Medical image universal segmentation method and device based on peripheral context information enhancement
By integrating the surrounding context information of medical images into the deep learning network and enhancing feature expression, the problem of inaccurate segmentation in existing methods is solved, and high-precision medical image segmentation is achieved, which is suitable for CT and MRI images.
Patent Information
- Application Number
- CN202510698582.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-28
- Publication Date
- 2025-10-17
AI Technical Summary
Existing deep learning-based medical image segmentation methods lack effective utilization of surrounding contextual information of the image, resulting in inaccurate segmentation and loss of details, making it difficult to meet the high-precision requirements of clinical applications.
By extracting different scale features of the initial medical image and downsampling it, the surrounding context information of the image is obtained and fused with image features of different scales. After enhancing the image features, the deep learning network is trained until the preset loss function converges to obtain a medical image segmentation model.
It improves the accuracy and versatility of medical image segmentation, meets the medical field's demand for high-precision image segmentation, is suitable for the segmentation processing of images such as CT and MRI, and improves the accuracy and reliability of segmentation.
Smart Images
Figure CN120807901A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of medical image processing, in particular to a medical image general segmentation method and device based on peripheral context information enhancement. BACKGROUND
[0002] Medical image segmentation is a key technology in medical image processing, aiming to separate the region of interest (such as lesion tissue, organ, etc.) from the background in medical images. Accurate medical image segmentation is of great significance for doctors to accurately diagnose diseases, develop personalized treatment plans, and evaluate treatment effectiveness.
[0003] Currently, medical image segmentation techniques mainly include threshold-based segmentation methods, region-based segmentation methods, edge-based segmentation methods, and deep learning-based segmentation methods. Threshold-based segmentation methods are simple and fast, but have poor segmentation results for medical images with uneven gray scale distribution; region-based segmentation methods are sensitive to image noise; edge-based segmentation methods are difficult to handle medical images with fuzzy edges; deep learning-based segmentation methods have achieved good results in some cases, but when dealing with complex medical images, due to the lack of effective use of peripheral context information, they are prone to inaccurate segmentation, loss of details, and other problems, making it difficult to meet the high-precision requirements of clinical applications.
[0004] Therefore, there is an urgent need for a method that can effectively utilize peripheral context information to improve medical image segmentation accuracy and versatility. SUMMARY
[0005] The present application provides a medical image general segmentation method and device based on peripheral context information enhancement to solve the problem of lack of effective use of peripheral context information in existing deep learning-based segmentation methods, which is prone to inaccurate segmentation, loss of details, and other problems, making it difficult to meet the high-precision requirements of clinical applications.
[0006] The first aspect of the present application provides a medical image general segmentation method based on peripheral context information enhancement, comprising the following steps: extracting image features of different scales in an initial medical image, and downsampling the image features of different scales to obtain peripheral context information; fusing the peripheral context information with the image features of different scales to obtain fused enhanced image features; training a pre-constructed deep learning network architecture using the fused enhanced image features until the preset loss function converges, obtaining a medical image segmentation model; inputting the medical image to be segmented into the medical image segmentation model to obtain the corresponding segmentation result.
[0007] Optionally, the different scales of image features in the initial medical image are extracted, and the different scales of image features are down-sampled to obtain the image peripheral context information, including:
[0008] An initial medical image is acquired, and the initial medical image is pre-processed to obtain a pre-processed medical image; a pre-constructed context information extraction model based on a convolutional neural network is used to extract features of the pre-processed medical image to obtain the different scales of image features, and the different scales of image features are down-sampled to obtain the image peripheral context information.
[0009] Optionally, the context information extraction model based on the convolutional neural network includes a plurality of convolutional layers and a pooling layer, wherein the plurality of convolutional layers are used to extract features of the pre-processed medical image to obtain the different scales of image features, and the pooling layer is used to down-sample the different scales of image features to obtain the image peripheral context information.
[0010] Optionally, the image peripheral context information and the different scales of image features are fused to obtain the fused enhanced image features, including:
[0011] The image peripheral context information and the different scales of image features are element-wise added or spliced to make the different scales of image features fuse the image peripheral context information to obtain the fused enhanced image features.
[0012] The second aspect embodiment of the present application provides a medical image general segmentation device based on peripheral context information enhancement, including:
[0013] An extraction and down-sampling module is configured to extract different scales of image features in an initial medical image, and down-sample the different scales of image features to obtain image peripheral context information; a fusion module is configured to fuse the image peripheral context information and the different scales of image features to obtain fused enhanced image features; a training module is configured to train a pre-constructed deep learning network architecture using the fused enhanced image features until a preset loss function converges to obtain a medical image segmentation model; and a segmentation module is configured to input a medical image to be segmented into the medical image segmentation model to obtain a corresponding segmentation result.
[0014] Optionally, the extraction and down-sampling module includes:
[0015] The preprocessing unit is configured to acquire an initial medical image and preprocess the initial medical image to obtain a preprocessed medical image; and the feature extraction and down-sampling unit is configured to perform feature extraction on the preprocessed medical image by using a pre-constructed context information extraction model based on a convolutional neural network to obtain image features of different scales, and perform down-sampling on the image features of different scales to obtain the image peripheral context information.
[0016] Optionally, the context information extraction model based on the convolutional neural network comprises a plurality of convolutional layers and a pooling layer, wherein the plurality of convolutional layers are configured to perform feature extraction on the preprocessed medical image to obtain the image features of different scales, and the pooling layer is configured to perform down-sampling on the image features of different scales to obtain the image peripheral context information.
[0017] Optionally, the fusion module comprises:
[0018] The image peripheral context information is added to or spliced with the image features of different scales element by element, so that the image features of different scales fuse the image peripheral context information to obtain the fused enhanced image features.
[0019] A third aspect of the present application provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the medical image general segmentation method based on peripheral context information enhancement as described in the above embodiments.
[0020] A fourth aspect of the present application provides a computer readable storage medium, which stores a computer program executable by a processor to implement the medical image general segmentation method based on peripheral context information enhancement as described above.
[0021] The medical image general segmentation method and device based on peripheral context information enhancement provided by the embodiments of the present application can fully utilize the context information of the peripheral region of the target region in the medical image, enhance the understanding and expression of the model for the features of the target region, and thus improve the accuracy and generality of medical image segmentation, so as to meet the demand of the medical field for high-precision image segmentation; the method and device can be applied to the segmentation processing of various medical images such as CT and MRI images, and can effectively improve the accuracy and reliability of medical image segmentation, and provide strong support for medical diagnosis, disease analysis and treatment plan making.
[0022] Additional aspects and advantages of the present application will be made apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS
[0023] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the references to the figures, in which:
[0024] Figure 1 A flowchart of a medical image general segmentation method based on peripheral context information enhancement provided by an embodiment of the present application;
[0025] Figure 2 A block schematic diagram of a medical image general segmentation device based on peripheral context information enhancement provided by an embodiment of the present application;
[0026] Figure 3 A structural schematic diagram of an electronic device apparatus provided by an embodiment of the present application. DETAILED DESCRIPTION
[0027] Embodiments of the present application are described in detail below with reference to the attached drawings, which show by way of example, embodiments in which the same or similar elements have the same or similar reference numbers and in which:
[0028] A medical image general segmentation method and device based on peripheral context information enhancement of an embodiment of the present application are described below with reference to the accompanying drawings.
[0029] Figure 1 A flowchart of a medical image general segmentation method based on peripheral context information enhancement provided by an embodiment of the present application.
[0030] As shown in Figure 1 the medical image general segmentation method based on peripheral context information enhancement includes the following steps:
[0031] In step S101, image features of different scales in an initial medical image are extracted, and the image features of different scales are down-sampled to obtain image peripheral context information.
[0032] In some embodiments, extracting image features of different scales in an initial medical image and down-sampling the image features of different scales to obtain image peripheral context information includes:
[0033] An initial medical image is obtained, and the initial medical image is preprocessed to obtain a preprocessed medical image;
[0034] The pre-constructed context information extraction model based on a convolutional neural network is used for feature extraction on the preprocessed medical image to obtain image features of different scales, and the image features of different scales are down-sampled to obtain the image peripheral context information.
[0035] In actual execution, the initial medical image is acquired and normalized to normalize the pixel values of the initial medical image to a specific range, such as [0, 1], to eliminate the gray difference between different images; at the same time, the initial medical image is also denoised, and a suitable filtering algorithm such as median filtering, Gaussian filtering, etc. is used to remove noise interference in the image and improve image quality.
[0036] Further, the pre-constructed context information extraction model based on a convolutional neural network CNN is used, which includes multiple convolutional layers and pooling layers. The preprocessed medical image is feature-extracted by the convolutional layer to obtain image features of different scales; the image features of different scales are down-sampled by the pooling layer to reduce the size and calculation of the feature map, while retaining the main feature information of the image, to obtain the image peripheral context information. In the convolution and pooling process, the size, number of convolution kernels and the step length of the pooling layer are adjusted to effectively extract the image peripheral context information.
[0037] In step S102, the image peripheral context information is fused with the image features of different scales to obtain the fused enhanced image features.
[0038] In some embodiments, the image peripheral context information is fused with the image features of different scales to obtain the fused enhanced image features, including:
[0039] The image peripheral context information is element-wise added or spliced with the image features of different scales to make the image features of different scales fuse the image peripheral context information to obtain the fused enhanced image features.
[0040] In actual execution, the extracted image peripheral context information is fused with the image features of different scales of the initial medical image. Specifically, the image features of different scales of the initial medical image output by the context information extraction model based on a convolutional neural network are element-wise added or spliced with the image peripheral context information after a certain convolution operation, so that the initial medical image can fully fuse the peripheral context information to obtain the fused enhanced image features, which enhances the feature expression ability of the target region.
[0041] In step S103, the pre-constructed deep learning network architecture is trained using the fused enhanced image features until the preset loss function converges to obtain the medical image segmentation model.
[0042] In actual execution, the fused enhanced image features are taken as the input of a pre-constructed deep learning network architecture (such as U-Net, SegNet, etc.) to train the model and construct a medical image segmentation model. In the model training process, a suitable loss function such as a cross-entropy loss function, a Dice loss function, etc. is adopted, combined with a large number of fused enhanced image features for training, and the parameters of the model are constantly adjusted through a back propagation algorithm to minimize the error between the segmentation result of the model and the labeled result, thereby improving the segmentation accuracy and generalization ability of the medical image segmentation model.
[0043] In step S104, the medical image to be segmented is input into the medical image segmentation model to obtain the corresponding segmentation result.
[0044] In actual execution, the medical image to be segmented is preprocessed, and then the preprocessed medical image to be segmented is input into the trained medical image segmentation model to classify each pixel in the preprocessed segmented medical image and divide it into different regions to obtain the corresponding segmentation result, thereby realizing general segmentation of medical images.
[0045] In summary, the medical image general segmentation method based on peripheral context information enhancement according to the embodiments of the present application fully utilizes the context information of the target region periphery in the medical image, enhances the understanding and expression of the model for the target region features, thereby improving the accuracy and generality of medical image segmentation to meet the demand of the medical field for high-precision image segmentation; it can be applied to segmentation processing of various medical images such as CT, MRI, etc., and can effectively improve the accuracy and reliability of medical image segmentation, providing strong support for medical diagnosis, disease analysis and treatment plan formulation, etc.
[0046] Next, a medical image general segmentation device based on peripheral context information enhancement according to embodiments of the present application is described with reference to the accompanying drawings.
[0047] Figure 2 A block diagram of a medical image general segmentation device based on peripheral context information enhancement according to embodiments of the present application is shown.
[0048] As shown in Figure 2 The medical image general segmentation device based on peripheral context information enhancement 20 includes an extraction and downsampling module 201, a fusion module 202, a training module 203, and a segmentation module 204.
[0049] The extraction and down-sampling module 201 is configured to extract image features of different scales in the initial medical image, and down-sample the image features of different scales to obtain image peripheral context information. The fusion module 202 is configured to fuse the image peripheral context information with the image features of different scales to obtain fused enhanced image features. The training module 203 is configured to train a pre-constructed deep learning network architecture by using the fused enhanced image features until a preset loss function converges, to obtain a medical image segmentation model. The segmentation module 204 is configured to input a medical image to be segmented into the medical image segmentation model to obtain a corresponding segmentation result.
[0050] In some embodiments, the extraction and down-sampling module 201 comprises:
[0051] a preprocessing unit configured to obtain an initial medical image, and pre-process the initial medical image to obtain a pre-processed medical image;
[0052] an extraction and down-sampling unit configured to extract features of the pre-processed medical image by using a pre-constructed context information extraction model based on a convolutional neural network, to obtain image features of different scales, and down-sample the image features of different scales to obtain image peripheral context information.
[0053] In some embodiments, the context information extraction model based on the convolutional neural network comprises a plurality of convolutional layers and a pooling layer, wherein the plurality of convolutional layers are configured to extract features of the pre-processed medical image to obtain image features of different scales, and the pooling layer is configured to down-sample the image features of different scales to obtain the image peripheral context information.
[0054] In some embodiments, the fusion module 202 comprises:
[0055] The image peripheral context information and the image features of different scales are added or spliced element by element, so that the image features of different scales are fused with the image peripheral context information to obtain the fused enhanced image features.
[0056] It should be noted that the foregoing explanation and description of the embodiment of the medical image universal segmentation method based on peripheral context information enhancement also applies to the embodiment of the medical image universal segmentation device based on peripheral context information enhancement, which will not be described here.
[0057] The medical image general segmentation device based on peripheral context information enhancement provided by the embodiment of the present application can fully utilize the context information of the target region periphery in the medical image, enhance the understanding and expression of the model to the target region features, and thus improve the accuracy and generality of the medical image segmentation, so as to meet the demand of the medical field for high-precision image segmentation. The device can be applicable to the segmentation processing of various medical images such as CT, MRI and the like, can effectively improve the accuracy and reliability of the medical image segmentation, and provides strong support for medical diagnosis, disease analysis and treatment scheme formulation and the like.
[0058] Figure 3 A structural schematic diagram of an electronic device provided by an embodiment of the present application.
[0059] The electronic device can include a memory 301, a processor 302, and a computer program stored on the memory 301 and executable on the processor 302.
[0060] The processor 302 implements the medical image general segmentation method based on peripheral context information enhancement provided in the above embodiments when executing the program.
[0061] Further, the electronic device further includes:
[0062] A communication interface 303 for communication between the memory 301 and the processor 302.
[0063] The memory 301 is used to store the computer program executable on the processor 302.
[0064] The memory 301 can include a high-speed RAM memory, and can also include a non-volatile memory such as at least one disk memory.
[0065] If the memory 301, the processor 302 and the communication interface 303 are independently implemented, the communication interface 303, the memory 301 and the processor 302 can be connected to each other through a bus and complete the communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 3 Only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
[0066] Optionally, in a specific implementation, if the memory 301, the processor 302 and the communication interface 303 are integrated on a chip, the memory 301, the processor 302 and the communication interface 303 can complete mutual communication through an internal interface.
[0067] The processor 302 can be a central processing unit (CPU) or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present application.
[0068] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the medical image general segmentation method based on peripheral context information enhancement as described above.
[0069] In the description of the present application, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present application, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, the skilled in the art can combine and combine the different embodiments or examples described in the present application and the features of the different embodiments or examples without contradiction.
[0070] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one feature. In the description of the present application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise specifically limited.
[0071] Any process or method descriptions in flow charts or described elsewhere herein can be understood as representing code modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions (or steps) in the process, and that the various embodiments of the application can include different implementations of code modules, segments, or portions of code that implement the specified functions (or steps) of the algorithms in different ways as will occur to those skilled in the art. The acts of a method can be performed by one or more programmable processors executing a computer program to perform functions by operating on input data and generating output.
[0072] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of instructions to implement logic functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. In the context of this specification, a "computer-readable medium" can be any means that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a machine-readable storage device (e.g., magnetic, optical or other) a machine-readable storage diskette (e.g., floppy, flexible or other), a machine-readable storage card (e.g., ROM, EEPROM, flash memory or other), a machine- readable storage tape (e.g., magnetic, optical or other), a machine-readable storage medium (e.g., a portable electronic device, a computer diskette, a computer memory, a programmable logic device, an application-specific integrated circuit, a programmable logic controller, a digital signal processor, a microprocessor, a microprocessor array or other), or a machine- readable interface device (e.g., a wired or wireless interface device). The computer-readable medium can also be paper or other suitable material upon which the program is printed, as the program can be electronically captured, via optical scanning of the paper or other suitable medium, then compiled, interpreted or otherwise processed in a suitable manner into a useable digital data. Further, it should be appreciated that a "computer-readable medium" can be any combination of these or other computer- readable media.
[0073] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the embodiments described above, the steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. If implemented in hardware, and in another embodiment, any of the following technologies, or combinations thereof, can be used: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application-specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), and / or the like.
[0074] Those of skill in the art could readily implement the above described example methods with all or a subset of the recited steps or elements. Those with skill in the art would further understand that they can be stored in a computer readable storage medium, a program
[0075] In addition, each function unit in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software function module. When the integrated module is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium.
[0076] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.
Claims
1. A general medical image segmentation method based on surrounding context information enhancement, characterized in that: The following steps are involved: Extracting image features of different scales from the initial medical image and downsampling the image features of different scales to obtain image surrounding context information; fusing the image surrounding context information with the image features of different scales to obtain fused enhanced image features; The pre-built deep learning network architecture is trained using the fused enhanced image features until a preset loss function converges to obtain a medical image segmentation model; The medical image to be segmented is input into the medical image segmentation model to obtain a corresponding segmentation result.
2. The general medical image segmentation method based on surrounding context information enhancement according to claim 1, characterized in that: The extracting image features of different scales from the initial medical image and downsampling the image features of different scales to obtain image surrounding context information includes: Acquiring an initial medical image, and preprocessing the initial medical image to obtain a preprocessed medical image; The pre-processed medical image is subjected to feature extraction using a pre-built context information extraction model based on a convolutional neural network to obtain image features of different scales, and the image features of different scales are down-sampled to obtain context information surrounding the image.
3. The general medical image segmentation method based on surrounding context information enhancement according to claim 2, characterized in that: The context information extraction model based on convolutional neural network includes multiple convolutional layers and pooling layers, wherein the multiple convolutional layers are used to extract features of the preprocessed medical image to obtain image features of different scales, and the pooling layers are used to downsample the image features of different scales to obtain the surrounding context information of the image.
4. The general medical image segmentation method based on surrounding context information enhancement according to claim 1, characterized in that: The fusing of the image surrounding context information with the image features of different scales to obtain fused enhanced image features includes: The image surrounding context information is added or concatenated element by element with the image features of different scales, so that the image features of different scales are fused with the image surrounding context information to obtain the fused enhanced image features.
5. A general medical image segmentation device based on surrounding context information enhancement, characterized in that: include: An extraction and downsampling module is used to extract image features of different scales from the initial medical image and downsample the image features of different scales to obtain image surrounding context information; A fusion module, configured to fuse the image surrounding context information with the image features of different scales to obtain fused enhanced image features; A training module is used to train a pre-built deep learning network architecture using the fused enhanced image features until a preset loss function converges to obtain a medical image segmentation model; The segmentation module is used to input the medical image to be segmented into the medical image segmentation model to obtain the corresponding segmentation result.
6. The general medical image segmentation device based on surrounding context information enhancement according to claim 5, characterized in that: The extraction and downsampling module includes: a preprocessing unit, configured to acquire an initial medical image and preprocess the initial medical image to obtain a preprocessed medical image; An extraction and downsampling unit is used to perform feature extraction on the preprocessed medical image using a pre-built context information extraction model based on a convolutional neural network to obtain image features of different scales, and downsample the image features of different scales to obtain context information around the image.
7. The general medical image segmentation device based on surrounding context information enhancement according to claim 6, characterized in that: The context information extraction model based on convolutional neural network includes multiple convolutional layers and pooling layers, wherein the multiple convolutional layers are used to extract features of the preprocessed medical image to obtain image features of different scales, and the pooling layers are used to downsample the image features of different scales to obtain the surrounding context information of the image.
8. The general medical image segmentation device based on surrounding context information enhancement according to claim 5, characterized in that: The fusion module includes: The image surrounding context information is added or concatenated element by element with the image features of different scales, so that the image features of different scales are fused with the image surrounding context information to obtain the fused enhanced image features.
9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the general medical image segmentation method based on surrounding context information enhancement as described in any one of claims 1 to 4.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the general medical image segmentation method based on surrounding context information enhancement as described in any one of claims 1 to 4.