Medical image segmentation method and device based on high frequency features

Abstract

The disclosed embodiment provides a medical image segmentation method and device based on high-frequency features, which relates to the field of artificial intelligence technology. The medical image segmentation method based on high-frequency features includes: obtaining medical raw data, which includes original images and segmentation annotation data corresponding to the original images; performing data enhancement processing on the medical raw data to obtain an enhanced image set; inputting the enhanced image set into a trained multi-task network model; performing image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result. The technical solution provided by the disclosed embodiment can enhance the robustness of the image segmentation algorithm and improve the effect of image segmentation.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a medical image segmentation method and device based on high-frequency features. Background Art

[0002] In the field of artificial intelligence, the purpose of medical image segmentation is to segment the parts with special meanings in medical images and extract relevant features to provide a reliable basis for clinical diagnosis and pathological research, and to assist doctors in making more accurate diagnoses. The current medical image segmentation technology has the problems of poor segmentation effect and low robustness of segmentation algorithm. Summary of the invention

[0003] The main purpose of the embodiments of the present disclosure is to propose a medical image segmentation method and device based on high-frequency features, which can enhance the robustness of the image segmentation algorithm and improve the effect of image segmentation.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present disclosure proposes a medical image segmentation method based on high-frequency features, comprising:

[0005] Acquire original medical data, where the original medical data includes an original image and segmentation and annotation data corresponding to the original image;

[0006] Performing data enhancement processing on the raw medical data to obtain an enhanced image set;

[0007] Inputting the enhanced image set into the trained multi-task network model;

[0008] The enhanced image set is subjected to image segmentation processing by the multi-task network model to obtain an image segmentation result; the image segmentation result includes high-frequency feature data and segmentation feature data.

[0009] In some embodiments, the multi-task network model includes a target encoder and a target decoder, and performing image segmentation processing on the enhanced image set by the multi-task network model to obtain an image segmentation result includes:

[0010] Inputting the enhanced image set into the target encoder to generate encoding target data and encoding branch data;

[0011] The encoding target data and the encoding branch data are input into the target decoder to generate the image segmentation result.

[0012] In some embodiments, the target decoder includes a high frequency component decoder and an image segmentation decoder, and the step of inputting the encoded target data and the encoded branch data into the target decoder to generate the image segmentation result includes:

[0013] Inputting the encoding target data and the encoding branch data into the high-frequency component decoder to generate high-frequency feature data and high-frequency decoding branch data;

[0014] The encoding target data and the high-frequency decoding branch data are input into the image segmentation decoder to generate the segmentation feature data.

[0015] In some embodiments, the method further includes training the multi-task network model, specifically including:

[0016] Obtain the initial network model;

[0017] Inputting the enhanced image set into the initial network model to obtain high-frequency initial feature data and segmentation initial feature data;

[0018] Generate a target loss function according to the high-frequency initial feature data and the segmented initial feature data; the target loss function is a loss function of the multi-task network model;

[0019] The initial network model is trained according to the target loss function to obtain the multi-task network model.

[0020] In some embodiments, generating a target loss function according to the high-frequency initial feature data and the segmented initial feature data includes:

[0021] Performing high-pass filtering on the raw medical data to obtain a filtered image;

[0022] Generate a first loss function according to the high-frequency initial feature data and the filtered image; the first loss function is a loss function of a target encoder and a high-frequency component decoder;

[0023] Generate a second loss function according to the segmentation initial feature data and the segmentation annotation data; the second loss function is a loss function of the target encoder and the image segmentation decoder;

[0024] The target loss function is obtained according to the first loss function and the second loss function.

[0025] In some embodiments, obtaining the target loss function according to the first loss function and the second loss function includes:

[0026] Get the loss function generation coefficient;

[0027] The first loss function and the second loss function are weighted according to the loss function generation coefficient to obtain the target loss function.

[0028] In some embodiments, the data enhancement processing includes at least any one of the following: color space adjustment processing and image style transfer processing.

[0029] To achieve the above-mentioned purpose, the second aspect of the present disclosure proposes a medical image segmentation device based on high-frequency features, comprising:

[0030] A medical raw data acquisition module, used to acquire medical raw data, wherein the medical raw data includes an original image and segmentation and annotation data corresponding to the original image;

[0031] A data enhancement module, used for performing data enhancement processing on the raw medical data to obtain an enhanced image set;

[0032] A data input module, used for inputting the enhanced image set into a trained multi-task network model;

[0033] An image segmentation module is used to perform image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result; the image segmentation result includes high-frequency feature data and segmentation feature data.

[0034] To achieve the above objective, a third aspect of the present disclosure provides an electronic device, including:

[0035] at least one memory;

[0036] at least one processor;

[0037] at least one program;

[0038] The program is stored in a memory, and the processor executes at least one program to implement the method of the present disclosure as described in the first aspect above.

[0039] To achieve the above objective, a fourth aspect of the present disclosure provides a storage medium, which is a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute:

[0040] As the method according to the first aspect above.

[0041] The medical image segmentation method and device based on high-frequency features proposed in the embodiments of the present disclosure first obtain medical raw data, then perform data enhancement processing on the medical raw data to obtain an enhanced image set, and finally input the enhanced image set into a trained multi-task network model, and perform image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result. The technical solution provided by the embodiments of the present disclosure can enhance the robustness of the image segmentation algorithm and improve the image segmentation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of a medical image segmentation method based on high-frequency features provided in an embodiment of the present disclosure.

[0043] Figure 2 yes Figure 1 Flowchart of step S140 in .

[0044] Figure 3 yes Figure 2 Flowchart of step S220 in .

[0045] Figure 4 It is a partial flow chart of a medical image segmentation method based on high-frequency features provided in another embodiment of the present disclosure.

[0046] Figure 5 yes Figure 4 Flowchart of step S430 in .

[0047] Figure 6 yes Figure 5 Flowchart of step S540 in .

[0048] Figure 7 It is a schematic diagram of the principle of the medical image segmentation method based on high-frequency features provided in an embodiment of the present disclosure.

[0049] Figure 8 It is a module block diagram of a medical image segmentation device based on high-frequency features provided in an embodiment of the present disclosure.

[0050] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided by the embodiment of the present disclosure.

[0051] Figure numerals: medical raw data acquisition module 810, data enhancement module 820, data input module 830, image segmentation module 840, processor 901, memory 902, input / output interface 903, communication interface 904, bus 905. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0053] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0055] Image segmentation is an important field of artificial intelligence application, and it has important clinical value in medical images. Medical images include images of human tissues and organs obtained by various imaging methods such as X-rays, CT, MRI, and fundus color photos. They are an indispensable basis for medical institutions to carry out disease diagnosis, surgical planning, prognosis assessment, and follow-up. Medical image processing helps to make images more intuitive and clear, and improve diagnostic efficiency. Medical image segmentation is an important part and difficulty of it, and plays an important role in medical image analysis.

[0056] The purpose of medical image segmentation is to segment the parts with special meanings in medical images and extract relevant features to provide a reliable basis for clinical diagnosis and pathological research, and to assist doctors in making more accurate diagnoses. The image segmentation process divides the image into multiple regions, which have similar properties such as grayscale, color, texture, brightness, contrast, etc. Although there are many algorithms in the field of image processing that can effectively segment images, how to continuously improve the segmentation effect and promote the robustness of the algorithm is always of great value.

[0057] Based on this, the embodiments of the present disclosure provide a medical image segmentation method and device based on high-frequency features, which first acquires raw medical data, then performs data enhancement processing on the raw medical data to obtain an enhanced image set, and finally inputs the enhanced image set into a trained multi-task network model, and performs image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result. The technical solution provided by the embodiments of the present disclosure utilizes the structural information contained in the high-frequency components of the image to assist image segmentation and improve the overall segmentation effect, thereby enhancing the robustness of the image segmentation algorithm and improving the image segmentation effect.

[0058] The embodiments of the present disclosure provide a medical image segmentation method and apparatus based on high-frequency features, which are specifically described through the following embodiments. First, the medical image segmentation method based on high-frequency features in the embodiments of the present disclosure is described.

[0059] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0060] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0061] The medical image segmentation method based on high-frequency features provided in the embodiments of the present disclosure relates to the field of artificial intelligence technology, and in particular to the field of medical image segmentation technology. The medical image segmentation method based on high-frequency features provided in the embodiments of the present disclosure can be applied to a terminal, or to a server, or can be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, or a smart watch, etc.; the server can be an independent server, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms; the software can be an application that implements a medical image segmentation method based on high-frequency features, etc., but is not limited to the above forms.

[0062] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0063] The disclosed embodiment proposes a medical image segmentation method based on high-frequency features, comprising: acquiring original medical data, the original medical data comprising original images and segmentation annotation data corresponding to the original images; performing data enhancement processing on the original medical data to obtain an enhanced image set; inputting the enhanced image set into a trained multi-task network model; performing image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result; the image segmentation result comprises high-frequency feature data and segmentation feature data.

[0064] In some embodiments, the data enhancement processing includes at least any one of the following: color space adjustment processing and image style transfer processing.

[0065] Figure 1 is an optional flowchart of a medical image segmentation method based on high-frequency features provided in an embodiment of the present disclosure. Figure 1 The method may include but is not limited to steps S110 to S150, specifically including:

[0066] S110, obtaining raw medical data;

[0067] S120, performing data enhancement processing on the original medical data to obtain an enhanced image set;

[0068] S130, inputting the enhanced image set into the trained multi-task network model;

[0069] S140, performing image segmentation processing on the enhanced image set through a multi-task network model to obtain an image segmentation result.

[0070] In step S110, the medical original data includes the original image and the segmentation annotation data corresponding to the original image, wherein the original image is denoted as S and the segmentation annotation is denoted as M. S ,It should be explained that the original medical data includes but is not limited to ,fundus images and annotation data.

[0071] In step S120, if the original image is directly used as the input of the segmentation model, the robustness of the model will be limited. Therefore, it is necessary to add data enhancement operations, such as performing color space adjustment processing and image style transfer processing on the image to generate a randomly enhanced enhanced image set. The elements in the enhanced image set all have the same structure as the image, but have different image styles. The data enhancement processing is denoted as F T , the enhanced image set is denoted as {S' k |k=1,2,...,K}, specifically, S'1 is the first image in the enhanced image set, S' K is the Kth image in the enhanced image set.

[0072] In step S130, the enhanced image set is input into a trained multi-task network model, which includes a single encoder and a dual decoder, and the dual decoders are used to train the network to extract high-frequency components and segmentation structures of the image, respectively.

[0073] In step S140, the image segmentation result includes high-frequency feature data and segmentation feature data, wherein the segmentation feature data is recorded as the final result of the image segmentation, and the high-frequency feature data is recorded as The segmentation feature data is recorded as

[0074] The medical image segmentation method based on high-frequency features proposed in the embodiment of the present disclosure first obtains the original medical data, then performs data enhancement processing on the original medical data to obtain an enhanced image set, and finally inputs the enhanced image set into a trained multi-task network model, and performs image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result. The technical solution provided by the embodiment of the present disclosure can enhance the robustness of the image segmentation algorithm and improve the image segmentation effect.

[0075] In some embodiments, the multi-task network model includes a target encoder and a target decoder, and image segmentation processing is performed on the enhanced image set through the multi-task network model to obtain an image segmentation result, including: inputting the enhanced image set into the target encoder to generate encoded target data and encoded branch data; inputting the encoded target data and the encoded branch data into the target decoder to generate an image segmentation result.

[0076] Figure 2 is a flowchart of step S140 in some embodiments, Figure 2 The illustrated step S140 includes but is not limited to steps S210 to S220:

[0077] S210, inputting the enhanced image set into a target encoder to generate encoding target data and encoding branch data;

[0078] S220, inputting the encoded target data and the encoded branch data into a target decoder to generate an image segmentation result.

[0079] In steps S210 to S220, the multi-task network model includes a target encoder and a target decoder, wherein the target encoder is denoted as E, and the direct output result after encoding by the encoder is the encoded target data, which is denoted as f E L, the encoding branch data is a jump link from the target encoder to the target decoder, forming a U-Net-like structure to improve its performance.

[0080] In some embodiments, the target decoder includes a high-frequency component decoder and an image segmentation decoder, and the encoded target data and the encoded branch data are input into the target decoder to generate an image segmentation result, including: inputting the encoded target data and the encoded branch data into the high-frequency component decoder to generate high-frequency feature data and high-frequency decoding branch data; inputting the encoded target data and the high-frequency decoding branch data into the image segmentation decoder to generate segmentation feature data.

[0081] Figure 3 is a flowchart of step S220 in some embodiments, Figure 3 The illustrated step S220 includes but is not limited to steps S310 to S320:

[0082] S310, inputting the encoding target data and the encoding branch data into a high-frequency component decoder to generate high-frequency feature data and high-frequency decoding branch data;

[0083] S320, inputting the encoded target data and the high-frequency decoding branch data into the image segmentation decoder to generate segmentation feature data.

[0084] In steps S310 to S320, the target decoder includes a high frequency component decoder and an image segmentation decoder, wherein the high frequency component decoder is denoted as D H , which is used to extract high-frequency components in the image and use them for image segmentation. The image segmentation decoder is denoted as D S , used for image segmentation; the high-frequency decoding branch data is data used to output high-frequency components to the image segmentation decoder.

[0085] In a specific embodiment, the two tasks share a target encoder E, so they can promote each other's training of the encoder E and the high-frequency component decoder D H , Image segmentation decoder D S There are jump links between them, forming a U-Net-like structure to improve its performance; image segmentation decoder D S In addition to using the output of encoder E as input, the high-frequency component decoder D H The intermediate features of are taken as input, thereby utilizing high-frequency information to promote image segmentation.

[0086] In some embodiments, the method also includes training a multi-task network model, specifically including: obtaining an initial network model; inputting an enhanced image set into the initial network model to obtain high-frequency initial feature data and segmentation initial feature data; generating a target loss function based on the high-frequency initial feature data and the segmentation initial feature data; the target loss function is a loss function of the multi-task network model; and performing model training on the initial network model according to the target loss function to obtain a multi-task network model.

[0087] like Figure 4 As shown, Figure 4 is a flowchart of a medical image segmentation method based on high-frequency features provided by other embodiments. The medical image segmentation method based on high-frequency features also includes steps S410 to S440, specifically including:

[0088] S410, obtaining an initial network model;

[0089] S420, inputting the enhanced image set into the initial network model to obtain high-frequency initial feature data and segmentation initial feature data;

[0090] S430, generating a target loss function according to the high-frequency initial feature data and the segmented initial feature data;

[0091] S440, performing model training on the initial network model according to the target loss function to obtain a multi-task network model.

[0092] In step S410, the initial network model is an initial model before training.

[0093] In step S420, a part of the enhanced image set is selected as a training set to train the initial network model, and the output of the obtained model is high-frequency initial feature data and segmentation initial feature data.

[0094] In steps S430 to S440, the target loss function is the loss function of the multi-task network model. The target loss function is calculated according to the output of the model, and then the parameters of the model are adjusted according to the loss function to obtain a trained multi-task network model.

[0095] In some embodiments, a target loss function is generated based on high-frequency initial feature data and segmentation initial feature data, including: performing high-pass filtering on the original medical data to obtain a filtered image; generating a first loss function based on the high-frequency initial feature data and the filtered image; the first loss function is a loss function of a target encoder and a high-frequency component decoder; generating a second loss function based on the segmentation initial feature data and the segmentation annotation data; the second loss function is a loss function of a target encoder and an image segmentation decoder; and obtaining a target loss function based on the first loss function and the second loss function.

[0096] Figure 5 is a flowchart of step S430 in some embodiments, Figure 5 The illustrated step S430 includes but is not limited to steps S510 to S530:

[0097] S510, performing high-pass filtering on the raw medical data to obtain a filtered image;

[0098] S520, generating a first loss function according to the high-frequency initial feature data and the filtered image;

[0099] S530, generating a second loss function according to the segmentation initial feature data and the segmentation annotation data;

[0100] S540, obtaining a target loss function according to the first loss function and the second loss function.

[0101] In step S510, according to Retinex theory, common image noise and style mostly belong to low-frequency components, while image structure is stored in high-frequency components. Therefore, the high-frequency components of the image can be used to assist image segmentation. A high-pass filter is used to obtain the high-frequency components from the image for image segmentation. The filtered image is a filtered image, where the high-pass filtering process is denoted as H(·), the filtered image is denoted as H(S), and the filtered image is used to generate the first loss function.

[0102] In step S520, considering the original image S and the enhanced image set S' k have the same structural information, so the enhanced image set S' extracted by the model k The high-frequency components of should be consistent with the original image S, so the first loss function can be calculated as: The first loss function is the target encoder E and the high-frequency component decoder D H The loss function is, where L H represents the first loss function, K is the number of images in the enhanced image set, H(S) is the filtered image, Represents the high-frequency feature data obtained during the model training process.

[0103] In step S530, when the model has a certain robustness to image style and noise, the segmentation results of the enhanced image set should also be consistent with the segmentation annotation M S If the target encoder E and the image segmentation decoder D are consistent, S The second loss function is: Among them, L S represents the second loss function, K is the number of images in the enhanced image set, M S For segmentation annotation, Represents the segmentation feature data obtained during the model training process.

[0104] In some embodiments, obtaining a target loss function according to the first loss function and the second loss function includes: obtaining a loss function generation coefficient; and weighting the first loss function and the second loss function according to the loss function generation coefficient to obtain the target loss function.

[0105] Figure 6 is a flowchart of step S540 in some embodiments, Figure 6 The illustrated step S540 includes but is not limited to steps S610 to S620:

[0106] S610, obtaining loss function generation coefficients;

[0107] S620, weighting the first loss function and the second loss function according to the loss function generation coefficient to obtain a target loss function.

[0108] In steps S610 to S620, the loss function generates coefficients λ S , the weighted processing process is: L total =L H +λ S L S , where L H represents the first loss function, L S represents the second loss function, L total represents the target loss function.

[0109] In a specific embodiment, by training the segmentation network through the above-mentioned target loss function, high-frequency information can be used to guide image segmentation and promote the robustness of the model to image style and noise.

[0110] Figure 7 is a schematic diagram of the principle of the medical image segmentation method based on high-frequency features provided by the embodiment of the present disclosure, and the following reference is made to Figure 7 , a medical image segmentation method based on high-frequency features is described in detail with a specific embodiment. It is worth noting that the following description is only an exemplary description and not a specific limitation of the application.

[0111] After the model is trained, data enhancement processing F is performed first T , and obtain the enhanced image set {S' k |k=1,2,...,K}, then the enhanced image set S' k Input into the trained multi-task network model, first pass through the target encoder E, and output the encoded target data and It is a branch of the target encoder E, used to transmit the encoded branch data to the high-frequency component decoder D H , then the high frequency component decoder D H Generate high-frequency feature data based on the encoded target data and the encoded branch data and high frequency decoded branch data, and then the high frequency decoded branch data is further decoded by the high frequency component decoder D H The branch is transmitted to the image segmentation decoder D S , image segmentation decoder D S Generate segmentation feature data based on the encoded target data and high-frequency decoding branch data This enables the application of high-frequency information in image segmentation.

[0112] Among them, the target encoder E includes E 1 、E 2 、E l 、E L-1 and E L ; High frequency component decoder D H include Image Segmentation Decoder D S include

[0113] It should be noted that Figure 7 In and The meaning of is different from that in the calculation process of the loss function. Figure 7 In and It is the image segmentation result of the actual segmentation data, and in the calculation process of the loss function and It represents the intermediate data required in the training process of the training set. The two represent different data at different stages and cannot be understood as the same data or ambiguous.

[0114] The disclosed embodiment proposes a medical image segmentation device based on high-frequency features, including: a medical raw data acquisition module, used to acquire medical raw data, the medical raw data including original images and segmentation annotation data corresponding to the original images; a data enhancement module, used to perform data enhancement processing on the medical raw data to obtain an enhanced image set; a data input module, used to input the enhanced image set into a trained multi-task network model; an image segmentation module, used to perform image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result; the image segmentation result includes high-frequency feature data and segmentation feature data.

[0115] See also Figure 8 , Figure 8 The diagram illustrates an embodiment of a medical image segmentation device based on high-frequency features, which includes: a medical original data acquisition module 810, a data enhancement module 820, a data input module 830, and an image segmentation module 840, wherein the medical original data acquisition module 810 is connected to the data enhancement module 820, the data enhancement module 820 is connected to the data input module 830, and the data input module 830 is connected to the image segmentation module 840.

[0116] The specific implementation of the medical image segmentation device based on high-frequency features in this embodiment is basically consistent with the specific implementation of the medical image segmentation method based on high-frequency features described above, and belongs to the same inventive concept, and will not be repeated here.

[0117] The present disclosure also provides an electronic device, including:

[0118] at least one memory;

[0119] at least one processor;

[0120] at least one program;

[0121] The program is stored in the memory, and the processor executes the at least one program to implement the above-mentioned medical image segmentation method based on high-frequency features in the present disclosure. The electronic device can be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), a car computer, etc.

[0122] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0123] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present disclosure;

[0124] The memory 902 can be implemented in the form of ROM (Read Only Memory), static storage device, dynamic storage device or RAM (Random Access Memory). The memory 902 can store operating systems and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 902, and the processor 901 calls and executes the medical image segmentation method based on high-frequency features in the embodiments of the present disclosure;

[0125] Input / output interface 903, used to implement information input and output;

[0126] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.); and

[0127] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0128] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0129] The embodiment of the present disclosure also provides a storage medium, which is a computer-readable storage medium. The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned medical image segmentation method based on high-frequency features.

[0130] The medical image segmentation method and device based on high-frequency features proposed in the embodiments of the present disclosure first acquire the original medical data, then perform data enhancement processing on the original medical data to obtain an enhanced image set, and finally input the enhanced image set into a trained multi-task network model, and perform image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result. The technical solution provided in the embodiments of the present disclosure enhances the robustness of the image segmentation algorithm and improves the image segmentation effect.

[0131] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0132] The embodiments described in the embodiments of the present disclosure are intended to more clearly illustrate the technical solutions of the embodiments of the present disclosure and do not constitute a limitation on the technical solutions provided by the embodiments of the present disclosure. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present disclosure are also applicable to similar technical problems.

[0133] It can be understood by those skilled in the art that Figure 1-6 The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present disclosure, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.

[0134] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0135] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0136] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0137] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0138] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0139] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0140] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0141] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0142] The preferred embodiments of the present disclosure are described above with reference to the accompanying drawings, but the scope of the rights of the present disclosure is not limited thereto. Any modification, equivalent substitution and improvement made by those skilled in the art without departing from the scope and essence of the present disclosure should be within the scope of the rights of the present disclosure.

Claims

1. A medical image segmentation method based on high-frequency features, characterized in that: include: Acquire original medical data, wherein the original medical data includes an original image and segmentation and annotation data corresponding to the original image; Performing data enhancement processing on the raw medical data to obtain an enhanced image set; Inputting the enhanced image set into a trained multi-task network model; wherein the multi-task network model includes a target encoder and a target decoder, and the target decoder includes a high-frequency component decoder and an image segmentation decoder; Performing image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result; the image segmentation result includes high-frequency feature data and segmentation feature data, and the segmentation feature data is recorded as the final result of the image segmentation; The performing image segmentation processing on the enhanced image set by the multi-task network model to obtain an image segmentation result includes: Inputting the enhanced image set into the target encoder to generate encoding target data and encoding branch data; the direct output result after encoding by the encoder is the encoding target data, and the encoding branch data is a jump link from the target encoder to the target decoder; Inputting the encoded target data and the encoded branch data into the target decoder to generate the image segmentation result; The step of inputting the encoded target data and the encoded branch data into the target decoder to generate the image segmentation result comprises: Inputting the encoding target data and the encoding branch data into the high-frequency component decoder to generate high-frequency feature data and high-frequency decoding branch data; wherein the high-frequency decoding branch data is data for outputting high-frequency components to the image segmentation decoder; The encoding target data and the high-frequency decoding branch data are input into the image segmentation decoder to generate the segmentation feature data.

2. The method according to claim 1, characterized in that The method further includes training the multi-task network model, specifically including: Obtain an initial network model; Inputting the enhanced image set into the initial network model to obtain high-frequency initial feature data and segmentation initial feature data; Generate a target loss function according to the high-frequency initial feature data and the segmented initial feature data; the target loss function is a loss function of the multi-task network model; The initial network model is trained according to the target loss function to obtain the multi-task network model.

3. The method according to claim 2, characterized in that The generating a target loss function according to the high-frequency initial feature data and the segmented initial feature data comprises: Performing high-pass filtering on the raw medical data to obtain a filtered image; Generate a first loss function according to the high-frequency initial feature data and the filtered image; the first loss function is a loss function of a target encoder and a high-frequency component decoder; Generate a second loss function according to the segmentation initial feature data and the segmentation annotation data; the second loss function is a loss function of the target encoder and the image segmentation decoder; The target loss function is obtained according to the first loss function and the second loss function.

4. The method according to claim 3, characterized in that The obtaining the target loss function according to the first loss function and the second loss function includes: Get the loss function generation coefficient; The first loss function and the second loss function are weighted according to the loss function generation coefficient to obtain the target loss function.

5. The method according to any one of claims 1 to 4, characterized in that: The data enhancement processing includes at least any one of the following: color space adjustment processing and image style transfer processing.

6. A medical image segmentation device based on high-frequency features, characterized in that: include: A medical raw data acquisition module, used to acquire medical raw data, wherein the medical raw data includes an original image and segmentation and annotation data corresponding to the original image; A data enhancement module, used for performing data enhancement processing on the raw medical data to obtain an enhanced image set; A data input module, used to input the enhanced image set into a trained multi-task network model; wherein the multi-task network model includes a target encoder and a target decoder, and the target decoder includes a high-frequency component decoder and an image segmentation decoder; An image segmentation module, used to perform image segmentation processing on the enhanced image set through the multi-task network model to obtain an image segmentation result; the image segmentation result includes high-frequency feature data and segmentation feature data, and the segmentation feature data is recorded as the final result of the image segmentation; The performing image segmentation processing on the enhanced image set by the multi-task network model to obtain an image segmentation result includes: Inputting the enhanced image set into the target encoder to generate encoding target data and encoding branch data; the direct output result after encoding by the encoder is the encoding target data, and the encoding branch data is a jump link from the target encoder to the target decoder; Inputting the encoded target data and the encoded branch data into the target decoder to generate the image segmentation result; The step of inputting the encoded target data and the encoded branch data into the target decoder to generate the image segmentation result comprises: Inputting the encoding target data and the encoding branch data into the high-frequency component decoder to generate high-frequency feature data and high-frequency decoding branch data; wherein the high-frequency decoding branch data is data for outputting high-frequency components to the image segmentation decoder; The encoding target data and the high-frequency decoding branch data are input into the image segmentation decoder to generate the segmentation feature data.

7. An electronic device, characterized in that: include: at least one memory; at least one processor; at least one program; The program is stored in the memory, and the processor executes at least one of the programs to implement: A method as claimed in any one of claims 1 to 5.

8. A storage medium, the storage medium being a computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute: A method as claimed in any one of claims 1 to 5.

Images