Image segmentation method and device based on AI

Abstract

The invention relates to the technical field of image processing, and relates to an AI-based image segmentation method and device, and the method comprises the steps: obtaining a brain CT image sequence, a brain MRI image sequence, and a brain MRA image sequence; the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence are subjected to feature extraction, first feature information, second feature information and third feature information are obtained, the first feature information comprises three-dimensional structure feature information of brain blood vessels, and the second feature information comprises brain tissue distribution feature information; fusing the first feature information, the second feature information and the third feature information to obtain fused feature map information; enhancing the fused feature map information to obtain enhanced feature map information; and according to the enhanced feature map information, the microvessels in the brain are segmented, segmented image information is obtained, the features of multiple modes are integrated, and the segmentation precision of the brain microvessels is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an AI-based image segmentation method and device. Background Art

[0002] Brain tumor is a common disease of the nervous system, which has a serious impact on the function of the human nervous system. In the diagnosis of brain tumor, the microvessels in the image are segmented through vascular imaging technology. The malignancy of the tumor can be identified based on indicators such as microvessel density and morphology, thereby further judging the malignancy of the patient's tumor. However, the signal intensity of the newly formed microvessels in the brain is relatively weak and can be easily masked by the signals of the surrounding tissues, making it difficult to accurately identify and segment the microvessels. Summary of the invention

[0003] The purpose of the present invention is to provide an AI-based image segmentation method and device to improve the above-mentioned problems.

[0004] In order to achieve the above objectives, the present application provides the following technical solutions:

[0005] On the one hand, an embodiment of the present application provides an AI-based image segmentation method, the method comprising:

[0006] Acquire brain CT image sequences, brain MRI image sequences, and brain MRA image sequences;

[0007] performing feature extraction on the brain CT image sequence, the brain MRI image sequence, and the brain MRA image sequence respectively to obtain first feature information, second feature information, and third feature information, wherein the first feature information includes three-dimensional structural feature information of brain blood vessels, the second feature information includes brain tissue distribution feature information, and the third feature information includes morphological feature information of brain blood vessels;

[0008] Fusing the first feature information, the second feature information, and the third feature information to obtain fused feature map information;

[0009] Enhance the fused feature map information to obtain enhanced feature map information;

[0010] The microvessels in the brain are segmented according to the enhanced feature map information to obtain segmented image information.

[0011] In a second aspect, an embodiment of the present application provides an AI-based image segmentation device, the device comprising:

[0012] An acquisition module, used for acquiring a brain CT image sequence, a brain MRI image sequence, and a brain MRA image sequence;

[0013] a feature extraction module, used to perform feature extraction on the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence respectively to obtain first feature information, second feature information and third feature information, wherein the first feature information includes three-dimensional structural feature information of brain blood vessels, the second feature information includes brain tissue distribution feature information, and the third feature information includes morphological feature information of brain blood vessels;

[0014] A first processing module, used for fusing the first feature information, the second feature information and the third feature information to obtain fused feature map information;

[0015] A second processing module, used for enhancing the fused feature map information to obtain enhanced feature map information;

[0016] The third processing module is used to segment the microvessels in the brain according to the enhanced feature map information to obtain segmented image information.

[0017] In a third aspect, an embodiment of the present application provides an AI-based image segmentation device, the device comprising a memory and a processor. The memory is used to store a computer program; the processor is used to implement the steps of the above-mentioned AI-based image segmentation method when executing the computer program.

[0018] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned AI-based image segmentation method are implemented.

[0019] The beneficial effects of the present invention are:

[0020] The present invention acquires a brain CT image sequence, a brain MRI image sequence and a brain MRA image sequence, determines first feature information, second feature information and third feature information according to the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence respectively, and then fuses the first feature information, the second feature information and the third feature information to obtain fused feature map information. By fusing the fused feature map information and the features of multiple modalities, the masking of microvascular signals by the tissue surrounding the microvascular can be effectively determined, thereby improving the segmentation accuracy of the microvascular.

[0021] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 Schematic diagram of the process of the AI-based image segmentation method described in an embodiment of the present invention.

[0024] Figure 2 Schematic diagram of the structure of the AI-based image segmentation device described in an embodiment of the present invention.

[0025] Figure 3 Schematic diagram of the structure of the AI-based image segmentation device described in an embodiment of the present invention.

[0026] Labels in the figure: 800, AI-based image segmentation device; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component; 901, acquisition module; 902, feature extraction module; 903, first processing module; 904, second processing module; 905, third processing module. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0028] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0029] Embodiment 1:

[0030] This embodiment provides an AI-based image segmentation method. It can be understood that in this embodiment, a scene can be laid out, for example: detecting the patient's brain microvessels to determine the malignancy of the patient's tumor.

[0031] See also Figure 1 , the figure shows that the method includes step S1, step S2, step S3, step S4 and step S5.

[0032] Step S1, acquiring a brain CT image sequence, a brain MRI image sequence, and a brain MRA image sequence;

[0033] In this step, the tiny size of the newly formed microvessels is a key difficulty in segmentation. During the imaging process, due to resolution limitations, even advanced imaging technology may find it difficult to clearly distinguish the boundaries of microvessels. Brain CT image sequences can provide high-resolution images of microvascular anatomical structures, and brain MRI image sequences can provide brain tissue microstructure to locate tissue features around microvessels. MRA image sequences can directly display the morphology of low-resolution microvessels, and by integrating images of different modalities, more comprehensive information can be obtained, thereby providing rich clues for more accurate microvascular segmentation.

[0034] It should be noted that the sampling interval of the image sequence also has an impact on the newly formed microvessels in the brain. When the sampling time is short, such as scanning every few minutes or more than ten minutes, it is possible to more accurately capture dynamic information such as the growth, branching and diameter changes of the microvessels. This is very helpful for accurately segmenting microvessels, because the segmentation algorithm can use these continuous and subtle changes to more accurately track blood vessel boundaries. However, a longer sampling time may lead to the accumulation of artifacts and noise, which also interferes with the segmentation. Therefore, when acquiring brain CT image sequences, brain MRI image sequences, and brain MRA image sequences, the sampling time interval needs to be determined. The specific process is: calculating based on the image information after segmentation of two adjacent frames to obtain the change in the growth rate of the microvessels; judging whether the growth rate of the microvessels is normal based on the change in the growth rate of the microvessels, and obtaining a second judgment result; adjusting the sampling interval of the brain CT image sequence based on the second judgment result, the change in the growth rate of the patient's microvessels can be determined through the image information after segmentation of two adjacent frames, and further adjusting the sampling interval based on the change in the growth rate of the patient's microvessels. When the second judgment result is that the growth rate change is faster, a shorter time interval should be selected, such as a few minutes to half an hour. Because in this case, short time interval imaging can capture the dynamic changes of microvessels in a timely manner, including slight expansion or contraction of the diameter, formation of new branches, etc.; when the second judgment result is that the growth rate changes slowly, a longer time interval can be selected, such as hours or even days. By determining the sampling interval based on the patient's own situation through this step, not only the accuracy of the segmentation is ensured, but also the image artifacts are reduced.

[0035] Step S2, performing feature extraction on the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence respectively to obtain first feature information, second feature information and third feature information, wherein the first feature information includes three-dimensional structural feature information of brain blood vessels, the second feature information includes brain tissue distribution feature information, and the third feature information includes morphological feature information of brain blood vessels;

[0036] The step S2 also includes step S21, step S22 and step S23, which are specifically:

[0037] Step S21, preprocessing the brain MRI image sequence to obtain preprocessed second image information;

[0038] In this step, preprocessing includes but is not limited to noise removal, intensity normalization, and spatial registration.

[0039] Step S22, segmenting the preprocessed image to obtain a brain tissue image;

[0040] In this step, the preprocessed image is segmented, and the segmentation algorithm for obtaining the brain tissue image is not limited.

[0041] Step S23: extracting features from the brain tissue image to obtain second feature information.

[0042] The step S23 also includes step S231, step S232, step S233, step S234 and step S235, which are specifically:

[0043] Step S231, calculating the volumes of different brain regions according to the brain tissue image to obtain volume feature information;

[0044] Step S232, performing wavelet transform on the brain tissue image to obtain fourth feature information;

[0045] In this step, the brain tissue image is processed using a double-layer wavelet transform to obtain fourth characteristic information.

[0046] Step S233, constructing a gray level co-occurrence matrix based on the fourth feature information;

[0047] Step S234, extracting texture feature information of brain tissue according to the gray level co-occurrence matrix;

[0048] In this step, the texture feature information of brain tissue is obtained by calculating the joint probability, contrast, correlation, energy and other features of different directions and distances through the gray level co-occurrence matrix.

[0049] Step S235: Obtain the second feature information according to the volume feature information and the texture feature information.

[0050] The small size of newborn microvessels is a key difficulty in segmentation. During the imaging process, due to resolution limitations, even advanced imaging technology may have difficulty in clearly distinguishing the boundaries of microvessels. Moreover, the signal intensity of microvessels is relatively weak and easily interfered by noise. In magnetic resonance angiography (MRA), the blood flow signals of microvessels may be masked by the signals of surrounding tissues, making it difficult to accurately identify and segment microvessels. In this embodiment, AI technology is used to set a feature extraction algorithm to process the brain MRI image sequence to obtain second feature information, so as to locate the brain tissue around the newborn microvessels through the second feature information, thereby eliminating the influence of the tissue around the microvessels on the imaging of newborn microvessels, thereby improving the accuracy of segmentation of newborn microvessels in the brain.

[0051] Step S3, fusing the first feature information, the second feature information and the third feature information to obtain fused feature map information;

[0052] In this step, a fusion algorithm is constructed based on AI technology to fuse the first feature information, the second feature information and the third feature information, specifically: the first feature information, the second feature information and the third feature information are combined into a matrix; the principal component analysis algorithm is used to reduce the dimension of the matrix to obtain the fused feature map information, and the fused feature map information fuses information of different modalities for subsequent processing.

[0053] Step S4, enhancing the fused feature map information to obtain enhanced feature map information;

[0054] Due to the small size of the newly formed microvessels and the limitation of resolution, it is necessary to further enhance the fused feature map information to improve the segmentation accuracy. In this step, the network architecture of the enhanced model is designed based on AI technology. The specific enhancement process is as follows: sending the fused feature map information to the separable convolution layer to obtain the fifth feature information; performing standard convolution on the fifth feature information to obtain the sixth feature information; sending the fifth feature information and the sixth feature information to the fully connected layer to obtain the enhanced feature map information.

[0055] Step S5: segment the microvessels in the brain according to the enhanced feature map information to obtain segmented image information.

[0056] The present invention uses artificial intelligence technology to construct an intelligent image fusion and enhancement system. Different neural networks are used to extract features from images of different modalities, and then different feature information is fused to obtain fused feature map information. With the help of this fused feature map information, the problem of microvascular signal blurring caused by factors such as interference from different tissue signals and image noise can be effectively eliminated, and the clarity of the newly formed microvessels in the brain can be significantly improved. Based on this image with greatly improved clarity, the segmentation accuracy of newly formed microvessels in the brain is further improved by using an image segmentation algorithm based on artificial intelligence, providing extremely accurate and reliable data support for early diagnosis of brain diseases, disease monitoring, and evaluation of treatment effects.

[0057] The step S5 also includes step S51, step S52 and step S53, which are specifically:

[0058] Step S51, obtaining a preset optical flow calculation model;

[0059] Step S52, calculating the optical flow field between images at adjacent time points using the preset optical flow calculation model;

[0060] In this step, the optical flow calculation model is a model for estimating the pixel motion in an image sequence. The optical flow field between images at adjacent time points can be calculated through the preset optical flow calculation model. The optical flow field represents the motion vector of each pixel in the image, which is obtained by solving the partial differential equation of the optical flow model. In the calculation process, an iterative optimization algorithm, such as a gradient descent method, can be used to gradually approach the optimal solution of the optical flow field. In a specific embodiment, the Horn-Schunck model is used to continuously iterate and update the optical flow vector until the change in the optical flow vector between two iterations is less than a set threshold, and the optimal solution of the optical flow field is obtained.

[0061] Step S53: segmenting the microvessels in the enhanced feature map information based on the optical flow field.

[0062] The step S53 also includes step S531, step S532, step S533, step S534 and step S535, which are specifically:

[0063] Step S531, obtaining annotation information, where the annotation information includes at least two blood vessel seed points annotated in the first frame image;

[0064] Step S532: determining the location information of the seed point based on the annotation information;

[0065] Step S533, performing boundary growth according to the position information of the seed point and the optical flow field to obtain a segmentation boundary;

[0066] Step S534, optimizing the segmentation boundary to obtain an optimized segmentation boundary;

[0067] In this step, since noise or locally inaccurate optical flow calculation may cause small protrusions on the segmentation boundary, it is necessary to use a corrosion operation to process the powder segmentation boundary, and then use a dilation operation to restore the basic shape of the blood vessel while filling any small holes that may exist, to achieve optimization of the segmentation boundary, thereby improving the segmentation accuracy of the newly formed microvessels in the brain.

[0068] Step S535: segment the microvessels in the enhanced feature map information according to the optimized segmentation boundary.

[0069] In this embodiment, since the segmentation of new microvessels is a dynamic process, the traditional segmentation method cannot accurately segment the microvessels. Therefore, the optical flow method can calculate the motion vector of each pixel in the image, thereby estimating the growth direction and speed of the microvessels, and assisting the segmentation algorithm to more accurately divide the microvessel boundaries. The present invention takes into account the dynamic evolution process of the microvessels and improves the accuracy of the segmentation of new microvessels.

[0070] The step S535 also includes step S5351, step S5352, step S5353 and step S5354, which are specifically:

[0071] Step S5351, calculating the area of ​​microvessels according to the segmented image information to obtain area information;

[0072] In this step, the area of ​​each segmented region is calculated and then compared with the known normal size range of microvessels. If the area of ​​a segmented region is significantly larger than the expected area of ​​a normal microvessel, it may be that multiple microvessels or microvessels and surrounding tissues are mistakenly divided into one region, which is over-segmentation.

[0073] Step S5352, determining the shape of the microvessels according to the segmented image information to obtain shape information;

[0074] In this step, normal microvessels usually have relatively simple geometric shapes, such as approximately circular or tubular. If the segmented region has a complex shape, with multiple branches, irregular protrusions or depressions, it may be that the tissue around the blood vessel or other blood vessels are mistakenly included, resulting in over-segmentation.

[0075] Step S5353, judging whether the segmented image information is over-segmented according to the area information and the shape information, and obtaining a first judgment result;

[0076] Step S5354: when the first judgment result is that over-segmentation exists, gradually reducing the blood vessel seed points in the annotation information until there is no over-segmentation in the segmented image information.

[0077] In this embodiment, after the initial segmentation, the seed points are adjusted based on the segmentation results. If the segmentation results are over-segmented, it may be that too many seed points are selected, resulting in unreasonable expansion of the vascular area. In this case, the number of seed points can be reduced; in addition, it may be that the unreasonable position leads to unreasonable expansion of the vascular area. In this case, it is first necessary to judge whether the vascular area around each seed point has an abnormal movement pattern (such as the movement direction does not match the expected microvascular growth direction or the movement speed is too fast / too slow) based on the optical flow field, and the seed points corresponding to the abnormal movement pattern are screened, the rationality of the screened seed points is re-evaluated, and the seed points are re-selected at a more appropriate position in combination with the new optical flow information. Through this feedback mechanism, the selection of seed points is continuously optimized, thereby improving the accuracy of the segmentation of new microvessels in the brain.

[0078] Embodiment 2:

[0079] like Figure 2As shown, this embodiment provides an AI-based image segmentation device, which includes an acquisition module 901, a feature extraction module 902, a first processing module 903, a second processing module 904 and a third processing module 905, wherein specifically:

[0080] An acquisition module 901 is used to acquire a brain CT image sequence, a brain MRI image sequence, and a brain MRA image sequence;

[0081] A feature extraction module 902 is used to perform feature extraction on the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence respectively to obtain first feature information, second feature information and third feature information, wherein the first feature information includes three-dimensional structural feature information of brain blood vessels, the second feature information includes brain tissue distribution feature information, and the third feature information includes morphological feature information of brain blood vessels;

[0082] A first processing module 903 is used to fuse the first feature information, the second feature information and the third feature information to obtain fused feature map information;

[0083] A second processing module 904 is used to enhance the fused feature map information to obtain enhanced feature map information;

[0084] The third processing module 905 is used to segment the microvessels in the brain according to the enhanced feature map information to obtain segmented image information.

[0085] In a specific embodiment of the present disclosure, the feature extraction module further includes a first processing unit, a second processing unit and a third processing unit, which are specifically:

[0086] A first processing unit, configured to preprocess the brain MRI image sequence to obtain preprocessed second image information;

[0087] A second processing unit is used to segment the preprocessed image to obtain a brain tissue image;

[0088] The third processing unit is used to extract features from the brain tissue image to obtain second feature information.

[0089] In a specific embodiment of the present disclosure, the third processing unit further includes a fourth processing unit, a fifth processing unit, a sixth processing unit, a seventh processing unit and an eighth processing unit, wherein specifically:

[0090] a fourth processing unit, configured to calculate the volumes of different brain regions according to the brain tissue image to obtain volume feature information;

[0091] a fifth processing unit, configured to perform wavelet transform on the brain tissue image to obtain fourth feature information;

[0092] a sixth processing unit, configured to construct a gray level co-occurrence matrix based on the fourth feature information;

[0093] a seventh processing unit, configured to extract texture feature information of brain tissue according to the gray level co-occurrence matrix;

[0094] An eighth processing unit is used to obtain the second feature information according to the volume feature information and the texture feature information.

[0095] In a specific implementation of the present disclosure, the third processing module further includes an acquisition unit, a ninth processing unit and a tenth processing unit, which are specifically:

[0096] An acquisition unit, used to acquire a preset optical flow calculation model;

[0097] A ninth processing unit, configured to calculate the optical flow field between images at adjacent time points using the preset optical flow calculation model;

[0098] The tenth processing unit is used to segment the microvessels in the enhanced feature map information based on the optical flow field.

[0099] It should be noted that, regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0100] Embodiment 3:

[0101] Corresponding to the above method embodiment, this embodiment also provides an AI-based image segmentation device. The AI-based image segmentation device described below and the AI-based image segmentation method described above can refer to each other.

[0102] Figure 3 8 is a block diagram of an AI-based image segmentation device 800 according to an exemplary embodiment. Figure 3 As shown, the AI-based image segmentation device 800 may include: a processor 801 and a memory 802. The AI-based image segmentation device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0103] The processor 801 is used to control the overall operation of the AI-based image segmentation device 800 to complete all or part of the steps in the above-mentioned AI-based image segmentation method. The memory 802 is used to store various types of data to support the operation of the AI-based image segmentation device 800, which may include, for example, instructions for any application or method operating on the AI-based image segmentation device 800, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules can be keyboards, mice, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the AI-based image segmentation device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include: Wi-Fi module, Bluetooth module, NFC module.

[0104] In an exemplary embodiment, the AI-based image segmentation device 800 can be implemented by one or more application-specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to perform the above-mentioned AI-based image segmentation method.

[0105] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned AI-based image segmentation method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions can be executed by the processor 801 of the AI-based image segmentation device 800 to complete the above-mentioned AI-based image segmentation method.

[0106] Embodiment 4:

[0107] Corresponding to the above method embodiment, a readable storage medium is also provided in this embodiment. The readable storage medium described below and the AI-based image segmentation method described above can refer to each other.

[0108] A readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the AI-based image segmentation method of the above method embodiment.

[0109] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0111] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An AI-based image segmentation method, characterized in that: include: Acquire brain CT image sequences, brain MRI image sequences, and brain MRA image sequences; performing feature extraction on the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence respectively to obtain first feature information, second feature information and third feature information, wherein the first feature information includes three-dimensional structural feature information of brain blood vessels, the second feature information includes brain tissue distribution feature information, and the third feature information includes morphological feature information of brain blood vessels; Fusing the first feature information, the second feature information, and the third feature information to obtain fused feature map information; Enhance the fused feature map information to obtain enhanced feature map information; The microvessels in the brain are segmented according to the enhanced feature map information to obtain segmented image information.

2. The AI-based image segmentation method according to claim 1, characterized in that: Feature extraction is performed on the brain CT image sequence, the brain MRI image sequence, and the brain MRA image sequence to obtain first feature information, second feature information, and third feature information, including: Preprocessing the brain MRI image sequence to obtain preprocessed second image information; Performing segmentation processing on the preprocessed image to obtain a brain tissue image; Feature extraction is performed on the brain tissue image to obtain second feature information.

3. The AI-based image segmentation method according to claim 2, characterized in that: Extracting features from the brain tissue image to obtain second feature information includes: Calculating the volumes of different brain regions according to the brain tissue image to obtain volume feature information; Performing wavelet transform on the brain tissue image to obtain fourth feature information; Constructing a gray level co-occurrence matrix based on the fourth feature information; Extracting texture feature information of brain tissue according to the gray-level co-occurrence matrix; The second feature information is obtained according to the volume feature information and the texture feature information.

4. The AI-based image segmentation method according to claim 1, characterized in that: Segmenting the microvessels in the brain according to the enhanced feature map information to obtain segmented image information, including: Get the preset optical flow calculation model; Calculating the optical flow field between images at adjacent time points using the preset optical flow calculation model; The microvessels in the enhanced feature map information are segmented based on the optical flow field.

5. The AI-based image segmentation method according to claim 4, characterized in that: Segmenting the microvessels in the enhanced feature map information based on the optical flow field includes: Acquire annotation information, where the annotation information includes at least two blood vessel seed points annotated in the first frame of image; Determine the location information of the seed point based on the annotation information; Perform boundary growth according to the position information of the seed point and the optical flow field to obtain a segmentation boundary; Optimizing the segmentation boundary to obtain an optimized segmentation boundary; The microvessels in the enhanced feature map information are segmented according to the optimized segmentation boundary.

6. The AI-based image segmentation method according to claim 5, characterized in that: After segmenting the microvessels in the enhanced feature map information according to the optimized segmentation boundary, the method further includes: Calculating the area of ​​microvessels according to the segmented image information to obtain area information; determining the shape of the microvessels according to the segmented image information to obtain shape information; determining whether the segmented image information has over-segmentation according to the area information and the shape information, and obtaining a first determination result; When the first judgment result is that over-segmentation exists, the blood vessel seed points in the annotation information are gradually reduced until there is no over-segmentation in the segmented image information.

7. An AI-based image segmentation device, characterized in that: include: An acquisition module, used for acquiring a brain CT image sequence, a brain MRI image sequence and a brain MRA image sequence; a feature extraction module, used to perform feature extraction on the brain CT image sequence, the brain MRI image sequence and the brain MRA image sequence respectively to obtain first feature information, second feature information and third feature information, wherein the first feature information includes three-dimensional structural feature information of brain blood vessels, the second feature information includes brain tissue distribution feature information, and the third feature information includes morphological feature information of brain blood vessels; A first processing module, used for fusing the first feature information, the second feature information and the third feature information to obtain fused feature map information; A second processing module, used for enhancing the fused feature map information to obtain enhanced feature map information; The third processing module is used to segment the microvessels in the brain according to the enhanced feature map information to obtain segmented image information.

8. The AI-based image segmentation device according to claim 7, characterized in that: The feature extraction module comprises: A first processing unit is used to preprocess the brain MRI image sequence to obtain preprocessed second image information; A second processing unit is used to segment the preprocessed image to obtain a brain tissue image; The third processing unit is used to extract features from the brain tissue image to obtain second feature information.

9. The AI-based image segmentation device according to claim 8, characterized in that: The third processing unit comprises: a fourth processing unit, configured to calculate the volumes of different brain regions according to the brain tissue image to obtain volume feature information; a fifth processing unit, configured to perform wavelet transform on the brain tissue image to obtain fourth feature information; a sixth processing unit, configured to construct a gray level co-occurrence matrix based on the fourth feature information; a seventh processing unit, configured to extract texture feature information of brain tissue according to the gray-level co-occurrence matrix; An eighth processing unit is used to obtain the second feature information according to the volume feature information and the texture feature information.

10. The AI-based image segmentation device according to claim 7, characterized in that: The third processing module comprises: An acquisition unit, used to acquire a preset optical flow calculation model; A ninth processing unit, configured to calculate the optical flow field between images at adjacent time points using the preset optical flow calculation model; The tenth processing unit is used to segment the microvessels in the enhanced feature map information based on the optical flow field.

Images