Medical image recognition method and device for lung tumor markers and medium

Abstract

The invention discloses a medical image recognition method and device for a lung tumor marker and a medium, and relates to the technical field of image recognition, and the method comprises the steps: obtaining a to-be-recognized marker based on blood detection; obtaining different markers of the marker to be identified; obtaining heterogeneous mark images; acquiring image recognition features by using a mark analysis method; optimizing the image recognition features, and performing image recognition on the lung tumor markers by using the optimized recognition features; the method is used for solving the problems that in an existing medical image recognition method for the lung tumor markers, a recognition model is relatively conventional, the types of the lung tumor markers cannot be accurately recognized due to the fact that the features of the lung tumor markers in the medical image are not obtained, consequently, much uncertainty exists in the recognition process, and the recognition accuracy is high. And the identification accuracy is low.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, specifically to a medical image recognition method, device, and medium for lung tumor markers. Background Technology

[0002] Lung tumor markers are biomarkers in the blood that indicate elevated levels of specific proteins or biochemical molecules, used to aid in the diagnosis of lung cancer and other lung diseases. They are typically produced and secreted into bodily fluids, such as blood, by malignant cells when abnormal proliferation occurs in lung tissue. Common examples include carcinoembryonic antigen (CEA) and squamous cell carcinoma antigen (SCCA). The clinical significance of lung tumor marker testing lies in its ability to assist in the diagnosis or differential diagnosis of certain lung diseases. In cases of significantly elevated tumor markers, further imaging examinations, such as lung CT scans and pathological examinations, are usually recommended to confirm the diagnosis.

[0003] Existing methods for medical image recognition of lung tumor markers typically involve cropping and segmenting regions within medical images. These segments are then analyzed, and further cropping and model building are employed to identify the markers. While this approach improves efficiency, lung tumor markers are often indistinguishable from other markers in medical images. Furthermore, the lack of specific features of lung tumor markers within the images prevents assurance that the cropped regions contain the markers to be identified, leading to significant uncertainty in the recognition process. Even after identifying lung tumor markers, accurate identification of their type is often impossible, resulting in low accuracy. For example, in publication CN11... Patent application 4511569A discloses a medical image recognition method, apparatus, device, and medium based on tumor markers. This method involves cropping a first region of interest and then cropping a second region of interest within the first region of interest, followed by comprehensive recognition and judgment using a recognition model. Other medical image recognition methods for lung tumor markers typically use pre-built recognition models to identify medical images. These improved methods still suffer from limitations, such as conventional recognition models, lack of understanding of the characteristics of lung tumor markers in medical images, leading to significant uncertainties in the recognition process. Furthermore, even after identifying lung tumor markers, they cannot accurately identify the type of lung tumor marker, resulting in low recognition accuracy. Therefore, it is necessary to improve existing medical image recognition methods for lung tumor markers. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art by proposing a medical image recognition method, device, and medium for lung tumor markers. This addresses the problem that existing medical image recognition methods for lung tumor markers use relatively conventional recognition models, which lack the characteristics of lung tumor markers in medical images, leading to significant uncertainties in the recognition process. Furthermore, after identifying lung tumor markers, the methods cannot accurately identify the type of lung tumor markers, resulting in low recognition accuracy.

[0005] To achieve the above objectives, in a first aspect, this application provides a medical image recognition method for lung tumor markers, comprising the following steps:

[0006] The patient's lung tumor image is recorded as the patient's marker image; lung tumor markers present in the patient's body are obtained based on blood tests and recorded as markers to be identified; all markers to be identified are placed into a marker set.

[0007] For any marker α to be identified, markers other than α in the marker set are denoted as heterogeneous markers of α; lung tumor images of patients with only heterogeneous markers in their bodies are obtained from the database and denoted as heterogeneous marker images; marker analysis is used to analyze the patient marker images and heterogeneous marker images, and the image recognition features of marker α are obtained based on the analysis results.

[0008] Obtain the image recognition features of all the markers to be identified;

[0009] The image recognition features of all markers to be identified are optimized, and the preferred recognition features of each marker to be identified are obtained based on the optimization results. The preferred recognition features are then used to perform image recognition of lung tumor markers.

[0010] Furthermore, marker analysis includes:

[0011] The patient marker image is converted to grayscale and denoted as the grayscale marker image. The grayscale marker image is then binarized as follows: for any pixel in the grayscale marker image, if the pixel's grayscale value is less than the equilibrium grayscale value, the pixel's grayscale value is adjusted to a very dark grayscale value; if the pixel's grayscale value is greater than or equal to the equilibrium grayscale value, the pixel's grayscale value is adjusted to a very white grayscale value.

[0012] Obtain all closed contours in the binarized grayscale marker image and label them sequentially as marker contour BL1 to marker contour BL2. c For any given marker outline BL v v is a positive integer less than or equal to c and greater than or equal to 1, marking the outline BL.v The position is marked in the grayscale marker image, and the position of the marker within the marker outline BL in the grayscale marker image is marked. v The pixels in the diagram are denoted as outline pixels.

[0013] Furthermore, marker analysis also includes:

[0014] For any contour pixel, obtain the gray values ​​of all pixels in the eight neighborhoods of the contour pixel, and sort the gray values ​​of the pixels in the eight neighborhoods of the contour pixel in a clockwise direction, starting from the gray value of the pixel directly above the contour pixel. The set corresponding to the eight gray values ​​is recorded as the pixel feature code of the contour pixel; obtain the pixel feature codes of all contour pixels.

[0015] Obtain the pixel feature codes of all contour pixels in the grayscale logo image corresponding to all logo contours BL.

[0016] Furthermore, marker analysis also includes:

[0017] For any acquired heterologous marker image, the heterologous markers in the patient's body corresponding to the heterologous marker image are sequentially denoted as analytical marker FB1 to analytical marker FB2. n The outlier image is converted to grayscale, and the grayscale outlier image is denoted as the grayscale outlier image. The grayscale outlier image is then binarized, and multiple closed contours corresponding to the binarized outlier image are obtained and sequentially denoted as outlier contour YL1 to outlier contour YL. r ;

[0018] For any out-of-type contour YL b b is a positive integer less than or equal to r and greater than or equal to 1; the grayscale outlier image containing the outlier contour YL b The pixels in the image are denoted as outlier pixels; obtain the pixel feature codes of all outlier pixels;

[0019] All marker contours BL are analyzed sequentially using the contour pixel comparison method. The contour pixel comparison method is as follows: for any marker contour BL, when the pixel feature code of a contour pixel A corresponding to the marker contour BL matches the pixel feature code of the out-of-class contour YL, the pixel feature code of the marker contour BL matches the pixel feature code of the out-of-class contour YL. b When the pixel feature codes of any corresponding heterogeneous pixel B are the same, the contour pixel A is recorded as the same pixel outside heterogeneous pixel B, and the difference between the gray value of the contour pixel A and the gray value of heterogeneous pixel B is recorded as the difference between the same pixels.

[0020] Furthermore, marker analysis also includes:

[0021] When a different type of pixel B has multiple corresponding pixels, the corresponding pixel with the smallest difference is recorded as a pixel of the same type as the different type of pixel B. When a different type of pixel B has only one corresponding pixel, the corresponding pixel is recorded as a pixel of the same type as the different type of pixel B. Specifically, when any pixel of the same type as a different type of pixel B is obtained, it is used in subsequent analysis of the different type contour YL. b When analyzing outlier pixels other than outlier pixel B, remove outlier pixels of the same type as outlier pixel B from the outline pixels corresponding to the marker outline BL.

[0022] When the pixel feature code of all contour pixels in the mark contour BL matches the heterogeneous contour YL b When the pixel feature codes of any out-of-type pixel C are all different, the out-of-type pixel C is recorded as an independent pixel.

[0023] The out-of-type contour YL b The similarity value is the difference between the number of dissimilar pixels containing the same type of pixels and the number of independent pixels.

[0024] Furthermore, marker analysis also includes:

[0025] After analyzing all marker contours BL using the contour pixel comparison method, obtain the similarity values ​​of the same category, and record the marker contour BL corresponding to the largest similarity value as the undetermined contour; when the similarity value of the undetermined contour is greater than or equal to 0, record the undetermined contour as the out-of-category contour YL. b Lateral analysis features; when the similarity value of the same type corresponding to the undetermined contour is less than 0, no dissimilar contour YL is set. b Lateral analysis features;

[0026] All out-of-class contours YL that do not contain lateral analysis features are denoted as image recognition features of the target marker α, and all out-of-class contours YL that contain lateral analysis features are denoted as analysis markers FB1 to FB1. n Provisional image features;

[0027] Analyze all outlier marker images to obtain all image recognition features of the marker to be identified α and provisional image features of outlier marker FB in all outlier marker images;

[0028] Based on the analysis method for the marker α to be identified, all markers to be identified are analyzed.

[0029] Furthermore, the image recognition features of all the markers to be identified are optimized, and the preferred recognition features for each marker are obtained based on the optimization results. Image recognition of lung tumor markers using the preferred recognition features includes:

[0030] The image recognition features of all the markers to be identified are optimized using a marker optimization method;

[0031] The marker optimization method is as follows: For any marker α to be identified, the feature optimization algorithm is used to obtain the effective recognition parameters of each image recognition feature of the marker α. For any image recognition feature β of the marker α to be identified, the feature optimization algorithm is as follows: Where F is the effective recognition parameter, f2 is the number of image recognition features β among all image recognition features of the marker α to be identified, f1 is the number of all image recognition features of the marker α to be identified, f3 is the number of outlier marker images when the marker α to be identified is analyzed as marker FB, and f4 is the number of provisional image features of the marker α to be identified as image recognition features β when outlier marker images are analyzed.

[0032] Furthermore, the image recognition features of all the markers to be identified are optimized, and the preferred recognition features for each marker are obtained based on the optimization results. Image recognition of lung tumor markers using the preferred recognition features also includes:

[0033] The image feature with the largest effective recognition parameter of the marker α to be identified is recorded as the preferred recognition feature of the marker α to be identified;

[0034] The preferred identification features of all markers to be identified are obtained, and when performing image recognition of lung tumor markers, the lung tumor markers are identified based on the contours corresponding to the preferred identification features of each marker.

[0035] Secondly, this application also provides a medical image recognition system for lung tumor markers, including a marker set acquisition module, a recognition feature analysis module, and a marker recognition module;

[0036] The biomarker set acquisition module is used to record the patient's lung tumor image as the patient's biomarker image; to obtain the lung tumor biomarkers present in the patient's body based on blood tests and record them as biomarkers to be identified, and to put all biomarkers to be identified into the biomarker set;

[0037] The feature analysis module is used to identify any marker α, and to denote markers other than α in the marker set as heterogeneous markers of α; to obtain lung tumor images of patients in which only heterogeneous markers are present in the body based on the database, and to denote them as heterogeneous marker images; to analyze the patient marker images and heterogeneous marker images using the marker analysis method, and to obtain the image recognition features of marker α based on the analysis results;

[0038] Obtain the image recognition features of all the markers to be identified;

[0039] The marker recognition module is used to optimize the image recognition features of all markers to be identified, and obtain the preferred recognition features for each marker based on the optimization results. The preferred recognition features are then used to perform image recognition of lung tumor markers.

[0040] Thirdly, this application provides a storage medium on which a computer program is stored, which, when executed by a processor, performs the steps of the method described above.

[0041] The beneficial effects of this invention are as follows: First, this application obtains the biomarkers to be identified based on blood testing and places all the biomarkers to be identified into a biomarker set; then, for any biomarker α to be identified, it obtains the outlier biomarkers of biomarker α; it obtains outlier biomarker images based on a database; it uses biomarker analysis to analyze the patient biomarker images and outlier biomarker images, and obtains the image recognition features of biomarker α based on the analysis results, and then obtains the image recognition features of all biomarkers to be identified. The advantage of this is that by placing the biomarkers to be identified into a biomarker set, it is beneficial to conduct unified analysis of the biomarkers to be identified and to facilitate the subsequent acquisition of outlier biomarkers of the biomarkers to be identified; by acquiring outlier biomarker images and using biomarker analysis to obtain image recognition features, it is possible to obtain the features of the biomarkers to be identified in medical images, and by analyzing images containing the biomarkers to be identified and images not containing the biomarkers to be identified separately, the obtained features are more comprehensive and more closely match the biomarkers to be identified, making subsequent image recognition more accurate;

[0042] This application also optimizes the image recognition features of all the markers to be identified and obtains the preferred recognition features for each marker based on the optimization results. The preferred recognition features are then used to perform image recognition of lung tumor markers. The advantage of this is that by optimizing the image recognition features to obtain the preferred recognition features, the features in the medical image that best match the markers to be identified can be selected, thereby improving the accuracy of subsequent recognition of lung tumor markers. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the system of the present invention;

[0044] Figure 2 This is a flowchart illustrating the steps of the method of the present invention;

[0045] Figure 3 This is a schematic diagram illustrating the acquisition of image recognition features of contour pixels according to the present invention;

[0046] Figure 4 This is a schematic diagram illustrating the analysis of similar pixel values ​​in this invention;

[0047] Figure 5This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Example 1, please refer to Figure 1 As shown, this application provides a medical image recognition system for lung tumor markers, including a marker set acquisition module, a recognition feature analysis module, and a marker recognition module;

[0050] The biomarker set acquisition module is used to record the patient's lung tumor image as the patient's biomarker image; to obtain the lung tumor biomarkers present in the patient's body based on blood tests and record them as biomarkers to be identified, and to put all biomarkers to be identified into the biomarker set;

[0051] In the specific implementation process, lung tumor markers may include squamous cell carcinoma antigen (SCC), neuron-specific enolase, carcinoembryonic antigen (CEA), cytokeratin 19 fragment, and gastrin-releasing peptide precursor, etc. The markers to be identified can be labeled according to the lung tumor markers that may be present in the patient's body during the actual analysis.

[0052] The feature analysis module is used to identify any marker α, and to denote markers other than α in the marker set as heterogeneous markers of α; to obtain lung tumor images of patients in which only heterogeneous markers are present in the body based on the database, and to denote them as heterogeneous marker images; to analyze the patient marker images and heterogeneous marker images using the marker analysis method, and to obtain the image recognition features of marker α based on the analysis results;

[0053] Obtain the image recognition features of all the markers to be identified;

[0054] The feature analysis module includes a marker analysis unit, which stores the marker analysis method. The marker analysis method includes: converting the patient marker image to grayscale and recording the grayscale patient marker image as a grayscale marker image; and performing binarization on the grayscale marker image. The binarization process is as follows: for any pixel in the grayscale marker image, if the grayscale value of the pixel is less than the equilibrium grayscale value, the grayscale value of the pixel is adjusted to a very dark grayscale value; if the grayscale value of the pixel is greater than or equal to the equilibrium grayscale value, the grayscale value of the pixel is adjusted to a very white grayscale value.

[0055] In the specific implementation process, the grayscale value of extreme black is 0 and the grayscale value of extreme white is 255. By binarizing the grayscale marker image, it is helpful to obtain the graphic contour in the grayscale marker image, so as to analyze the contour of each region in the patient marker image in subsequent analysis, and thus obtain image recognition features.

[0056] Obtain all closed contours in the binarized grayscale marker image and label them sequentially as marker contour BL1 to marker contour BL2. c For any given marker outline BL v v is a positive integer less than or equal to c and greater than or equal to 1, marking the outline BL. v The position is marked in the grayscale marker image, and the position of the marker within the marker outline BL in the grayscale marker image is marked. v The pixels in the image are denoted as outline pixels;

[0057] For any contour pixel, obtain the gray values ​​of all pixels in the eight neighborhoods of the contour pixel, and sort the gray values ​​of the pixels in the eight neighborhoods of the contour pixel in a clockwise direction, starting from the gray value of the pixel directly above the contour pixel. The set corresponding to the eight gray values ​​is recorded as the pixel feature code of the contour pixel; obtain the pixel feature codes of all contour pixels.

[0058] In the specific implementation process, for example, during a data analysis, a contour pixel is obtained as follows: Figure 3 As shown, LX represents the contour pixel, and L1 to L8 represent the pixels corresponding to the eight neighboring regions of the contour pixel. The gray values ​​of L1 to L8 are 232, 234, 32, 32, 67, 32, 44, and 100, respectively. By sorting, the pixel feature code of the contour pixel is [232, 234, 32, 32, 67, 32, 44, 100]. By obtaining the pixel feature code, the features of the contour pixel can be extracted, which helps to obtain the features of the marker contour BL where the contour pixel is located, and then extract the image recognition features of the marker to be identified.

[0059] Obtain the pixel feature codes of all contour pixels in the grayscale logo image corresponding to all logo contours BL;

[0060] In the specific implementation process, since the contour pixels with the same gray value may be located in multiple marker contours BL, the contour pixels with the same gray value may have multiple pixel feature codes. Therefore, in this embodiment, it is allowed for the contour pixels with the same gray value to have multiple pixel feature codes.

[0061] For any acquired heterologous marker image, the heterologous markers in the patient's body corresponding to the heterologous marker image are sequentially denoted as analytical marker FB1 to analytical marker FB2.n The outlier image is converted to grayscale, and the grayscale outlier image is denoted as the grayscale outlier image. The grayscale outlier image is then binarized, and multiple closed contours corresponding to the binarized outlier image are obtained and sequentially denoted as outlier contour YL1 to outlier contour YL. r ;

[0062] For any out-of-type contour YL b b is a positive integer less than or equal to r and greater than or equal to 1; the grayscale outlier image containing the outlier contour YL b The pixels in the image are denoted as outlier pixels; obtain the pixel feature codes of all outlier pixels;

[0063] All marker contours BL are analyzed sequentially using the contour pixel comparison method. The contour pixel comparison method is as follows: for any marker contour BL, when the pixel feature code of a contour pixel A corresponding to the marker contour BL matches the pixel feature code of the out-of-class contour YL, the pixel feature code of the marker contour BL matches the pixel feature code of the out-of-class contour YL. b When the pixel feature codes of any corresponding heterogeneous pixel B are the same, the contour pixel A is recorded as the same pixel outside heterogeneous pixel B, and the difference between the gray value of the contour pixel A and the gray value of heterogeneous pixel B is recorded as the difference between the same pixels outside.

[0064] In a specific implementation process, for example, during a data processing operation, the pixel feature encoding of a contour pixel A in the marker contour BL is [232, 234, 32, 32, 67, 32, 44, 100], and the heterogeneous contour YL is obtained. b If there is an out-of-class pixel B with pixel feature encoding [232, 234, 32, 32, 67, 32, 44, 100], then it indicates that the marker contour BL and the out-of-class contour YL are similar. b There is a possibility that the contours are the same and the positions are the same. Therefore, contour pixel A and out-of-type pixel B can be marked. That is, contour pixel A is recorded as the same pixel outside out-of-type pixel B, so as to carry out subsequent analysis.

[0065] When a different type of pixel B has multiple corresponding pixels, the corresponding pixel with the smallest difference is recorded as a pixel of the same type as the different type of pixel B. When a different type of pixel B has only one corresponding pixel, the corresponding pixel is recorded as a pixel of the same type as the different type of pixel B. Specifically, when any pixel of the same type as a different type of pixel B is obtained, it is used in subsequent analysis of the different type contour YL. b When analyzing outlier pixels other than outlier pixel B, remove outlier pixels of the same type as outlier pixel B from the outline pixels corresponding to the marker outline BL.

[0066] In the specific implementation process, by recording the outermost pixel with the smallest difference as a pixel of the same class as a pixel of the same class from a different class, the probability of pixels of the same class from a different class being the same as pixels of the same class is maximized, thus improving the accuracy of the analysis; and by subsequently analyzing the different contours YL b When analyzing outlier pixels other than outlier pixel B, pixels of the same type as outlier pixel B are removed from the contour pixels corresponding to the marker contour BL. The purpose is to prevent the same type of pixels of the same outlier pixel from being the same, which would lead to analysis errors and affect the acquisition of the contour to be determined.

[0067] When the pixel feature code of all contour pixels in the mark contour BL matches the heterogeneous contour YL b When the pixel feature codes of any out-of-type pixel C are all different, the out-of-type pixel C is recorded as an independent pixel.

[0068] The out-of-type contour YL b The similarity value is calculated by subtracting the number of independent pixels from the number of dissimilar pixels that contain the same type of pixels.

[0069] After analyzing all marker contours BL using the contour pixel comparison method, obtain the similarity values ​​of the same category, and record the marker contour BL corresponding to the largest similarity value as the undetermined contour; when the similarity value of the undetermined contour is greater than or equal to 0, record the undetermined contour as the out-of-category contour YL. b Lateral analysis features; when the similarity value of the same type corresponding to the undetermined contour is less than 0, no dissimilar contour YL is set. b Lateral analysis features;

[0070] In the specific implementation process, for example, in a data processing operation, a heterogeneous contour YL is obtained. b like Figure 4 As shown, the contour in region PP1 contains heterogeneous pixels with the same type of pixels, and the number of heterogeneous pixels in region PP1 is 200. The contour in region PP2 contains independent pixels, and the number of independent pixels in region PP2 is 134. Calculations show that the value of the same type of pixels is greater than 0, indicating that the marker contour BL contains a large number of pixels that are heterogeneous with the contour YL. b The pixels in the same area are identical, so they can be further filtered by being recorded as undetermined contours.

[0071] All out-of-class contours YL that do not contain lateral analysis features are denoted as image recognition features of the target marker α, and all out-of-class contours YL that contain lateral analysis features are denoted as analysis markers FB1 to FB1. n Provisional image features;

[0072] Analyze all outlier marker images to obtain all image recognition features of the marker to be identified α and provisional image features of outlier marker FB in all outlier marker images;

[0073] Based on the analysis method for the marker α to be identified, all markers to be identified are analyzed.

[0074] The marker recognition module is used to optimize the image recognition features of all markers to be identified, and obtain the preferred recognition features for each marker based on the optimization results. The preferred recognition features are then used to perform image recognition of lung tumor markers.

[0075] The marker recognition module includes a marker recognition unit, which is configured with a marker recognition strategy. The marker recognition strategy includes: using a marker optimization method to optimize the image recognition features of all markers to be recognized.

[0076] The marker optimization method is as follows: For any marker α to be identified, the feature optimization algorithm is used to obtain the effective recognition parameters of each image recognition feature of the marker α. For any image recognition feature β of the marker α to be identified, the feature optimization algorithm is as follows: Where F is the effective recognition parameter, f2 is the number of image recognition features β among all image recognition features of the marker α to be identified, f1 is the number of all image recognition features of the marker α to be identified, f3 is the number of outlier marker images when the marker α to be identified is analyzed as marker FB, and f4 is the number of provisional image features of the marker α to be identified as image recognition features β when outlier marker images are analyzed.

[0077] In a specific implementation process, for example, during a data analysis, the number of all image recognition features of the marker α to be identified is 5, the number of image recognition features β among all image recognition features of the marker α to be identified is 3, the number of outlier marker images when the marker α to be identified is FB is 10, and the number of provisional image recognition features of the marker α to be identified as image recognition feature β when the outlier marker images when the marker α to be identified is FB is 5, then by calculation, the effective recognition parameter of the marker α to be identified for image recognition feature β is 1.1. By obtaining the effective recognition parameter, the correlation between the marker α to be identified and image recognition feature β can be judged from the image recognition features of the marker α to be identified and the provisional image features of the marker α to be identified. The larger the effective recognition parameter, the stronger the correlation between the marker α to be identified and image recognition feature β, and the more likely it is that image recognition feature β will be used as the preferred recognition feature of the marker α to be identified.

[0078] The image feature with the largest effective recognition parameter of the marker α to be identified is recorded as the preferred recognition feature of the marker α to be identified;

[0079] The preferred identification features of all markers to be identified are obtained, and when performing image recognition of lung tumor markers, the lung tumor markers are identified based on the contours corresponding to the preferred identification features of each marker.

[0080] Example 2, please refer to Figure 2 As shown, this application also provides a medical image recognition method for lung tumor markers, comprising the following steps:

[0081] Step S1: Record the patient's lung tumor image as the patient marker image; obtain the lung tumor markers present in the patient's body based on blood tests and record them as markers to be identified; put all markers to be identified into the marker set.

[0082] Step S2: For any marker α to be identified, the markers in the marker set other than the marker α to be identified are recorded as heterogeneous markers of the marker α to be identified; lung tumor images of patients in which only heterogeneous markers exist in the body are obtained from the database and recorded as heterogeneous marker images; the marker analysis method is used to analyze the patient marker images and heterogeneous marker images, and the image recognition features of the marker α to be identified are obtained based on the analysis results.

[0083] Obtain the image recognition features of all the markers to be identified;

[0084] The marker analysis method includes: step S201, converting the patient marker image to grayscale and recording the grayscale patient marker image as a grayscale marker image; performing binarization on the grayscale marker image, wherein the binarization is as follows: for any pixel in the grayscale marker image, when the grayscale value of the pixel is less than the equilibrium grayscale value, the grayscale value of the pixel is adjusted to a very black grayscale value; when the grayscale value of the pixel is greater than or equal to the equilibrium grayscale value, the grayscale value of the pixel is adjusted to a very white grayscale value;

[0085] Step S202: Obtain all closed contours in the binarized grayscale marker image and record them sequentially as marker contour BL1 to marker contour BL2. c For any given marker outline BL v v is a positive integer less than or equal to c and greater than or equal to 1, marking the outline BL. v The position is marked in the grayscale marker image, and the position of the marker within the marker outline BL in the grayscale marker image is marked. v The pixels in the image are denoted as outline pixels;

[0086] Step S203: For any contour pixel, obtain the gray values ​​of all pixels in the eight neighborhoods of the contour pixel, and sort the gray values ​​of the pixels in the eight neighborhoods of the contour pixel in a clockwise direction, starting from the gray value of the pixel directly above the contour pixel. The set corresponding to the eight gray values ​​is recorded as the pixel feature code of the contour pixel; obtain the pixel feature codes of all contour pixels.

[0087] Step S204: Obtain the pixel feature codes of all contour pixels in the grayscale marker image corresponding to all marker contours BL;

[0088] Step S205: For any acquired heterologous marker image, the heterologous markers in the patient's body corresponding to the heterologous marker image are sequentially recorded as analytical marker FB1 to analytical marker FB. n The outlier image is converted to grayscale, and the grayscale outlier image is denoted as the grayscale outlier image. The grayscale outlier image is then binarized, and multiple closed contours corresponding to the binarized outlier image are obtained and sequentially denoted as outlier contour YL1 to outlier contour YL. r ;

[0089] Step S206, for any heterogeneous contour YL b b is a positive integer less than or equal to r and greater than or equal to 1; the grayscale outlier image containing the outlier contour YL b The pixels in the image are denoted as outlier pixels; obtain the pixel feature codes of all outlier pixels;

[0090] Step S207: Analyze all marker contours BL sequentially using the contour pixel comparison method. The contour pixel comparison method is as follows: Step S2071: For any marker contour BL, when the pixel feature code of a contour pixel A corresponding to the marker contour BL matches the pixel feature code of the heterogeneous contour YL... b When the pixel feature codes of any corresponding heterogeneous pixel B are the same, the contour pixel A is recorded as the same pixel outside heterogeneous pixel B, and the difference between the gray value of the contour pixel A and the gray value of heterogeneous pixel B is recorded as the difference between the same pixels outside.

[0091] Step S2072: When multiple similar pixels exist for a different type of pixel B, the similar pixel with the smallest difference is recorded as a similar pixel of the different type of pixel B; when only one similar pixel exists for a different type of pixel B, the similar pixel of the different type of pixel B is recorded as a similar pixel of the different type of pixel B. Specifically, when any similar pixel of a different type of pixel B is obtained, it is used in subsequent processing of the different type contour YL... b When analyzing outlier pixels other than outlier pixel B, remove outlier pixels of the same type as outlier pixel B from the outline pixels corresponding to the marker outline BL.

[0092] Step S2073, when the pixel feature code of all contour pixels in the marker contour BL matches the heterogeneous contour YL b When the pixel feature codes of any out-of-type pixel C are all different, the out-of-type pixel C is recorded as an independent pixel.

[0093] Step S2074, the heterogeneous contour YL b The similarity value is calculated by subtracting the number of independent pixels from the number of dissimilar pixels that contain the same type of pixels.

[0094] Step S208: Obtain the similarity values ​​of all marker contours BL after analysis by the contour pixel comparison method, and record the marker contour BL corresponding to the largest similarity value as the undetermined contour; when the similarity value of the undetermined contour is greater than or equal to 0, record the undetermined contour as the heterogeneous contour YL. b Lateral analysis features; when the similarity value of the same type corresponding to the undetermined contour is less than 0, no dissimilar contour YL is set. b Lateral analysis features;

[0095] Step S209: Record all out-of-type contours YL that do not contain lateral analysis features as image recognition features of the target marker α, and record all out-of-type contours YL that contain lateral analysis features as analysis markers FB1 to FB1. n Provisional image features;

[0096] Step S210: Analyze all outlier marker images to obtain all image recognition features of the marker to be identified α and provisional image features of outlier marker FB in all outlier marker images;

[0097] Step S211: Based on the analysis method of the marker α to be identified, analyze all the markers to be identified.

[0098] Step S3: Optimize the image recognition features of all markers to be identified, and obtain the preferred recognition features for each marker based on the optimization results. Use the preferred recognition features to perform image recognition of lung tumor markers.

[0099] Step S3 includes: Step S301, using a marker optimization method to optimize the image recognition features of all markers to be identified;

[0100] Step S302, the marker optimization method is as follows: Step S3021, for any marker α to be identified, the effective recognition parameters of each image recognition feature of the marker α are obtained using a feature optimization algorithm. For any image recognition feature β of the marker α to be identified, the feature optimization algorithm is as follows: Where F is the effective recognition parameter, f2 is the number of image recognition features β among all image recognition features of the marker α to be identified, f1 is the number of all image recognition features of the marker α to be identified, f3 is the number of outlier marker images when the marker α to be identified is analyzed as marker FB, and f4 is the number of provisional image features of the marker α to be identified as image recognition features β when outlier marker images are analyzed.

[0101] Step S303: Record the image recognition feature with the largest effective recognition parameter of the marker α to be recognized as the preferred recognition feature of the marker α to be recognized;

[0102] Step S304: Obtain the preferred identification features of all markers to be identified, and when performing image recognition of lung tumor markers, identify the lung tumor markers based on the contour corresponding to the preferred identification features of each marker to be identified.

[0103] Example 3, please refer to Figure 5 As shown, Figure 5 A schematic diagram of an electronic device is provided, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory stores computer-readable instructions, and the processor can call the instructions in the memory. When the computer-readable instructions are executed by the processor, steps such as those in a medical image recognition method for lung tumor markers are performed to achieve the following functions: First, the markers to be identified are obtained based on blood tests, and all markers to be identified are placed into a marker set; then, for any marker α to be identified, outlier markers of marker α are obtained; outlier marker images are obtained based on a database; marker analysis is used to analyze the patient marker images and outlier marker images, and image recognition features of marker α to be identified are obtained based on the analysis results, and then image recognition features of all markers to be identified are obtained; furthermore, the image recognition features of all markers to be identified are optimized, and preferred recognition features of each marker to be identified are obtained based on the optimization results; and the preferred recognition features are used to perform image recognition of lung tumor markers.

[0104] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0105] Example 4: This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it performs the steps of the above-mentioned medical image recognition method for lung tumor markers to achieve the following functions: First, obtain the markers to be identified based on blood testing, and place all the markers to be identified into a marker set; then, for any marker α to be identified, obtain the heterologous markers of marker α; obtain heterologous marker images based on a database; analyze the patient marker images and heterologous marker images using a marker analysis method, obtain the image recognition features of marker α to be identified based on the analysis results, and then obtain the image recognition features of all the markers to be identified; furthermore, optimize the image recognition features of all the markers to be identified, and obtain the preferred recognition features of each marker to be identified based on the optimization results, and use the preferred recognition features to perform image recognition of lung tumor markers.

[0106] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the technical solutions described above, or the parts that contribute to the prior art, can be embodied in the form of software products. These computer software products can be stored in computer-readable storage media, such as ROM / RAM, magnetic disks, optical disks, etc., and include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0107] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A medical image recognition method for lung tumor markers, characterized in that, Includes the following steps: The patient's lung tumor image is recorded as the patient's marker image; lung tumor markers present in the patient's body are obtained based on blood tests and recorded as markers to be identified; all markers to be identified are placed into a marker set. For any marker α to be identified, markers other than α in the marker set are denoted as heterogeneous markers of α; lung tumor images of patients with only heterogeneous markers in their bodies are obtained from the database and denoted as heterogeneous marker images; marker analysis is used to analyze the patient marker images and heterogeneous marker images, and the image recognition features of marker α are obtained based on the analysis results. Obtain the image recognition features of all the markers to be identified; The image recognition features of all markers to be identified are optimized, and the preferred recognition features of each marker to be identified are obtained based on the optimization results. The preferred recognition features are then used to perform image recognition of lung tumor markers.

2. The medical image recognition method for lung tumor markers according to claim 1, characterized in that, The marker analysis method includes: The patient marker image is converted to grayscale and denoted as the grayscale marker image. The grayscale marker image is then binarized as follows: for any pixel in the grayscale marker image, if the pixel's grayscale value is less than the equilibrium grayscale value, the pixel's grayscale value is adjusted to a very dark grayscale value; if the pixel's grayscale value is greater than or equal to the equilibrium grayscale value, the pixel's grayscale value is adjusted to a very white grayscale value. Obtain all closed contours in the binarized grayscale marker image and label them sequentially as marker contour BL1 to marker contour BL2. c For any given marker outline BL v v is a positive integer less than or equal to c and greater than or equal to 1, marking the outline BL. v The position is marked in the grayscale marker image, and the position of the marker within the marker outline BL in the grayscale marker image is marked. v The pixels in the diagram are denoted as outline pixels.

3. The medical image recognition method for lung tumor markers according to claim 2, characterized in that, The marker analysis method also includes: For any contour pixel, obtain the gray values ​​of all pixels in the eight neighborhoods of the contour pixel, and sort the gray values ​​of the pixels in the eight neighborhoods of the contour pixel in a clockwise direction, starting from the gray value of the pixel directly above the contour pixel. The set corresponding to the eight gray values ​​is recorded as the pixel feature code of the contour pixel; obtain the pixel feature codes of all contour pixels. Obtain the pixel feature codes of all contour pixels in the grayscale logo image corresponding to all logo contours BL.

4. The medical image recognition method for lung tumor markers according to claim 3, characterized in that, The marker analysis method also includes: For any acquired heterologous marker image, the heterologous markers in the patient's body corresponding to the heterologous marker image are sequentially denoted as analytical marker FB1 to analytical marker FB2. n The outlier image is converted to grayscale, and the grayscale outlier image is denoted as the grayscale outlier image. The grayscale outlier image is then binarized, and multiple closed contours corresponding to the binarized outlier image are obtained and sequentially denoted as outlier contour YL1 to outlier contour YL. r ; For any out-of-type contour YL b b is a positive integer less than or equal to r and greater than or equal to 1; the grayscale outlier image containing the outlier contour YL b The pixels in the image are denoted as outlier pixels; obtain the pixel feature codes of all outlier pixels; All marker contours BL are analyzed sequentially using the contour pixel comparison method. The contour pixel comparison method is as follows: for any marker contour BL, when the pixel feature code of a contour pixel A corresponding to the marker contour BL matches the pixel feature code of the out-of-class contour YL, the pixel feature code of the marker contour BL matches the pixel feature code of the out-of-class contour YL. b When the pixel feature codes of any corresponding heterogeneous pixel B are the same, the contour pixel A is recorded as the same pixel outside heterogeneous pixel B, and the difference between the gray value of the contour pixel A and the gray value of heterogeneous pixel B is recorded as the difference between the same pixels.

5. A medical image recognition method for lung tumor markers according to claim 4, characterized in that, The marker analysis method also includes: When a different type of pixel B has multiple corresponding pixels, the corresponding pixel with the smallest difference is recorded as a pixel of the same type as the different type of pixel B. When a different type of pixel B has only one corresponding pixel, the corresponding pixel is recorded as a pixel of the same type as the different type of pixel B. Specifically, when any pixel of the same type as a different type of pixel B is obtained, it is used in subsequent analysis of the different type contour YL. b When analyzing outlier pixels other than outlier pixel B, remove outlier pixels of the same type as outlier pixel B from the outline pixels corresponding to the marker outline BL. When the pixel feature code of all contour pixels in the mark contour BL matches the heterogeneous contour YL b When the pixel feature codes of any out-of-type pixel C are all different, the out-of-type pixel C is recorded as an independent pixel. The outlier contour YL b The similarity value is the difference between the number of dissimilar pixels containing the same type of pixels and the number of independent pixels.

6. A medical image recognition method for lung tumor markers according to claim 5, characterized in that, The marker analysis method also includes: After analyzing all marker contours BL using the contour pixel comparison method, obtain the similarity values ​​of the same category, and record the marker contour BL corresponding to the largest similarity value as the undetermined contour; when the similarity value of the undetermined contour is greater than or equal to 0, record the undetermined contour as the out-of-category contour YL. b Lateral analysis features; when the similarity value of the same type corresponding to the undetermined contour is less than 0, no heterogeneous contour YL is set. b Lateral analysis features; All out-of-class contours YL that do not contain lateral analysis features are denoted as image recognition features of the target marker α, and all out-of-class contours YL that contain lateral analysis features are denoted as analysis markers FB1 to FB1. n Provisional image features; Analyze all outlier marker images to obtain all image recognition features of the marker to be identified α and provisional image features of outlier marker FB in all outlier marker images; Based on the analysis method for the marker α to be identified, all markers to be identified are analyzed.

7. A medical image recognition method for lung tumor markers according to claim 6, characterized in that, The image recognition features of all biomarkers to be identified are optimized, and the preferred recognition features for each biomarker are obtained based on the optimization results. Image recognition of lung tumor biomarkers using the preferred recognition features includes: The image recognition features of all the markers to be identified are optimized using a marker optimization method; The marker optimization method is as follows: For any marker α to be identified, the feature optimization algorithm is used to obtain the effective recognition parameters of each image recognition feature of the marker α. For any image recognition feature β of the marker α to be identified, the feature optimization algorithm is as follows: Where F is the effective recognition parameter, f2 is the number of image recognition features β among all image recognition features of the marker α to be identified, f1 is the number of all image recognition features of the marker α to be identified, f3 is the number of outlier marker images when the marker α to be identified is analyzed as marker FB, and f4 is the number of provisional image features of the marker α to be identified as image recognition features β when outlier marker images are analyzed.

8. A medical image recognition method for lung tumor markers according to claim 7, characterized in that, The image recognition features of all biomarkers to be identified are optimized, and the preferred recognition features for each biomarker are obtained based on the optimization results. Image recognition of lung tumor biomarkers using the preferred recognition features also includes: The image feature with the largest effective recognition parameter of the marker α to be identified is recorded as the preferred recognition feature of the marker α to be identified; The preferred identification features of all markers to be identified are obtained, and when performing image recognition of lung tumor markers, the lung tumor markers are identified based on the contours corresponding to the preferred identification features of each marker.

9. A medical image recognition system for lung tumor markers, used to implement the medical image recognition method for lung tumor markers as described in any one of claims 1-8, characterized in that, It includes a marker set acquisition module, a recognition feature analysis module, and a marker recognition module; The biomarker set acquisition module is used to record the patient's lung tumor image as the patient's biomarker image; to obtain the lung tumor biomarkers present in the patient's body based on blood tests and record them as biomarkers to be identified, and to put all biomarkers to be identified into the biomarker set; The feature analysis module is used to identify any marker α, and to denote markers other than α in the marker set as heterogeneous markers of α; to obtain lung tumor images of patients in which only heterogeneous markers are present in the body based on the database, and to denote them as heterogeneous marker images; to analyze the patient marker images and heterogeneous marker images using the marker analysis method, and to obtain the image recognition features of marker α based on the analysis results; Obtain the image recognition features of all the markers to be identified; The marker recognition module is used to optimize the image recognition features of all markers to be identified, and obtain the preferred recognition features for each marker based on the optimization results. The preferred recognition features are then used to perform image recognition of lung tumor markers.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it performs the steps of the method as described in any one of claims 1-8.

Images