing; Measure the grayscale value of each grid, calculate the average grayscale according to the measurement results, determine the abnormal grid according to the average grayscale and the preset standard grayscale, and mark the abnormal grid within the preset standard abnormal range to form a marked abnormal grid; Calculate the average center point based on the center point coordinates of all annotated anomaly grids; Calculate the standard deviation based on the distance from the center point of each anomaly grid to the average center point, and cancel the anomaly grids whose standard deviation from the center point to the average center point is greater than the standard deviation threshold; Connect the remaining several anomaly-marked grids and the boundaries of the corresponding adjacent grids to form an anomaly range, and mark the grids within the anomaly range to form a corrected anomaly range; Calculate the average grayscale difference between the abnormal grid and the adjacent grids at the boundary of the corrected abnormal range in the horizontal direction, end the marking in the horizontal direction when the average grayscale difference is greater than the preset standard grayscale difference, calculate the segmentation ratio according to the average grayscale difference and the preset standard grayscale difference when the average grayscale difference is less than or equal to the preset standard grayscale difference, divide the adjacent grids into a plurality of sub-grids according to the segmentation ratio, calculate the average grayscale difference between the abnormal grid and the adjacent sub-grids at the boundary of the corrected abnormal range in the horizontal direction, end the marking when the average grayscale difference is greater than the preset standard grayscale difference, and continue to calculate the new segmentation ratio when the average grayscale difference is less than or equal to the preset standard grayscale difference until the marking is ended; Calculate the average grayscale difference between the abnormal grid and the adjacent grids at the boundary of the corrected abnormal range in the vertical direction, end the marking in the vertical direction when the average grayscale difference is greater than the preset standard grayscale difference, calculate the segmentation ratio according to the average grayscale difference and the preset standard grayscale difference when the average grayscale difference is less than or equal to the preset standard grayscale difference, divide the adjacent grids into a plurality of sub-grids according to the segmentation ratio, calculate the average grayscale difference between the abnormal grid and the adjacent sub-grids at the boundary of the corrected abnormal range in the vertical direction, end the marking when the average grayscale difference is greater than the preset standard grayscale difference, and continue to calculate the new segmentation ratio when the average grayscale difference is less than or equal to the preset standard grayscale difference until the marking is ended; After completing the annotation in all directions, the CT image with annotations is output.

2. The neurosurgery-based CT image recognition and annotation method according to claim 1, characterized in that: Calculating the average center point according to the center point coordinates of all annotated abnormal grids includes: Calculate the average value of the horizontal axis of the center point coordinates of all marked abnormal grids to obtain the horizontal coordinate of the average center point; Calculate the average value of the vertical axis of the center point coordinates of all marked abnormal grids to obtain the vertical coordinate of the average center point.

Images