Pulmonary ground-glass nodule recognition method, system, terminal and storage medium

Abstract

The present invention provides a method, system, terminal and storage medium for identifying pulmonary ground-glass nodules, including: using a pre-constructed image recognition model to identify pulmonary ground-glass nodules from CT images to obtain an image recognition result; receiving the detection result of serum tumor markers, and inputting the image recognition result and the detection result of serum tumor markers into a pre-constructed clustering detection model to obtain an identification result of pulmonary ground-glass nodules. The present invention greatly improves the identification accuracy of pulmonary ground-glass nodules and provides objective data support for the diagnosis of pulmonary ground-glass nodules.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a method, system, terminal and storage medium for identifying pulmonary ground glass nodules. Background Art

[0002] With the popularization of high-resolution CT, a large number of pulmonary nodules have been discovered, and a large part of the ground glass nodules (GGN) have finally been confirmed as early lung cancer. Since ground glass nodules are relatively blurred in CT, and the treatment methods for ground glass nodules of different degrees are different, if the degree of ground glass nodules is not accurately grasped, it will lead to improper treatment methods and cause adverse effects on patients. The existing ground glass nodules rely heavily on doctors' experience, resulting in poor diagnostic accuracy of ground glass nodules. Summary of the Invention

[0003] In view of the above deficiencies of the prior art, the present invention provides a method, system, terminal and storage medium for identifying pulmonary ground glass nodules to solve the above technical problems.

[0004] In a first aspect, the present invention provides a method for identifying pulmonary ground glass nodules, including:

[0005] Identifying pulmonary ground glass nodules from CT images by using a pre-constructed image recognition model to obtain an image recognition result;

[0006] Receiving the detection results of serum tumor markers, and inputting the image recognition result and the detection results of serum tumor markers into a pre-constructed clustering detection model to obtain an identification result of pulmonary ground glass nodules.

[0007] Further, the image recognition model includes:

[0008] A contour detection sub-model for identifying contour lines in CT images by using the canny contour detection algorithm, screening out closed contour lines from all contour lines, and segmenting the region where the closed contour lines are located from the CT image as the recognition target;

[0009] A neural network sub-model for processing the recognition target to obtain the type and probability of the pulmonary ground glass nodules to which the recognition target belongs.

[0010] Further, the training method of the neural network sub-model includes:

[0011] Collecting a large number of CT images, performing edge detection on all CT images, extracting closed contour line regions therefrom, marking the types of pulmonary ground glass nodules to which the closed contour line regions belong, and saving them to a training set;

[0012] Training the neural network sub-model by using the training set.

[0013] Further, receive the serum tumor marker test results, and input the image recognition results and the serum tumor marker test results into a pre-constructed clustering detection model to obtain the pulmonary ground-glass nodule recognition results, including:

[0014] Receive the serum tumor marker test results uploaded by the user from the external interface, and the serum tumor marker test results include the detection results of the contents of carcinoembryonic antigen and carbohydrate antigen;

[0015] Select the maximum probability from the probabilities of all recognition targets output by the neural network sub-model;

[0016] Input the serum tumor marker test results and the type of pulmonary ground-glass nodule corresponding to the maximum probability into a pre-constructed clustering detection model to obtain the pulmonary ground-glass nodule recognition results.

[0017] Further, the construction method of the clustering detection model includes:

[0018] Collect a large amount of diagnostic data of patients with pulmonary ground-glass nodules and diagnostic data of patients without pulmonary ground-glass nodules. The diagnostic data includes image recognition results and serum tumor marker test results, and save all the collected data to the clustering training set;

[0019] Construct a K-Means clustering model, and specify the clustering center of the K-Means clustering model as the number of types of pulmonary ground-glass nodules;

[0020] Use the clustering training set to train the K-Means clustering model.

[0021] In a second aspect, the present invention provides a pulmonary ground-glass nodule recognition system, including:

[0022] An image recognition unit, configured to recognize pulmonary ground-glass nodules from CT images by using a pre-constructed image recognition model to obtain image recognition results;

[0023] A result recognition unit, configured to receive the serum tumor marker test results, and input the image recognition results and the serum tumor marker test results into a pre-constructed clustering detection model to obtain the pulmonary ground-glass nodule recognition results.

[0024] Further, the image recognition model includes:

[0025] A contour detection sub-model, configured to use the canny contour detection algorithm to recognize the contour lines in the CT image, screen out the closed contour lines from all the contour lines, and segment the area where the closed contour lines are located from the CT image as the recognition target;

[0026] A neural network sub-model for processing the recognition target to obtain the type and probability of the ground-glass nodule in the lung to which the recognition target belongs.

[0027] Further, the training method of the neural network sub-model includes:

[0028] Collect a large number of CT images, perform edge detection on all CT images, extract the closed contour line regions therefrom, mark the types of ground-glass nodules in the lung to which the closed contour line regions belong, and save them to the training set.

[0029] Use the training set to train the neural network sub-model.

[0030] Further, the result recognition unit is used for:

[0031] Receive the serum tumor marker test results uploaded by the user from the external interface, and the serum tumor marker test results include the test results of the carcinoembryonic antigen and carbohydrate antigen contents.

[0032] Select the maximum probability from the probabilities of all recognition targets output by the neural network sub-model.

[0033] Input the serum tumor marker test results and the type of ground-glass nodule corresponding to the maximum probability into a pre-constructed clustering detection model to obtain the ground-glass nodule recognition result.

[0034] Further, the construction method of the clustering detection model includes:

[0035] Collect the diagnostic data of a large number of patients with ground-glass nodules in the lung and the diagnostic data of patients without ground-glass nodules in the lung. The diagnostic data includes the image recognition results and the serum tumor marker test results, and save all the collected data to the clustering training set.

[0036] Construct a K-Means clustering model and specify the clustering center of the K-Means clustering model as the number of types of ground-glass nodules in the lung.

[0037] Use the clustering training set to train the K-Means clustering model.

[0038] In a third aspect, a terminal is provided, including:

[0039] A processor and a memory, wherein,

[0040] The memory is used to store a computer program,

[0041] The processor is used to call and run the computer program from the memory, so that the terminal executes the method of the above terminal.

[0042] Fourthly, a computer storage medium is provided. Instructions are stored in the computer-readable storage medium, and when they run on a computer, the computer is made to execute the methods described in the above aspects.

[0043] The beneficial effects of the present invention are as follows.

[0044] The method, system, terminal and storage medium for identifying pulmonary ground-glass nodules provided by the present invention utilize a neural network model to identify CT images, and based on the image recognition results, in combination with the detection results of serum tumor markers, an accurate recognition result of pulmonary ground-glass nodules is obtained through a clustering detection model. The present invention greatly improves the recognition accuracy of pulmonary ground-glass nodules and provides objective data support for the diagnosis of pulmonary ground-glass nodules.

[0045] In addition, the design principle of the present invention is reliable, the structure is simple, and it has a very broad application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0047] Figure 1 It is a schematic flow chart of the method according to an embodiment of the present invention.

[0048] Figure 2 It is a schematic block diagram of the system according to an embodiment of the present invention.

[0049] Figure 3 It is a schematic structural diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] Figure 1 It is a schematic flow chart of the method according to an embodiment of the present invention. Among them, Figure 1 The execution subject can be a system for identifying pulmonary ground-glass nodules.

[0052] As Figure 1 shown, the method includes:

[0053] Step 110: Use a pre-constructed image recognition model to identify pulmonary ground-glass nodules from CT images, and obtain an image recognition result.

[0054] Step 120: Receive the serum tumor marker test result, and input the image recognition result and the serum tumor marker test result into a pre-constructed clustering detection model to obtain a pulmonary ground-glass nodule recognition result.

[0055] This method uses a neural network model to identify CT images, and based on the image recognition result, combines the serum tumor marker test result, and obtains an accurate pulmonary ground-glass nodule recognition result through a clustering detection model. The present invention greatly improves the recognition accuracy of pulmonary ground-glass nodules and provides objective data support for the diagnosis of pulmonary ground-glass nodules.

[0056] For the convenience of understanding the present invention, the principle of the pulmonary ground-glass nodule recognition method of the present invention will be further described below in combination with the process of recognizing pulmonary ground-glass nodules in the embodiments.

[0057] Specifically, the pulmonary ground-glass nodule recognition method includes:

[0058] S1. Use a pre-constructed image recognition model to identify pulmonary ground-glass nodules from CT images, and obtain an image recognition result.

[0059] In this embodiment, pulmonary ground-glass nodules are classified into the following categories: pure ground-glass density nodules (type Ⅰ), mixed density nodules with a solid component less than 50% (type Ⅱ), mixed density nodules with a solid component greater than or equal to 50% (type Ⅲ), scattered solid density nodules (type Ⅳ), and solid nodules (type Ⅴ). For pure ground-glass nodules (pGGN), the maximum diameter of 10 mm is used as the cut-off value to distinguish invasive adenocarcinoma and pre-invasive lesions, and the sensitivity and specificity are 53.33% and 100% respectively; for part-solid nodules, the maximum diameter of the solid component of 3 mm is used to diagnose pre-invasive lesions and MIA, and the specificity is 100%. The ground-glass density is an important sign for predicting the prognosis.

[0060] Since pulmonary ground-glass nodules may only occupy a partial area in CT images, recognizing the entire CT image will result in very low recognition accuracy. Therefore, the image recognition model provided by the present invention includes: a contour detection sub-model, which is used to use the canny contour detection algorithm to identify the contour lines in the CT image, screen out the closed contour lines from all the contour lines, and segment the area where the closed contour lines are located from the CT image as the recognition target; a neural network sub-model, which is used to process the recognition target to obtain the type and probability of the pulmonary ground-glass nodules to which the recognition target belongs.

[0061] Among them, the Canny contour detection algorithm for identifying contour lines in CT images includes:

[0062] 1) Image denoising. Since noise is where the gray level changes greatly, it is easily recognized as a false edge.

[0063] 2) Calculate the image gradient to obtain possible edges. Calculating the image gradient can obtain the edges of the image because the gradient is where the gray level changes significantly, and the edges are also where the gray level changes significantly.

[0064] 3) Non-maximum suppression. Usually, the areas where the gray level changes are relatively concentrated. Keep the one with the largest gray level change in the gradient direction within the local range, and do not keep the others, so that a large number of points can be removed. Turn the edge with multiple pixel widths into an edge with a single pixel width. That is, the "fat edge" becomes the "thin edge".

[0065] 4) Dual-threshold screening. After non-maximum suppression, there are still many possible edge points. Further set a dual threshold, namely the low threshold (low) and the high threshold (high). Those with a gray level change greater than high are set as strong edge pixels, and those lower than low are removed. Those between low and high are set as weak edges. Further judge, if there are strong edge pixels in its neighborhood, keep them, otherwise, remove them.

[0066] Based on the above steps, the contour lines in the CT image are obtained. The closed contour lines are screened out from all the contour lines, and the area where the closed contour lines are located is segmented from the CT image as the recognition target. There may be multiple recognition targets in one CT image. If there is no recognition target, it is output that there are no nodules.

[0067] The neural network sub-model adopted by the present invention is a convolutional neural network model (CNN). The training method of the neural network sub-model includes: collecting a large number of CT images, performing edge detection on all CT images, extracting the closed contour line area therefrom, marking the type of pulmonary ground-glass nodule to which the closed contour line area belongs, and then saving it to the training set; using the training set to train the neural network sub-model.

[0068] S2. Receive the serum tumor marker detection results, and input the image recognition results and the serum tumor marker detection results into a pre-constructed clustering detection model to obtain the pulmonary ground-glass nodule recognition results.

[0069] Receive the serum tumor marker detection results uploaded by the user from the external interface. The serum tumor marker detection results include the detection results of the contents of carcinoembryonic antigen and carbohydrate antigens (CA125, CA153). The serum tumor marker detection results can improve the sensitivity and specificity of lung cancer diagnosis. Selectively combining CT scanning and serum tumor markers has great value for the early diagnosis of lung cancer.

[0070] Select the maximum probability from the probabilities of all recognition targets output by the neural network sub-model; input the serum tumor marker test result and the type of pulmonary ground-glass nodule corresponding to the maximum probability into a pre-constructed clustering detection model to obtain the recognition result of the pulmonary ground-glass nodule. In this embodiment, the type of pulmonary ground-glass nodule corresponding to the maximum probability is selected as the influencing factor of the final recognition result, reducing the computational amount of the clustering detection model. In other embodiments of the present invention, the types of pulmonary ground-glass nodules with probabilities exceeding the set threshold can also be used as the influencing factors of the final recognition result respectively. That is, if the image recognition result is multiple types of pulmonary ground-glass nodules, the multiple types of pulmonary ground-glass nodules need to be input into the clustering detection model in sequence, and each type of pulmonary ground-glass nodule corresponds to a recognition result of the pulmonary ground-glass nodule.

[0071] The construction method of the clustering detection model includes: collecting a large amount of diagnostic data of patients with pulmonary ground-glass nodules and diagnostic data of patients without pulmonary ground-glass nodules. The diagnostic data includes the image recognition result and the serum tumor marker test result, and saving all the collected data to the clustering training set; constructing a K-Means clustering model and specifying the clustering center of the K-Means clustering model as the number of types of pulmonary ground-glass nodules. In this embodiment, the number of types of pulmonary ground-glass nodules is 5; training the K-Means clustering model using the clustering training set.

[0072] The process of the K-Means clustering algorithm is as follows: (1) Specify the number of clusters to be divided, that is, the K value (the number of classes); (2) Randomly select K data objects as the initial clustering centers; (3) Calculate the distances from the remaining each data object to these K clustering centers, and divide the data objects into the cluster where the center closest to it is located; (4) Adjust the new classes and calculate the new clustering centers; (5) Loop steps (3) and (4) until the clustering centers no longer change and the clustering ends.

[0073] As Figure 2 shown, the system 200 includes:

[0074] An image recognition unit 210, configured to recognize pulmonary ground-glass nodules from CT images using a pre-constructed image recognition model to obtain an image recognition result;

[0075] A result recognition unit 220, configured to receive the serum tumor marker test result and input the image recognition result and the serum tumor marker test result into a pre-constructed clustering detection model to obtain the recognition result of the pulmonary ground-glass nodule.

[0076] Optionally, as an embodiment of the present invention, the image recognition model includes:

[0077] The contour detection sub-model is used to identify the contour lines in the CT image by using the canny contour detection algorithm, screen out the closed contour lines from all the contour lines, and segment the area where the closed contour lines are located from the CT image as the recognition target;

[0078] The neural network sub-model is used to process the recognition target to obtain the type and probability of the pulmonary ground-glass nodule to which the recognition target belongs.

[0079] Optionally, as an embodiment of the present invention, the training method of the neural network sub-model includes:

[0080] Collect a large number of CT images, perform edge detection on all CT images, extract the closed contour line area from them, mark the type of the pulmonary ground-glass nodule to which the closed contour line area belongs, and save it to the training set;

[0081] Use the training set to train the neural network sub-model.

[0082] Optionally, as an embodiment of the present invention, the result recognition unit is used for:

[0083] Receive the serum tumor marker detection results uploaded by the user from the external interface, and the serum tumor marker detection results include the detection results of the carcinoembryonic antigen and carbohydrate antigen contents;

[0084] Select the maximum probability from the probabilities of all recognition targets output by the neural network sub-model;

[0085] Input the serum tumor marker detection results and the type of the pulmonary ground-glass nodule corresponding to the maximum probability into the pre-constructed clustering detection model to obtain the pulmonary ground-glass nodule recognition result.

[0086] Optionally, as an embodiment of the present invention, the construction method of the clustering detection model includes:

[0087] Collect the diagnosis data of a large number of patients with pulmonary ground-glass nodules and the diagnosis data of patients without pulmonary ground-glass nodules. The diagnosis data includes the image recognition results and the serum tumor marker detection results, and save all the collected data to the clustering training set;

[0088] Construct a K-Means clustering model, and specify the clustering center of the K-Means clustering model as the number of types of pulmonary ground-glass nodules;

[0089] Use the clustering training set to train the K-Means clustering model.

[0090] Figure 3 It is a schematic structural diagram of a terminal 300 provided by an embodiment of the present invention. The terminal 300 can be used to execute the pulmonary ground-glass nodule recognition method provided by the embodiment of the present invention.

[0091] Among them, the terminal 300 may include: a processor 310, a memory 320, and a communication unit 330. These components communicate through one or more buses. Those skilled in the art can understand that the structure of the server shown in the figure does not constitute a limitation on the present invention. It can be a bus structure, a star structure, and may also include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0092] Among them, the memory 320 can be used to store the execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile storage terminal or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disc. When the execution instructions in the memory 320 are executed by the processor 310, the terminal 300 can execute some or all of the steps in the above method embodiments.

[0093] The processor 310 is the control center of the storage terminal, connecting various parts of the entire electronic terminal through various interfaces and lines. By running or executing the software programs and / or modules stored in the memory 320, and by calling the data stored in the memory, it executes various functions of the electronic terminal and / or processes data. The processor may be composed of an integrated circuit (IC), for example, it may be composed of a single packaged IC, or may be composed of multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may only include a central processing unit (CPU). In the embodiment of the present invention, the CPU may be a single arithmetic core or may include multiple arithmetic cores.

[0094] The communication unit 330 is used to establish a communication channel, so that the storage terminal can communicate with other terminals. It receives user data sent by other terminals or sends user data to other terminals.

[0095] The present invention also provides a computer storage medium. Among them, the computer storage medium can store a program, and when the program is executed, it can include some or all of the steps in the various embodiments provided by the present invention. The storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc.

[0096] Therefore, the present invention uses a neural network model to identify CT images, and based on the image recognition results, combined with the detection results of serum tumor markers, an accurate recognition result of pulmonary ground-glass nodules is obtained through a clustering detection model. The present invention greatly improves the recognition accuracy of pulmonary ground-glass nodules, provides objective data support for the diagnosis of pulmonary ground-glass nodules, and the technical effects that can be achieved in this embodiment can be referred to the description above and will not be elaborated here.

[0097] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solutions in the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc, etc., which can store program codes, including several instructions for causing a computer terminal (which can be a personal computer, a server, or a second terminal, a network terminal, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0098] For the same or similar parts among the various embodiments in this specification, reference can be made to each other. In particular, for the terminal embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0099] In several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the systems or units can be in electrical, mechanical or other forms.

[0100] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, in each embodiment of the present invention, each functional unit may be integrated in a processing unit, may exist independently as individual units physically, or two or more units may be integrated in one unit.

[0102] Although the present invention has been described in detail by referring to the accompanying drawings and in conjunction with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, those of ordinary skill in the art can make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions should all be within the scope of the present invention. / Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for identifying pulmonary ground-glass nodules, characterized in that Including: Using a pre - constructed image recognition model to identify ground - glass nodules in CT images, and obtaining an image recognition result; Receiving the detection results of serum tumor markers, and inputting the image recognition result and the detection results of serum tumor markers into a pre - constructed clustering detection model to obtain a ground - glass nodule recognition result; The image recognition model includes: A contour detection sub - model, which is used to identify contour lines in CT images using the canny contour detection algorithm, screen out closed contour lines from all contour lines, and segment the area where the closed contour lines are located from the CT image as the recognition target; A neural network sub - model, which is used to process the recognition target to obtain the type and probability of the ground - glass nodules to which the recognition target belongs; Receiving the detection results of serum tumor markers, and inputting the image recognition result and the detection results of serum tumor markers into a pre - constructed clustering detection model to obtain a ground - glass nodule recognition result, including: Receiving the detection results of serum tumor markers uploaded by the user from an external interface, and the detection results of serum tumor markers include the detection results of carcinoembryonic antigen and carbohydrate antigen; Selecting the maximum probability from the probabilities of all recognition targets output by the neural network sub - model; Inputting the detection results of serum tumor markers and the type of ground - glass nodules corresponding to the maximum probability into a pre - constructed clustering detection model to obtain a ground - glass nodule recognition result; The construction method of the clustering detection model includes: Collecting a large amount of diagnostic data of patients with ground - glass nodules and diagnostic data of patients without ground - glass nodules. The diagnostic data includes image recognition results and detection results of serum tumor markers, and saving all the collected data to a clustering training set; Constructing a K - Means clustering model, and specifying the clustering center of the K - Means clustering model as the number of types of ground - glass nodules; Training the K - Means clustering model using the clustering training set.

2. The method according to claim 1, wherein The training method of the neural network sub - model includes: Collecting a large amount of CT images, performing edge detection on all CT images, extracting the closed contour line area from them, marking the type of ground - glass nodules to which the closed contour line area belongs, and then saving it to the training set; Training the neural network sub - model using the training set.

3. A pulmonary ground-glass nodule recognition system, characterized in that, Including: An image recognition unit, which is used to identify ground - glass nodules in CT images using a pre - constructed image recognition model to obtain an image recognition result; A result recognition unit, which is used to receive the detection results of serum tumor markers, and input the image recognition result and the detection results of serum tumor markers into a pre - constructed clustering detection model to obtain a ground - glass nodule recognition result; The image recognition model includes: A contour detection sub - model, which is used to identify contour lines in CT images using the canny contour detection algorithm, screen out closed contour lines from all contour lines, and segment the area where the closed contour lines are located from the CT image as the recognition target; A neural network sub - model, which is used to process the recognition target to obtain the type and probability of the ground - glass nodules to which the recognition target belongs: Receive the test results of serum tumor markers, and input the image recognition results and the test results of serum tumor markers into a pre-constructed clustering detection model to obtain the recognition results of pulmonary ground-glass nodules, including: Receive the test results of serum tumor markers uploaded by the user from the external interface, and the test results of serum tumor markers include the test results of the contents of carcinoembryonic antigen and carbohydrate antigen; Select the maximum probability from the probabilities of all recognition targets output by the neural network sub-model; Input the test results of serum tumor markers and the type of pulmonary ground-glass nodule corresponding to the maximum probability into a pre-constructed clustering detection model to obtain the recognition results of pulmonary ground-glass nodules; The construction method of the clustering detection model includes: Collect a large amount of diagnostic data of patients with pulmonary ground-glass nodules and diagnostic data of patients without pulmonary ground-glass nodules. The diagnostic data includes image recognition results and test results of serum tumor markers, and save all the collected data to the clustering training set; Construct a K-Means clustering model, and specify the clustering center of the K-Means clustering model as the number of types of pulmonary ground-glass nodules; Use the clustering training set to train the K-Means clustering model.

4. The system according to claim 3, characterized in that The training method of the neural network sub-model includes: Collect a large number of CT images, perform edge detection on all CT images, extract the closed contour line area from them, mark the closed contour line area with the type of pulmonary ground-glass nodule to which it belongs, and then save it to the training set; Use the training set to train the neural network sub-model.

5. A terminal, characterized in that, Include: A processor; A memory for storing the execution instructions of the processor; Wherein, the processor is configured to execute the method according to any one of claims 1-2.

6. A computer-readable storage medium storing a computer program, characterized in that, When the program is executed by the processor, it implements the method according to any one of claims 1-2.

Images