Method and device for identifying abnormal region in CT image, medium and product

Abstract

The invention discloses a method and device for recognizing an abnormal region in a CT image, a medium and a product, and relates to the technical field of abnormal region recognition, and the method comprises the steps: decomposing a to-be-detected CT image into a plurality of to-be-detected super voxels through employing a linear iteration clustering algorithm; determining any to-be-tested super voxel as a current to-be-tested super voxel; determining an abnormal value of the current to-be-detected super voxel based on a spectral space function principle and a healthy super voxel library; the abnormal value is used for expressing the probability that the super voxel region in the CT image to be detected is an abnormal region. According to the method, the abnormal region in the CT image is identified based on the spectral-space function, so that the accuracy and efficiency of identifying the abnormal region of the CT image can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of abnormal region recognition, and in particular to a method, device, medium and product for identifying abnormal regions in CT images. Background Art

[0002] In the process of trauma emergency treatment, the type, severity and location of trauma are highly random. Manual film reading will not only cause delays in the rescue process, but may also introduce subjective errors and cause misjudgment or missed judgment. Fast and accurate trauma detection and segmentation algorithms are the basis of clinical intelligent decision-making, which can optimize the rescue process and improve the survival rate of patients.

[0003] At present, some medical image anomaly detection algorithms have emerged, which can automatically detect lesions or damaged areas in images. Supervised medical image anomaly detection algorithms usually require pixel-level or case-level labels to supervise model optimization, but medical image annotation is costly, and manual annotation is not only time-consuming and labor-intensive, but may also introduce annotation errors. Unsupervised medical image anomaly detection algorithms usually reconstruct healthy images of patients, and the difference between the reconstructed healthy images and the input images can reflect the abnormal areas. Related models include methods based on variational autoencoders (VAE), generative adversarial models (GANs), and diffusion models. The variational autoencoder-based method is difficult to restore the normal tissue details in the patient's CT image, causing some normal tissues to be identified as abnormal tissues. The training process of the generative adversarial model is unstable. At present, the diffusion model has become a popular method, but its reconstruction time is long and it is difficult to meet the needs of clinical applications. At the same time, most models focus on tumor detection, and lack research on detection algorithms for multi-size and multi-type injuries. Summary of the invention

[0004] The purpose of this application is to provide a method, device, medium and product for identifying abnormal areas in CT images, which can improve the accuracy and efficiency of abnormal area identification in CT images based on spectral space functions.

[0005] To achieve the above-mentioned purpose, the present application provides the following solutions: In the first aspect, the present application provides a method for identifying abnormal regions in CT images, the method for identifying abnormal regions in CT images is applied to a CT machine, and the method for identifying abnormal regions in CT images includes: constructing a healthy supervoxel library. The healthy supervoxel library includes multiple healthy supervoxel images. Any of the healthy supervoxel images is constructed by decomposing a CT image corresponding to a healthy human body into multiple healthy supervoxels using a linear iterative clustering algorithm.

[0006] Acquire the CT image to be tested.

[0007] The CT image to be tested is decomposed into a plurality of supervoxels to be tested by using a linear iterative clustering algorithm.

[0008] Any supervoxel to be tested is determined as the current supervoxel to be tested.

[0009] Based on the spectral space function principle and the healthy supervoxel library, the abnormal value of the supervoxel to be tested is determined. The abnormal value is used to express the probability that the supervoxel region in the CT image to be tested is an abnormal region.

[0010] The current supervoxel to be tested is updated, and the step of "determining the abnormal value of the current supervoxel to be tested based on the spectral space function principle and the healthy supervoxel library" is returned until all supervoxels to be tested are traversed to obtain the abnormal values ​​of all supervoxels to be tested.

[0011] Optionally, the determining the abnormal value of the current supervoxel to be tested based on the spectral space function principle and the healthy supervoxel library includes: setting the number of iterations i=1.

[0012] Construct an empty set as the target healthy supervoxel set.

[0013] Based on the principle of spectral space function, the healthy supervoxel corresponding to the current supervoxel space to be tested in the i-th healthy supervoxel map is determined as the current target healthy supervoxel.

[0014] The current target healthy supervoxel is taken as the i-th element and added to the target healthy supervoxel set.

[0015] The supervoxel grayscale value of the current healthy supervoxel is determined. The supervoxel grayscale value is the average grayscale value of all voxels in the supervoxel. The supervoxel is a target healthy supervoxel or a supervoxel to be tested.

[0016] Increase the value of the iteration number i by 1, and return to the step "based on the principle of spectral space function, determine the healthy supervoxel in the i-th healthy supervoxel map corresponding to the current supervoxel space to be tested as the current target healthy supervoxel" until the iteration number i is equal to the total number m of healthy supervoxel maps in the healthy supervoxel library.

[0017] Gaussian fitting is performed on the supervoxel grayscale values ​​of the target healthy supervoxel set to determine the supervoxel grayscale value mean and supervoxel grayscale value variance of the target healthy supervoxel set.

[0018] The abnormal value of the supervoxel to be tested is determined according to the supervoxel gray value mean value, the supervoxel gray value variance value and the supervoxel gray value of the supervoxel to be tested currently.

[0019] Optionally, the supervoxel grayscale value mean is: .

[0020] in, is the mean supervoxel grayscale value of the target healthy supervoxel set, is the supervoxel grayscale value of the i-th target healthy supervoxel.

[0021] Optionally, the supervoxel grayscale value variance is: .

[0022] in, is the supervoxel grayscale value variance of the target healthy supervoxel set.

[0023] Optionally, the abnormal value is: .

[0024] in, is the abnormal value of the supervoxel to be tested, is the supervoxel grayscale value of the supervoxel to be tested.

[0025] Optionally, based on the spectral space function principle, determining the healthy supervoxel in the i-th healthy supervoxel map corresponding to the current supervoxel space to be tested as the current target healthy supervoxel includes: The CT image of the i-th healthy human body is determined as the current healthy CT image.

[0026] Based on the principle of spectral space function, the spatial mapping relationship between the CT image to be tested and the current healthy CT image is determined.

[0027] Based on the spatial mapping relationship, the healthy super voxel in the i-th healthy super voxel map corresponding to the current super voxel space to be tested is determined as the current target healthy super voxel.

[0028] Optionally, the spatial mapping relationship is: .

[0029] in, is the spatial mapping relationship between the supervoxels in the CT image to be tested V_p and the healthy CT image V_h, is the spectral basis function of the CT image V_p to be tested, is the low-dimensional spectrum correspondence between the CT image to be tested V_p and the healthy CT image V_h, is the supervoxel feature D of the healthy CT image V_h h The spectral space coordinates of =Φ h D h , is the spectral basis function of the healthy CT image V_h, is the supervoxel feature D of the CT image V_p to be tested p The spectral space coordinates of =Φ p D p ; Spectral basis functions of V_h The pseudo-inverse.

[0030] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned method for identifying abnormal regions in CT images.

[0031] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for identifying abnormal regions in CT images.

[0032] In a fourth aspect, the present application provides a computer program product, including a computer program, which implements the above-mentioned method for identifying abnormal areas in CT images when executed by a processor.

[0033] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a method, device, medium and product for identifying abnormal areas in CT images. The method uses a spectral space function correspondence algorithm to reduce the computational cost and time, and is conducive to the application in medical imaging calculations with large amounts of data; no manual labeling is required, thus avoiding the manpower and material resources spent on data labeling; there is no restriction on the application site and the type of trauma, and the method can detect and segment multiple traumas of clinical concern, including "cerebral hematoma" and "pleural effusion". BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 This is a flow chart of a method for identifying abnormal regions in a CT image in one embodiment of the present application. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0037] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0038] In an exemplary embodiment, Figure 1 As shown, a method for identifying abnormal regions in a CT image is provided. The method for identifying abnormal regions in a CT image is applied to a CT machine. The method for identifying abnormal regions in a CT image includes: Step 101: construct a healthy supervoxel library. The healthy supervoxel library includes multiple healthy supervoxel images. Any healthy supervoxel image is constructed by decomposing a CT image of a corresponding healthy human body into multiple healthy supervoxels using a linear iterative clustering algorithm.

[0039] Step 102: Acquire a CT image to be tested.

[0040] Step 103: Decompose the CT image to be tested into a plurality of supervoxels to be tested using a linear iterative clustering algorithm.

[0041] Step 104: Determine any supervoxel to be tested as the current supervoxel to be tested.

[0042] Step 105: Based on the spectral space function principle and the healthy supervoxel library, determine the abnormal value of the supervoxel to be tested. The abnormal value is used to express the probability that the supervoxel area in the CT image to be tested is an abnormal area.

[0043] Step 106: Update the current supervoxel to be tested, and return to step 105 until all supervoxels to be tested are traversed to obtain the abnormal values ​​of all supervoxels to be tested.

[0044] Step 105 includes: Step 105-1: Set the number of iterations i=1.

[0045] Step 105 - 2: Construct an empty set as the target healthy supervoxel set.

[0046] Step 105-3: Based on the principle of spectral space function, determine the healthy supervoxel in the i-th healthy supervoxel map that corresponds to the current supervoxel space to be tested as the current target healthy supervoxel. Step 105-3 includes: determining the CT image of the i-th healthy human body as the current healthy CT image. Based on the principle of spectral space function, determine the spatial mapping relationship between the CT image to be tested and the current healthy CT image. Based on the spatial mapping relationship, determine the healthy supervoxel in the i-th healthy supervoxel map that corresponds to the current supervoxel space to be tested as the current target healthy supervoxel.

[0047] Among them, the spatial mapping relationship is: .

[0048] in, is the spatial mapping relationship between the supervoxels in the CT image to be tested V_p and the healthy CT image V_h, is the spectral basis function of the CT image V_p to be tested, is the low-dimensional spectrum correspondence between the CT image to be tested V_p and the healthy CT image V_h, is the supervoxel feature D of the healthy CT image V_h h The spectral space coordinates of =Φ h D h , is the spectral basis function of the healthy CT image V_h, is the supervoxel feature D of the CT image V_p to be tested p The spectral space coordinates of =Φ p D p ; Spectral basis functions of V_h The pseudo-inverse.

[0049] Step 105 - 4 : Add the current target healthy supervoxel as the i-th element to the target healthy supervoxel set.

[0050] Step 105-5: Determine the supervoxel grayscale value of the current healthy supervoxel. The supervoxel grayscale value is the average grayscale value of all voxels in the supervoxel. The supervoxel is a target healthy supervoxel or a supervoxel to be tested.

[0051] Step 105 - 6 : Increase the value of the iteration number i by 1, and return to step 105 - 3 until the iteration number i is equal to the total number m of healthy supervoxel maps in the healthy supervoxel library.

[0052] Step 105-7: Perform Gaussian fitting on the supervoxel grayscale values ​​of the target healthy supervoxel set to determine the supervoxel grayscale value mean and supervoxel grayscale value variance of the target healthy supervoxel set. The supervoxel grayscale value mean is: .

[0053] in, is the mean supervoxel grayscale value of the target healthy supervoxel set, is the supervoxel grayscale value of the i-th target healthy supervoxel.

[0054] The supervoxel grayscale value variance is: .

[0055] in, is the supervoxel grayscale value variance of the target healthy supervoxel set.

[0056] Step 105 - 8 : Determine the abnormal value of the current supervoxel to be tested according to the supervoxel grayscale value mean value, the supervoxel grayscale value variance value and the supervoxel grayscale value of the current supervoxel to be tested.

[0057] The outliers are: .

[0058] in, is the abnormal value of the supervoxel to be tested, is the supervoxel grayscale value of the supervoxel to be tested.

[0059] This embodiment uses a spectral domain image registration algorithm to introduce the image information of healthy people, and determines the abnormal (trauma) area on the whole body CT by comparing the apparent differences of corresponding voxels. The spectral function image correspondence technology is used to establish a dense correspondence between the CT images of patients and healthy people, introduce the image information of healthy people, and calculate the voxel grayscale distribution of the healthy people's images. The abnormal area in the patient's CT image is judged based on the voxel grayscale distribution. The supervoxel correspondence between CT images is calculated using spectral correspondence technology to reduce the computational cost and avoid additional rigid registration preprocessing. Below, taking the construction of a healthy supervoxel library of CT images of 20 healthy people as an example, the abnormal area identification method in the CT image shown in this embodiment is specifically described.

[0060] CT images are obtained, including patient CT images (or CT images to be tested) V_p and healthy human CT images V_h. The size of the healthy human CT image library is about 20 cases.

[0061] The patient CT image V_p and the kth healthy human CT image V_h^k are decomposed into supervoxels using a linear iterative clustering algorithm to obtain the patient supervoxel map G h (S h , E h ) and healthy supervoxel map G p (S p , E p ), S h represents the supervoxel set in the patient supervoxel map, E h represents the edge set in the patient supervoxel graph; the number of supervoxels in the patient supervoxel graph is ; S p represents the set of supervoxels in the healthy supervoxel graph, E p represents the edge set in the healthy supervoxel graph; the number of supervoxels in the healthy supervoxel graph is .

[0062] The way to construct a supervoxel graph is to treat each supervoxel as a graph vertex and connect adjacent supervoxels with edges.

[0063] The edge weights of the supervoxel graph are calculated as follows: for a supervoxel and its neighbor supervoxel, the apparent features are recorded as and , these two supervoxels are connected by an edge with an edge weight ,in, represents all the neighbors of this supervoxel.

[0064] In order to distinguish tissues of different densities and tissues of the same density but different locations, this application extracts the apparent features and contextual features of supervoxels respectively. The grayscale distribution histogram of supervoxels is used to describe the apparent features of supervoxels. , where n is the number of supervoxels contained in the CT image, and b is the dimension of the grayscale distribution histogram.

[0065] In order to describe the context of a supervoxel, this application extracts a certain number of neighboring supervoxels around the supervoxel as context supervoxels, and calculates the chi-square distance of the grayscale distribution histogram between the supervoxel and the context supervoxel as the context feature of the supervoxel. , where q is the number of context supervoxels.

[0066] The feature of each supervoxel is the concatenation of the apparent feature and the context feature, i.e., D = {D_his, D_con}. For the patient CT image V_p and the kth healthy human CT image V_h^k, the features D_p and D_h^k are extracted respectively.

[0067] In order to ensure that the spectral space of the image is as stable and compact as possible, this application uses the characteristic function of the CT image supervoxel graph Laplacian operator to define the spectral space of the image. The spectral basis functions of the images V_h and V_p are Φ h and Φ p , the spectral basis function can be used to project the image supervoxel features into the spectral space to obtain the spectral space coordinates of the features =Φ h D h and =Φ p D p The spectral space coordinates of the aligned supervoxel features determine the low-dimensional spectral correspondence C between the images: (1) After obtaining the spectral spatial correspondence C between the two images, the spatial correspondence of the supervoxels can be restored: (2) Elements in Supervoxel With Supervoxel The probability of having a corresponding relationship.

[0068] Using the image supervoxel correspondence algorithm, calculate the correspondence between the patient's CT image and m healthy human CT images The spatial correspondence between the supervoxels (As shown in Equation 2 above), for the supervoxel in the patient's CT image Corresponding healthy supervoxel sets are available Use Gaussian distribution to fit the grayscale distribution of CT images of healthy people, and count the mean and variance of these 20 corresponding supervoxels to get the grayscale distribution of the healthy people image at the supervoxel of this patient. The grayscale distribution of the supervoxel is obtained and supervoxel grayscale value variance By calculating the paralog likelihood function of the supervoxel of the patient's CT image to this distribution, the supervoxel of the patient's CT image can be obtained. Anomaly score: ,The abnormal score includes the probability that the supervoxel is a supervoxel in the trauma area. The higher score indicates that the supervoxel is in the trauma area.

[0069] This embodiment uses image registration technology to establish the image distribution function of healthy people to determine the abnormal tissue area of ​​the patient, and can realize unsupervised automatic labeling of the trauma area, avoiding a large amount of supervised data labeling work; the use of spectral function image correspondence algorithm can avoid errors in the image registration results caused by differences in patient posture and position, and avoid additional image rigid registration preprocessing, while reducing the computational cost.

[0070] In an exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for identifying abnormal areas in a CT image is implemented.

[0071] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0072] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0073] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0074] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0075] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0076] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for identifying abnormal regions in CT images, characterized in that: The abnormal region recognition method in a CT image is applied to a CT machine, and the abnormal region recognition method in a CT image comprises: Constructing a healthy supervoxel library; the healthy supervoxel library includes a plurality of healthy supervoxel images; any of the healthy supervoxel images is constructed by decomposing a CT image corresponding to a healthy human body into a plurality of healthy supervoxels using a linear iterative clustering algorithm; Acquire a CT image to be tested; Decomposing the CT image to be tested into a plurality of supervoxels to be tested by using a linear iterative clustering algorithm; Determine any supervoxel to be tested as the current supervoxel to be tested; Based on the spectral space function principle and the healthy supervoxel library, determining the abnormal value of the current supervoxel to be tested; the abnormal value is used to express the probability that the supervoxel area in the CT image to be tested is an abnormal area; The current supervoxel to be tested is updated, and the process returns to the step of "determining the abnormal value of the current supervoxel to be tested based on the spectral space function principle and the healthy supervoxel library" until all supervoxels to be tested are traversed to obtain the abnormal values ​​of all supervoxels to be tested.

2. The method for identifying abnormal regions in CT images according to claim 1, characterized in that: The determining of the abnormal value of the current supervoxel to be tested based on the spectral space function principle and the healthy supervoxel library comprises: Let the number of iterations i = 1; Construct an empty set as the target healthy supervoxel set; Based on the principle of spectral space function, the healthy supervoxel in the i-th healthy supervoxel map corresponding to the current supervoxel space to be tested is determined as the current target healthy supervoxel; Add the current target healthy supervoxel as the i-th element to the target healthy supervoxel set; Determine a supervoxel grayscale value of a current healthy supervoxel; the supervoxel grayscale value is an average of grayscale values ​​of all voxels in the supervoxel; the supervoxel is a target healthy supervoxel or a supervoxel to be tested; Increase the value of the iteration number i by 1, and return to step "based on the principle of spectral space function, determine the healthy supervoxel in the i-th healthy supervoxel map corresponding to the current supervoxel space to be tested as the current target healthy supervoxel" until the iteration number i is equal to the total number m of healthy supervoxel maps in the healthy supervoxel library; Performing Gaussian fitting processing on the supervoxel grayscale values ​​of the target healthy supervoxel set to determine the supervoxel grayscale value mean and supervoxel grayscale value variance of the target healthy supervoxel set; The abnormal value of the supervoxel to be tested is determined according to the supervoxel gray value mean value, the supervoxel gray value variance value and the supervoxel gray value of the supervoxel to be tested currently.

3. The method for identifying abnormal regions in CT images according to claim 2, characterized in that: The supervoxel grayscale value mean is: ； in, is the mean supervoxel grayscale value of the target healthy supervoxel set, is the supervoxel grayscale value of the i-th target healthy supervoxel.

4. The method for identifying abnormal regions in CT images according to claim 3, characterized in that: The supervoxel grayscale value variance is: ； in, is the supervoxel grayscale value variance of the target healthy supervoxel set.

5. The method for identifying abnormal regions in CT images according to claim 4, characterized in that: The abnormal values ​​are: ； in, is the abnormal value of the supervoxel to be tested, is the supervoxel grayscale value of the supervoxel to be tested.

6. The method for identifying abnormal regions in CT images according to claim 1, characterized in that: Based on the principle of spectral space function, the healthy supervoxel corresponding to the current supervoxel space to be tested in the i-th healthy supervoxel map is determined as the current target healthy supervoxel, including: Determine the CT image of the i-th healthy human body as the current healthy CT image; Based on the principle of spectral space function, the spatial mapping relationship between the CT image to be tested and the current healthy CT image is determined; Based on the spatial mapping relationship, the healthy super voxel in the i-th healthy super voxel map corresponding to the current super voxel space to be tested is determined as the current target healthy super voxel.

7. The method for identifying abnormal regions in CT images according to claim 6, characterized in that: The spatial mapping relationship is: ； in, is the spatial mapping relationship between the supervoxels in the CT image to be tested V_p and the healthy CT image V_h, is the spectral basis function of the CT image V_p to be tested, is the low-dimensional spectrum correspondence between the CT image to be tested V_p and the healthy CT image V_h, , is the supervoxel feature D of the healthy CT image V_h h The spectral space coordinates of =Φ h D h , is the spectral basis function of the healthy CT image V_h, is the supervoxel feature D of the CT image V_p to be tested p The spectral space coordinates of =Φ p D p ; is the spectral basis function of the healthy CT image V_h The pseudo-inverse.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for identifying abnormal areas in CT images according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the abnormal region identification method in a CT image according to any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the abnormal region identification method in a CT image according to any one of claims 1 to 7 is implemented.

Images