Medical image analysis method and equipment based on fallopian tube anatomical segmentation

Abstract

The invention relates to a medical image analysis method and equipment based on fallopian tube anatomical segmentation, and the method comprises the steps: obtaining a laparoscopic image, extracting the flowing direction and speed information of blood, and generating a blood flowing direction diagram; dividing the fallopian tube into three areas of an umbrella part, an euphausiid part and an isthmus part, extracting and marking structure edges of the three areas, generating an anatomical structure boundary diagram, performing image fusion processing on the blood flow direction diagram and the anatomical structure boundary diagram, and identifying a blood flow abnormal concentration area; and carrying out image segmentation and pathological feature identification on the abnormal blood flow concentration region, judging whether a pathological tissue exists or not, if so, marking the region, and generating prompt information. According to the technical scheme, by analyzing the blood flowing direction and speed and combining anatomical segmentation of the fallopian tube, the position of a bleeding point can be accurately recognized, an abnormal blood flowing area can be found in time, and a doctor is assisted in making a more accurate and safer decision in the operation process.

Description

Technical Field

[0001] This application relates to the field of medical imaging, and in particular to a medical image analysis method and device based on fallopian tube anatomical segmentation. Background Technology

[0002] In laparoscopic surgery, especially procedures involving the fallopian tubes, rapid and accurate identification and management of bleeding are crucial. The fallopian tubes, an important component of the female reproductive system, have a complex anatomical structure, including the fimbriae, ampulla, and isthmus, each with distinct physiological characteristics and functions. Precise localization of bleeding points in different anatomical locations, and making appropriate judgments based on blood flow diagrams, anatomical segments, and physiological characteristics, are extremely important during surgery.

[0003] However, traditional surgical methods rely heavily on the surgeon's experience and real-time visual observation, facing challenges such as time constraints and high operational difficulty. In emergency situations, doctors may be unable to quickly locate the bleeding point, easily leading to misdiagnosis or mistreatment, or even resorting to a "one-size-fits-all" approach, resulting in unnecessary tissue damage, functional loss, and even affecting the patient's fertility. Especially for different areas of the fallopian tube, where blood supply and pathological manifestations vary significantly, traditional methods cannot accurately differentiate and provide personalized treatment. Therefore, how to use advanced image analysis technology to quickly locate bleeding points, analyze blood flow trajectories, and combine this with anatomical structures and pathological characteristics to assist doctors in making more precise and rational treatment decisions has become an important issue for improving surgical efficiency and safety. Summary of the Invention

[0004] The purpose of this application is to solve the problems mentioned above, such as the difficulty in quickly locating bleeding points in emergency situations, the difficulty in accurately distinguishing the blood flow characteristics in different areas of the fallopian tube, and the inability to determine in real time whether there is any pathological tissue or foreign body residue.

[0005] According to one aspect of this application, a medical image analysis method based on fallopian tube anatomical segmentation is provided, comprising: Acquire laparoscopic images, analyze the blood distribution in the laparoscopic images, extract the blood flow direction and velocity information, and generate a blood flow direction map; The fallopian tube structure in the laparoscopic image is divided into three regions: fimbriae, ampullae, and isthmus. The structural edges of the three regions are extracted and marked to generate an anatomical boundary map. The fimbriae include the connection between the fallopian tube and the ovary, the isthmus includes the connection between the fallopian tube and the uterus, and the ampullae is the connection between the fimbriae and the isthmus. Image fusion processing is performed on the blood flow direction map and the anatomical structure boundary map to mark the contact boundaries between blood and the three regions, analyze the contact relationship between the blood flow trajectory and the structural edges of the three regions, and identify areas of abnormal blood flow concentration. The image segmentation and pathological feature recognition are performed on the area of ​​concentrated abnormal blood flow to determine whether there is pathological tissue. If so, the area is marked and a prompt message is generated.

[0006] Preferably, after generating the blood flow direction map, the method further includes: constructing a vector field model based on the blood flow direction and velocity information, and inversely calculating the spatial coordinates of the bleeding points. Specifically, Extract the points with maximum velocity gradients from the blood flow pattern and mark them as candidate bleeding points; The flow field distribution of different combinations of candidate bleeding points is simulated, and the bleeding points are determined by maximum likelihood estimation.

[0007] Preferably, after determining the bleeding point, the method further includes: Determine whether the bleeding point is located on the abdomen of the ampulla; If so, calculate the bleeding rate and amount of bleeding at the bleeding point, determine the size of the wound at the bleeding point, and if the wound size is less than a preset wound size threshold, output a prompt message suggesting conservative treatment.

[0008] Preferably, the step of extracting and marking the structural edges of the three regions to generate an anatomical boundary map includes: The anatomical segments of the fallopian tubes in the laparoscopic images are identified; An edge detection algorithm or a deep learning-based edge detection model is applied to extract the edges of each anatomical structure and mark the edges.

[0009] Preferably, the edge detection algorithm includes: Canny operator, Sobel operator, HED algorithm or CombineNet algorithm.

[0010] Preferably, the image fusion processing of the blood flow direction map and the anatomical structure boundary map includes: The blood flow direction map and the anatomical structure boundary map are fused at the channel level using multi-channel data fusion technology. The fusion methods include weighted averaging, maximum value synthesis, or fusion filtering. Perform boundary enhancement processing on the fused image; The relative position and contact between blood flow and anatomical structure boundaries are analyzed using spatial relationship modeling algorithms. Output the fused image as a multidimensional data graph.

[0011] Preferably, the identification of areas with concentrated abnormal blood flow includes: Edge extraction of the blood flow region includes: pixel-level feature extraction of the blood flow region; Based on blood flow velocity, flow rate, and blood diffusion range, the concentration of blood flow in each blood flow region is calculated; Determine whether the concentration of blood flow exceeds a preset concentration threshold. If so, mark the area as an area of ​​abnormally concentrated blood flow.

[0012] Preferably, the calculation of the concentration of blood flow in each blood flow region includes: calculating the concentration of blood flow in the blood flow region based on one or more of the following algorithms: K-Means clustering algorithm, DBSCAN algorithm, Otsu algorithm, Canny edge detection algorithm, U-Net algorithm, and CNN algorithm.

[0013] Preferably, image segmentation and pathological feature identification of the region with concentrated abnormal blood flow include: Image segmentation technology is used to initially extract the areas of abnormal blood flow, separating the blood areas from other normal tissue areas; Extract the shape and boundary features of the blood region; Using deep learning-based image classification, object detection, and lesion detection methods, the system identifies whether the areas of abnormal blood flow concentration belong to cysts, adhesions, or foreign bodies. Abnormal blood flow areas with pathological characteristics are labeled, and prompts including lesion type, disease diagnosis suggestions, and treatment plan suggestions are generated.

[0014] This invention also provides a medical image analysis device based on fallopian tube anatomical segmentation, which, using the aforementioned medical image analysis method based on fallopian tube anatomical segmentation, includes: The image acquisition and processing module is used to acquire laparoscopic images and perform preprocessing to generate blood flow direction maps and fallopian tube anatomical structure boundary maps. The identification and analysis module is used to analyze the blood flow direction map and the anatomical structure boundary map. Through image segmentation and pathological feature recognition technology, it locates bleeding points, assesses the wound size and bleeding rate of the bleeding points, and determines whether pathological tissue exists. Decision support module: Based on the analysis results of the identification and analysis module, it generates corresponding treatment suggestions; Interactive modules are used to display analysis results in the form of images, charts, or sound.

[0015] This application offers the following advantages: By acquiring laparoscopic images and analyzing blood flow direction and velocity, combined with the anatomical segmentation of the fallopian tube (fimbria, isthmus, ampulla), the location of bleeding points can be accurately identified. After confirming the bleeding point, the severity of bleeding can be further assessed by calculating the bleeding rate, flow rate, and wound size. If the wound is small, the system will automatically provide conservative treatment recommendations, preventing doctors from making overtreatment decisions in emergency situations and thus reducing harm to the patient. Through image segmentation and pathological feature identification of areas with abnormal blood flow, the presence of pathological tissue (such as cysts, adhesions, or residual fertilized egg fragments) can be determined. This function provides doctors with more accurate intraoperative diagnostic information, helping them avoid missing lesion areas or making incorrect treatments.

[0016] By employing image fusion, blood flow trajectory analysis, and modeling of regional contact relationships, the system can promptly identify areas of abnormal blood flow, assisting doctors in making more precise and safer decisions during surgery. Especially in emergency situations, it helps doctors avoid making indiscriminate treatment decisions and reduces the risk of misdiagnosis.

[0017] The entire process utilizes a combination of technologies such as image fusion, edge detection, and deep learning to effectively improve the efficiency of intraoperative judgment, reduce the workload and operation time of doctors, and improve the overall efficiency of the surgery.

[0018] This invention can not only detect blood flow and tissue structure, but also provide treatment suggestions based on the analysis results, supporting doctors to make accurate decisions in emergency situations, reducing errors and risks, and further improving the quality and safety of medical services. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a logic block diagram of a medical image analysis method based on fallopian tube anatomical segmentation according to an embodiment of this application. Detailed Implementation

[0021] To facilitate understanding of the present application, a more comprehensive description of the present application will be provided below with reference to the accompanying drawings. The accompanying drawings illustrate preferred embodiments of the present application. However, the present application may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the disclosure of the present application.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0023] Please refer to Figure 1 One embodiment of this application provides a medical image analysis method based on fallopian tube anatomical segmentation, comprising: S10. Acquire laparoscopic images, analyze the blood distribution in the laparoscopic images, extract the blood flow direction and velocity information, and generate a blood flow direction map.

[0024] In this step, it's important to note that laparoscopic images are acquired using laparoscopic equipment, typically employing high-resolution video or still images. These images should possess sufficient clarity to ensure accurate capture of key elements such as blood flow, anatomical structures, and blood vessels. Identifying and extracting blood regions from the image usually relies on image processing techniques, such as thresholding and color space conversion (e.g., RGB to HSV space), to distinguish the color difference between blood and surrounding normal tissue. Edge detection algorithms (such as the Canny operator and Sobel operator) are used to clearly define the boundaries of blood regions, ensuring accurate extraction of the blood distribution area. Deep learning models (such as U-Net) can be used for automated segmentation of blood regions, further improving accuracy, especially in small blood flow areas in complex environments. In an optional embodiment, optical flow can be used to estimate the direction and velocity of blood flow using pixel motion in video sequences or consecutive image frames. Optical flow calculates the motion vector of each pixel in the image, reflecting the direction and velocity of blood flow.

[0025] Particle tracking algorithms can also be used to estimate blood flow velocity by tracking particles or fluid markers in the blood. By analyzing successive image frames, a velocity vector field of blood flow can be generated.

[0026] Furthermore, by analyzing local regions, a region of interest (ROI) can be selected in the image, and the flow trend of blood can be inferred by analyzing the pixel changes within that region.

[0027] The extracted flow direction and velocity information form a vector field that can accurately describe the flow path and velocity distribution of blood.

[0028] Based on the extracted blood flow direction and velocity data, a blood flow direction map is generated. This map reflects the blood flow path and velocity information and is usually presented in the form of a vector map, with each vector representing a direction and velocity.

[0029] This diagram can visually show the direction and speed of blood flow, as well as areas of concentrated or dispersed flow, providing data support for subsequent steps (such as locating bleeding points and analyzing areas of abnormal flow).

[0030] Furthermore, after generating the blood flow direction map, the process also includes: constructing a vector field model based on the blood flow direction and velocity information, and inversely calculating the spatial coordinates of the bleeding points, specifically including: S11. Extract the points with maximum velocity gradients from the blood flow direction map and mark them as candidate bleeding points.

[0031] In this step, it's important to note that in the blood flow direction diagram, the velocity gradient represents the intensity of the velocity change and reflects the region of blood flow variation. Areas with large velocity gradients are often associated with potential bleeding points, as bleeding areas typically cause abnormal changes in local blood flow.

[0032] Gradient calculation methods (such as the Sobel operator and the Laplacian operator) are used to extract the intensity of changes in flow velocity. By calculating the local gradient at each point in the blood flow pattern, the regions with the most dramatic velocity changes can be identified.

[0033] Based on the maxima of the velocity gradient, these regions are labeled as candidate hemorrhage points. These regions are likely places where blood flow changes drastically, usually indicating that blood is flowing out from the hemorrhage point, and may be related to the location of the hemorrhage point.

[0034] After candidate bleeding points are marked, these points can be further filtered and verified to improve the accuracy of the final bleeding point location.

[0035] S12. Simulate the flow field distribution of different combinations of candidate bleeding points, and determine the bleeding points through maximum likelihood estimation.

[0036] In this step, it should be noted that numerical simulation methods are used to simulate the flow field distribution of candidate bleeding points. By simulating combinations of candidate bleeding points, the blood flow patterns at these points are calculated. The simulation results provide blood flow paths and trends under different bleeding point assumptions.

[0037] Flow field simulation can be based on fluid dynamics models and combined with the physical properties of blood flow (such as blood viscosity, flow velocity, and blood vessel morphology) for accurate calculations.

[0038] The maximum likelihood estimation (MLE) method is used to determine the most likely bleeding point based on the simulated flow field distribution and actual observed blood flow information. The MLE method identifies the bleeding point that best matches the data by evaluating the degree of matching between different candidate points and actual blood flow.

[0039] Maximum likelihood estimation can effectively filter out candidate regions that are unlikely to be bleeding points, thus improving the accuracy of the final bleeding point determination.

[0040] After identifying the bleeding point, the following is also included: S13. Determine whether the bleeding point is in the ampulla of Vater.

[0041] In this step, it's important to note that, based on the aforementioned information on fallopian tube anatomical segments, it's crucial to determine whether the candidate bleeding point is located in the ampulla of the fallopian tube. As the middle segment of the fallopian tube, the ampulla has a relatively large diameter and possesses a certain degree of blood flow and tissue regeneration capacity. Therefore, if the bleeding point is located in this area and the incision is small, conservative treatment can avoid unnecessary surgical intervention. This strategy minimizes harm to the patient and aligns with best practices in treatment, especially in emergency situations where rapid and accurate diagnosis can help doctors make more appropriate decisions. By analyzing laparoscopic images, the blood flow direction map is matched with the anatomical boundary map to determine whether the candidate bleeding point is located within the defined area of ​​the ampulla.

[0042] Specific judgment criteria can be based on anatomical annotations or algorithmic rules. For example, edge detection-based segmentation algorithms can be used to accurately divide the various regions of the fallopian tube, thereby achieving accurate ampulla localization.

[0043] S14. If yes, calculate the bleeding rate and amount of bleeding at the bleeding point, determine the size of the wound at the bleeding point, and output a prompt message suggesting conservative treatment if the wound size is less than the preset wound size threshold.

[0044] In this step, it's important to note that the aforementioned velocity information, combined with a fluid dynamics model, is used to estimate the bleeding velocity and volume. Then, by combining the previously provided blood flow direction and velocity data, the wound size at the bleeding point is estimated. Wound size reflects the severity of the bleeding; a larger wound usually indicates a larger volume of bleeding, potentially requiring immediate intervention.

[0045] Based on the distribution characteristics of blood flow, image segmentation and morphological processing techniques are used to estimate the area of ​​the bleeding point wound or other relevant indicators. If the wound size is determined to be smaller than a preset threshold (i.e., a smaller bleeding point), conservative treatment recommendations are output based on clinical experience and preset rules. This type of treatment typically includes methods such as drug control and local hemostasis to avoid excessive intervention or unnecessary surgical procedures.

[0046] The thresholds for wound size, bleeding rate, etc., should be set based on medical data and clinical experience, and can be calibrated using historical case data.

[0047] After identifying the bleeding point, the system should generate corresponding treatment suggestions based on the doctor's operational decision-making process, such as whether surgery is needed or whether conservative treatment should be adopted.

[0048] S20. The fallopian tube structure in the laparoscopic image is divided into three regions: fimbriae, ampullae, and isthmus. The structural edges of the three regions are extracted and marked to generate an anatomical boundary map. The fimbriae include the connection between the fallopian tube and the ovary, the isthmus includes the connection between the fallopian tube and the uterus, and the ampulla is the connection between the fimbriae and the isthmus.

[0049] In this step, it's important to note that segmenting and identifying the fallopian tube structures in the laparoscopic images is crucial for accurately locating the various anatomical regions of the fallopian tube. This provides a reliable anatomical basis for subsequent blood flow analysis and bleeding point localization. Specific details are as follows: The fimbriae: Located at the distal end of the fallopian tube, the fimbriae connect to the ovary. Their ends are umbrella-shaped, extending outwards and close to the ovary. The fimbriae are relatively broad and have finger-like appendages. In image processing, identifying the fimbriae helps distinguish their boundaries from the ovary or other tissues.

[0050] The isthmus is the narrowest part of the fallopian tube, connecting to the uterine horn. The isthmus has a smaller diameter and, compared to the fimbriae and ampulla, its lumen is narrower and less elastic. Because this area is narrow, any abnormal bleeding will quickly become visible in the localized area, making accurate identification of the isthmus crucial for diagnosis.

[0051] Ampulla: Located between the fimbriae and isthmus, the ampulla is in the middle segment of the fallopian tube. It lies in a relatively wide section of the middle fallopian tube. The ampulla is relatively large in structure and has a large lumen diameter. This area is thicker and has a higher blood flow, therefore, bleeding from it may be more severe. Due to its relatively obvious characteristics, precise localization helps in the rapid identification of bleeding points and abnormal areas.

[0052] By extracting and labeling the structural edges of these three regions, accurate anatomical boundary maps can be generated, laying the foundation for subsequent blood flow direction maps and the identification of abnormal flow areas.

[0053] In one specific embodiment, extracting and marking the structural edges of three regions to generate an anatomical boundary map includes: S21. Identify the anatomical segments of the fallopian tubes in laparoscopic images.

[0054] S22. Apply edge detection algorithms or deep learning-based edge detection models to extract the edges of each anatomical structure and label the edges. Edge detection algorithms include: Canny operator, Sobel operator, HED algorithm, or CombineNet algorithm.

[0055] In this embodiment, it should be noted that before anatomical segmentation, the laparoscopic images are preprocessed, including noise reduction, contrast enhancement, and brightness adjustment, to ensure image quality and facilitate subsequent structural identification.

[0056] Different parts of the fallopian tube in laparoscopic images are identified by applying segmentation methods based on anatomical features (such as color, texture, and shape). The anatomical segmentation process divides the fallopian tube into three main regions: the fimbriae, the ampulla, and the isthmus.

[0057] Anatomical structures can be segmented using trained deep learning models (such as convolutional neural networks, CNNs) or feature extraction can be performed using classic image processing techniques. Deep learning models can accurately identify and segment these anatomical structures by learning from large amounts of labeled data.

[0058] The Canny operator is a classic edge detection algorithm that effectively detects edge information in images. Its advantage lies in its ability to extract detailed information from images, making it particularly suitable for use in regions with dramatic edge changes.

[0059] The Sobel operator extracts edges by calculating the gradient of image grayscale values. It is suitable for detecting areas with large pixel variations in images, and is particularly effective for recognizing linear structures.

[0060] HED (Holistically-Nested Edge Detection) is a deep learning-based edge detection algorithm that can effectively identify edges in complex backgrounds, making it particularly suitable for complex medical images. This algorithm uses a multi-layered convolutional neural network (CNN) for end-to-end edge detection, enabling it to obtain more detailed and accurate edge information.

[0061] The CombineNet algorithm is another deep learning-based edge detection method that enhances the accuracy and robustness of edge detection by combining the outputs of multiple neural network models, making it suitable for complex medical image analysis.

[0062] By applying the edge detection algorithm described above, the system extracts the edges of the fallopian tube anatomical region in the laparoscopic image. The extracted edges are then marked, forming a clear boundary. This boundary information will serve as a crucial basis for subsequent fusion of the blood flow direction map and the anatomical structure boundary map.

[0063] After the edges are extracted, they will be marked with different colors or symbols to clearly distinguish different regions (umbilicus, ampulla, isthmus). This will help doctors quickly identify key anatomical structures in laparoscopic images and improve diagnostic efficiency.

[0064] S30. Perform image fusion processing on the blood flow direction map and the anatomical structure boundary map, mark the contact boundaries between blood and the three regions, analyze the contact relationship between the blood flow trajectory and the structural edges of the three regions, and identify areas of concentrated blood flow abnormalities. In this step, it should be noted that the purpose of image fusion is to combine the blood flow direction map and the anatomical structure boundary map to form a fused image, facilitating subsequent analysis of the relationship between the blood flow trajectory and the anatomical structures. The fused image can more clearly demonstrate the contact and interaction between blood flow and the anatomical structures of the fallopian tube (fimbria, ampulla, isthmus), thereby helping doctors better locate bleeding points and related abnormalities.

[0065] Image fusion aims to simultaneously display blood flow direction and structural boundary information in the same image, allowing doctors to intuitively understand the spatial relationship between blood flow and fallopian tube anatomy. Especially in complex medical scenarios, it helps to quickly locate areas of abnormal blood flow and provides a basis for clinical decision-making.

[0066] After image fusion processing, the contact boundaries between blood flow areas and anatomical structures can be marked on the fused image. These boundaries reflect the interaction between blood and the three regions of the fallopian tube. Through automated algorithms or manual annotation, doctors can quickly identify areas of abnormal blood flow, especially areas of concentrated bleeding.

[0067] Contact boundaries can be marked with different colors or dashed frames for easy identification by doctors. For example, areas of blood flow may be marked in red, anatomical boundary areas in blue or green, and overlapping areas in yellow or other striking colors.

[0068] In an optional embodiment, image fusion processing of the blood flow direction map and the anatomical structure boundary map includes: S31. Using multi-channel data fusion technology, the blood flow direction map and the anatomical structure boundary map are fused at the channel level. The fusion methods include weighted averaging, maximum value synthesis, or fusion filtering. It should be noted in this step that the blood flow direction map and the anatomical structure boundary map usually originate from different image channels. To fuse these two images, multi-channel data fusion technology can be used to combine blood flow information and anatomical structure boundary information into a single image channel. The fusion methods include: weighted average fusion, which weights the blood flow information and structural boundary information for each pixel, adjusting the weights of the two information sources in the fused image as needed; maximum value synthesis, which selects the information with the larger value from the blood flow direction map and the anatomical structure boundary map for each pixel, retaining the more important information source; and fusion filtering, which uses filters to filter and fuse the two images, smoothing noise during the fusion process while preserving boundary and flow direction information.

[0069] S32. Perform boundary enhancement processing on the fused image. The fused image may need further enhancement of its edge information. By using edge detection techniques, such as the Canny algorithm or the Sobel operator, the structural boundaries in the fused image can be further strengthened, making the boundaries of anatomical structures (umbilicus, ampulla, isthmus) and blood flow more obvious.

[0070] S33. Spatial relationship modeling algorithms are used to analyze the relative position and contact between blood flow and anatomical structure boundaries. These algorithms (such as image-based spatial analysis and geometric model construction) can analyze the relative position and contact between the blood flow trajectory and anatomical structure boundaries. This process helps determine whether blood flows along anatomical structures and identifies areas of abnormal concentration. During the analysis, special attention is paid to whether blood flow has strong contact with the edges of anatomical structures in certain areas. For example, blood pooling near the ampulla of Vater may indicate significant bleeding in that area, requiring close monitoring by the physician. Flow direction maps reflect the velocity and direction of blood flow; by comparing them with anatomical structure boundaries, it can be determined whether the flow is obstructed or guided by the anatomical structures. If blood flow stagnates or suddenly changes direction in certain areas, this may indicate the presence of bleeding points.

[0071] S34. Output the fused image in the form of a multidimensional data graph.

[0072] Furthermore, identifying areas of concentrated abnormal blood flow includes: Edge extraction of blood flow areas includes: S35. Pixel-level feature extraction of blood flow areas can be performed using the following methods to extract edge and region features: Edge detection algorithms, such as the Canny edge detection algorithm, the Sobel operator, and the HED algorithm (Holistically-Nested Edge Detection), are applied to images of blood flow areas to obtain details of the blood flow edges.

[0073] Deep learning models: Based on U-Net or other convolutional neural network (CNN) structures, the model is trained to perform pixel-level segmentation of images to more accurately extract blood flow areas and their edges.

[0074] S36. Based on blood flow velocity, flow volume, and blood diffusion range, calculate the concentration of blood flow within each blood flow region. This helps identify areas of concentrated blood flow as potential areas of abnormal blood flow.

[0075] The calculation methods include: Flow velocity: The intensity of blood flow is calculated based on the direction and velocity information of blood flow at each pixel in the image.

[0076] Blood flow rate: Blood flow rate is estimated based on the blood velocity and the volume of the area through which it passes.

[0077] Blood diffusion range: Calculate the spatial distribution of blood flow based on the blood diffusion range.

[0078] Based on these calculation results, the concentration value of the blood flow area can be generated.

[0079] S37. Determine whether the concentration of blood flow exceeds the preset concentration threshold. If so, mark the area as an area of ​​abnormally concentrated blood flow.

[0080] Optionally, calculating the concentration of blood flow within each blood flow region includes: using one or more of the following algorithms: K-Means clustering, DBSCAN, Otsu's algorithm, Canny edge detection, U-Net, and CNN. Specifically, K-Means clustering clusters pixels within the blood flow region and calculates the concentration of blood flow in each cluster based on its center location and distribution. DBSCAN uses density clustering to identify regions with concentrated blood flow, better handling irregularly shaped or noisy blood flow regions. Otsu's algorithm is used for global thresholding of images, automatically determining the concentration threshold within blood flow regions to distinguish abnormal areas. Canny edge detection helps extract edges from blood flow regions for further concentration analysis. U-Net uses a deep learning network to accurately segment blood regions, thus calculating the concentration of each region. CNN utilizes the feature learning capabilities of convolutional neural networks to identify abnormal patterns in blood flow and calculate concentration.

[0081] Once it is determined that the concentration of blood flow exceeds a threshold, the area can be marked as an area of ​​abnormally concentrated blood flow using image labeling technology.

[0082] In this embodiment, the calculation of the aforementioned "blood flow concentration" is specifically implemented by combining image features such as the velocity field, diffusion area, and pixel density of the blood region in the following manner: 1. Clustering algorithm-based concentration calculation (K-Means or DBSCAN): Feature vectors are constructed from the blood pixel regions extracted from the blood flow direction map. The features include: position coordinates (x, y), flow velocity, flow direction angle, and color channel (such as RGB or HSV value).

[0083] K-Means or DBSCAN are used to cluster the blood regions to obtain several blood sub-region clusters; Let the number of pixels in each sub-region be... The area is The flow concentration of this sub-region The definition of is:

[0084] The concentration of the overall blood flow region can be defined as the concentration of blood flow in all clusters. Maximum value or weighted average: or

[0085] 2. Based on edge detection algorithm (Canny) and diffusion range determination: The Canny algorithm is used to detect the edges of the blood region and extract the closed boundaries. Calculate the area A enclosed by the boundary, where N is the number of blood pixels, and define the concentration as:

[0086] If the concentration C is greater than the empirically set threshold, it is determined to be an area of ​​abnormal blood flow concentration.

[0087] 3. Extract concentrated regions based on deep learning models (U-Net or CNN): Using a pre-trained or custom-trained U-Net network, semantic segmentation is performed on blood regions in an image, outputting a probability map. , where P∈[0,1] represents the probability that each pixel is a blood region; Binary image B(x,y) is obtained by threshold segmentation, and continuous connected regions are counted. Calculate the average probability density or area-to-pixel ratio for each region, i.e., the concentration:

[0088] 4. Otsu's algorithm as a reference for determining concentration thresholds: The Otsu method was used to perform automatic grayscale thresholding on blood region images to determine the optimal segmentation threshold. Used to automatically determine the concentration threshold for judging "abnormal concentration" of blood. .

[0089] Because the blood distribution in abnormal bleeding areas exhibits spatial characteristics such as rapid local velocity, small diffusion range, and high pixel density, the concentration C defined above can be used to reflect potential bleeding points and lesion sites, thereby aiding in diagnosis.

[0090] Furthermore, the boundary shape of the region can be analyzed through morphological processing or region growth methods to confirm whether the region is an actual area of ​​abnormal blood flow.

[0091] S40. Perform image segmentation and pathological feature recognition on areas with concentrated abnormal blood flow to determine whether pathological tissue exists. If so, mark the area and generate a prompt message.

[0092] In this step, it should be noted that the process of image segmentation and pathological feature recognition for areas with concentrated abnormal blood flow mainly includes the following key steps: S41. Image segmentation: Image segmentation technology is used to initially extract areas of concentrated abnormal blood flow, separating the blood area from other normal tissue areas. The purpose of image segmentation is to separate areas of abnormal blood flow from other normal tissue areas, providing clear regional boundaries for the identification of pathological features.

[0093] Common image segmentation methods include: Threshold segmentation: Separating blood regions from the background in an image by setting a threshold.

[0094] Edge detection: The edges of the blood region are extracted using methods such as Canny edge detection and Sobel operator to ensure that abnormal areas are clearly identifiable.

[0095] Deep learning methods: Using deep learning models (such as U-Net) to learn global and local features of images in order to accurately segment areas with abnormal blood flow.

[0096] S42. Extract the shape and boundary features of the blood region. Extracting pathological features of areas with abnormal blood flow typically includes: Morphological characteristics, such as the size, shape, and smoothness of the boundaries of the region, help identify the possibility of abnormal tissue.

[0097] Texture features: Extract texture features from images using methods such as gray-level co-occurrence matrix (GLCM) to determine whether a region has malignant or pathological features.

[0098] Color characteristics: By analyzing the color difference between blood and surrounding tissues, it is possible to identify whether there is abnormal blood or diseased areas.

[0099] Applications of deep learning models: Using trained deep neural networks (such as convolutional neural networks CNN, ResNet, VGG, etc.) to classify segmented regions and determine whether they have pathological features (such as cysts, adhesions, foreign bodies, etc.).

[0100] In this step, pathological features of areas with abnormal blood flow are extracted, specifically including: 1. Shape characteristics of the extracted blood region: The blood region is segmented into a binary graph B(x,y), and for each connected region R... i Calculate the following features: area:

[0101] Perimeter (boundary length): The boundary line is obtained using an edge detection algorithm (such as Canny), and the perimeter is the sum of the number of edge pixels.

[0102] Compactness: Used to assess whether the shape of the bleeding area is approximately regular.

[0103] in For the area, For the perimeter, The closer it is to 1, the closer it is to a circle.

[0104] Irregularity can be defined as:

[0105] A higher value indicates a more irregular bleeding boundary, which usually foreshadows pathological abnormalities.

[0106] 2. Extract boundary features (sharpness of blood edge): Boundary strength is calculated using image gradient. Let the image grayscale be I(x,y), then the gradient magnitude is:

[0107] Take the boundary pixel set Ei and calculate the average boundary gradient strength:

[0108] Higher Values ​​typically indicate well-defined boundaries, while blurred boundaries may indicate diffuse bleeding or a lesion area.

[0109] 3. Methods for extracting pathological features: Based on the above shape and boundary features, combined with the velocity, density, and diffusion direction of blood flow, a set of pathological feature indicators is constructed as shown in Table 1 below:

[0110] Table 1 In clinical medicine, pathological bleeding areas typically present with: irregular bleeding patterns and blurred boundaries; high local blood concentration and abnormal flow velocity; and extensive and disordered blood diffusion within the area.

[0111] This invention utilizes image processing techniques to quantify the above phenomena into identifiable feature indicators, thereby enabling automatic identification and localization of pathological bleeding areas.

[0112] S43. Using deep learning-based image classification, target detection, and lesion detection methods, identify whether areas with abnormal blood flow concentration belong to cysts, adhesions, or foreign bodies. Mark abnormal blood flow areas with pathological characteristics and generate prompt information including lesion type, disease diagnosis suggestions, and treatment plan suggestions.

[0113] Lesion detection: By analyzing the pathological features of areas with abnormal blood flow, it is determined whether pathological tissue (such as cysts, adhesions, foreign bodies, etc.) exists in the area. In this process, the deep learning model uses the labeled data from its training to determine whether the region conforms to a certain pathological feature.

[0114] Classification and Labeling: Once an abnormal pathological area is identified, the system can label the area and generate corresponding prompts, including: Lesion type: such as whether it is a cyst, adhesion, etc.

[0115] Disease diagnosis suggestions: Based on the identification results, the system can generate preliminary diagnostic suggestions to help doctors determine the pathological condition.

[0116] Treatment recommendations: Based on the diagnosis results, generate treatment recommendations, such as whether further examinations are needed, whether surgery is required, etc.

[0117] In this step, to identify whether areas of concentrated abnormal blood flow belong to pathological tissues such as cysts, adhesions, or foreign bodies, a deep learning-based image recognition model is used for image classification, target detection, and lesion identification. This method includes the following steps: 1. Model selection and structural design This invention may employ one of the following network structures: (1) Image classification model: ResNet-50 Input: Image fragments of concentrated blood flow regions segmented from laparoscopic images (uniform size 224×224RGB).

[0118] Output: Classification result, belonging to one of the following categories: [normal, cyst, adhesion, foreign body, other lesions].

[0119] Loss function: Cross-Entropy Loss:

[0120] Where C is the number of categories. For real labels, Predict probabilities for the model.

[0121] (2) Object detection model: YOLOv5 or Faster R-CNN Input: Complete laparoscopic image or image of an area with abnormal blood flow; Output: The position coordinates [x, y, w, h] of each bounding box, and the category label (cyst, adhesion, foreign body, etc.); Loss function: Classification loss (cross entropy); Bounding box regression loss (GIoU or SmoothL1); The total loss is:

[0122] This represents the classification loss function, used to measure the accuracy of the model in identifying whether the abnormal blood area is a lesion type such as a cyst, adhesion, or foreign body. Examples of loss functions include Cross-Entropy Loss and Focal Loss. This represents the bounding box regression loss function, used to measure the deviation between the predicted lesion region bounding box and the true bounding box, such as IoU Loss or GIoU Loss; These are the loss weight coefficients, used to control the proportion of the contribution of classification loss and regression loss to the total loss, respectively. They are usually set during the training phase through hyperparameter tuning or automatic weighted learning strategies.

[0123] (3) Lesion detection model: U-Net or DeepLabV3+ Input: Laparoscopic image or map of area with abnormal blood flow; Output: Pixel-level segmentation map, with lesion areas marked (using labels such as "adhesion area" and "foreign body area"); Loss function: Dice loss + BCE (Binary Cross Entropy):

[0124] Where P is the predicted region and G is the ground truth region.

[0125] 2. Data Preparation and Training Process Data acquisition: Collect labeled laparoscopic surgical image data, indicating the lesion area and type; Preprocessing includes image enhancement, normalization, and data augmentation (rotation, scaling, cropping). Training methods: Use the Adam or SGD optimizer; Learning rate scheduling; Batch size is typically set to 16~32; Use GPUs to accelerate training (such as NVIDIA A100); Evaluation metrics: The model performance is evaluated using standards such as mAP (mean Average Precision), IoU (Intersection over Union), and F1-score.

[0126] 3. Model Application and Hint Generation After training, the model is deployed to the image recognition and analysis module to automatically identify whether areas of abnormal blood flow in newly input laparoscopic images belong to pathological tissue and generate diagnostic information, specifically including: Lesion type: Classification or detection label output by the model (e.g., "foreign body", "adhesion"); Diagnostic recommendations are provided based on the location of the lesion and the confidence level of the model, such as "chronic tubal adhesions may be present"; Treatment options: Based on the lesion area and confidence level, suggestions may include "conservative treatment is recommended" or "further laparoscopic intervention is recommended".

[0127] 4. Combining medical principles According to common medical knowledge: Adhesion: This usually manifests as blurred structural boundaries and irregular connections between tissues, which can be effectively extracted by segmentation models such as U-Net; Cysts: characterized by localized fluid filling and regular outlines, their appearance can be easily identified by ResNet or YOLO; Foreign objects: characterized by high contrast and abnormal texture, classification networks can quickly learn their edge features.

[0128] Furthermore, it also includes: S44, generating prompt information, marking pathological areas on the image, and generating text or voice prompt information. The prompt information may include: This area may be a lesion area, and further diagnosis is recommended.

[0129] If a specific type of lesion is present (such as a cyst or adhesion), it indicates what further action the doctor may need to take.

[0130] This invention also provides a medical image analysis device based on fallopian tube anatomical segmentation, which, using the aforementioned medical image analysis method based on fallopian tube anatomical segmentation, includes: The image acquisition and processing module is used to acquire laparoscopic images and perform preprocessing to generate blood flow direction maps and fallopian tube anatomical structure boundary maps. The identification and analysis module is used to analyze the blood flow direction map and anatomical structure boundary map. Through image segmentation and pathological feature recognition technology, it locates bleeding points, assesses the wound size and bleeding rate of bleeding points, and determines whether pathological tissue is present. Decision support module: Based on the analysis results from the identification and analysis module, it generates corresponding treatment suggestions; Interactive modules are used to display analysis results in the form of images, charts, or sound.

[0131] In this embodiment, it should be noted that the image acquisition and processing module uses a laparoscopic device for image acquisition to ensure high-resolution laparoscopic images. The acquired images must have sufficient clarity and detail for subsequent processing. During image preprocessing, operations such as noise reduction, contrast enhancement, and size normalization may be necessary to improve the accuracy and robustness of subsequent image analysis. After image preprocessing, a blood flow direction map and a fallopian tube anatomical boundary map are generated, providing basic data for subsequent analysis.

[0132] The identification and analysis module performs a comprehensive analysis of the generated blood flow direction map and fallopian tube anatomical structure boundary map. Specific steps include: Blood flow analysis: By analyzing information such as the speed, direction and spread of blood flow, potential bleeding points can be identified, and areas of abnormal concentration of blood flow can be further analyzed.

[0133] Anatomical structure analysis: Image segmentation and edge detection techniques were applied to extract and mark different anatomical regions of the fallopian tube, such as the fimbriae, ampulla, and isthmus, providing a basis for subsequent pathological feature identification.

[0134] Pathological feature identification: Based on deep learning or traditional machine learning methods, identify and analyze the presence of pathological tissues, such as cysts and adhesions, assess the nature of abnormal areas, and determine whether further treatment is needed.

[0135] Based on the results from the identification and analysis module, the decision support module generates treatment recommendations. These recommendations may include: Conservative treatment recommendation: When the wound is small, blood flow is not significant, or the pathological tissue is minor, the system may recommend conservative treatment.

[0136] Further examination or surgery: When a large wound, abnormal blood flow, or severe pathological tissue is found, the system may recommend further examination or surgical treatment.

[0137] Personalized support: The system generates personalized treatment recommendations based on the patient's specific circumstances (such as age, medical history, etc.) to provide more accurate decision support.

[0138] The interactive module delivers analysis results to doctors or patients through images, charts, or sound. Interaction methods may include: Image display: By highlighting bleeding points, pathological areas, and related anatomical structures, doctors can visually see the problem area.

[0139] Charts and graphs: These charts display data such as blood flow velocity, volume, and concentration, helping doctors better understand the bleeding situation.

[0140] Voice prompts: If needed, the system can also provide voice prompts to remind doctors of key areas or disease types in real time.

[0141] User interaction experience: The interaction module also has a simple and easy-to-use user interface, which supports doctors to perform interactive operations, view and analyze different results, and make further diagnoses or decisions.

[0142] Implementing the technical solution in steps S10-S40 of this embodiment, by acquiring laparoscopic images and analyzing the direction and speed of blood flow, combined with the anatomical segmentation of the fallopian tube (fimbria, isthmus, ampulla), the location of the bleeding point can be accurately identified. After confirming the bleeding point, the severity of bleeding can be further assessed by calculating the bleeding rate, flow rate, and wound size. If the wound is small, the system will automatically output conservative treatment suggestions to avoid doctors making overtreatment decisions in emergency situations, thereby reducing harm to the patient. By segmenting images and identifying pathological features in areas of abnormal blood flow, the presence of pathological tissue (such as cysts, adhesions, or residual fertilized egg fragments) can be determined. This function provides doctors with more accurate intraoperative diagnostic information, helping them avoid missing lesion areas or making incorrect treatments.

[0143] By employing image fusion, blood flow trajectory analysis, and modeling of regional contact relationships, the system can promptly identify areas of abnormal blood flow, assisting doctors in making more precise and safer decisions during surgery. Especially in emergency situations, it helps doctors avoid making indiscriminate treatment decisions and reduces the risk of misdiagnosis.

[0144] The entire process utilizes a combination of technologies such as image fusion, edge detection, and deep learning to effectively improve the efficiency of intraoperative judgment, reduce the workload and operation time of doctors, and improve the overall efficiency of the surgery.

[0145] This invention can not only detect blood flow and tissue structure, but also provide treatment suggestions based on the analysis results, supporting doctors to make accurate decisions in emergency situations, reducing errors and risks, and further improving the quality and safety of medical services.

[0146] The embodiments described above are merely illustrative of several implementations of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the appended claims.

Claims

1. A medical image analysis method based on fallopian tube anatomical segmentation, characterized in that, include: Acquire laparoscopic images, analyze the blood distribution in the laparoscopic images, extract the blood flow direction and velocity information, and generate a blood flow direction map; The fallopian tube structure in the laparoscopic image is divided into three regions: fimbriae, ampullae, and isthmus. The structural edges of the three regions are extracted and marked to generate an anatomical boundary map. The fimbriae include the connection between the fallopian tube and the ovary, the isthmus includes the connection between the fallopian tube and the uterus, and the ampullae is the connection between the fimbriae and the isthmus. Image fusion processing is performed on the blood flow direction map and the anatomical structure boundary map to mark the contact boundaries between blood and the three regions, analyze the contact relationship between the blood flow trajectory and the structural edges of the three regions, and identify areas of abnormal blood flow concentration. The image segmentation and pathological feature recognition are performed on the area of ​​concentrated abnormal blood flow to determine whether there is pathological tissue. If so, the area is marked and a prompt message is generated.

2. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 1, characterized in that, After generating the blood flow direction map, the process further includes: constructing a vector field model based on the blood flow direction and velocity information, and inversely calculating the spatial coordinates of the bleeding points. Specifically, Extract the points with maximum velocity gradients from the blood flow pattern and mark them as candidate bleeding points; The flow field distribution of different combinations of the candidate bleeding points is simulated, and the bleeding points are determined by maximum likelihood estimation.

3. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 1, characterized in that, After determining the bleeding point through maximum likelihood estimation, the process also includes: Determine whether the bleeding point is located on the abdomen of the ampulla; If so, calculate the bleeding rate and amount of bleeding at the bleeding point, determine the size of the wound at the bleeding point, and if the wound size is less than a preset wound size threshold, output a prompt message suggesting conservative treatment.

4. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 1, characterized in that, The step of extracting and marking the structural edges of the three regions to generate an anatomical boundary map includes: The anatomical segments of the fallopian tubes in the laparoscopic images are identified; An edge detection algorithm or a deep learning-based edge detection model is applied to extract the edges of each anatomical structure and mark the edges.

5. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 4, characterized in that, The edge detection algorithms include: Canny operator, Sobel operator, HED algorithm, or CombineNet algorithm.

6. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 1, characterized in that, The image fusion processing of the blood flow direction map and the anatomical structure boundary map includes: The blood flow direction map and the anatomical structure boundary map are fused at the channel level using multi-channel data fusion technology. The fusion methods include weighted averaging, maximum value synthesis, or fusion filtering. Perform boundary enhancement processing on the fused image; The relative position and contact between blood flow and anatomical structure boundaries are analyzed using spatial relationship modeling algorithms. Output the fused image as a multidimensional data graph.

7. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 1, characterized in that, The identified areas of concentrated abnormal blood flow include: Edge extraction of the blood flow region includes: pixel-level feature extraction of the blood flow region; Based on blood flow velocity, flow rate, and blood diffusion range, the concentration of blood flow in each blood flow region is calculated; Determine whether the concentration of blood flow exceeds a preset concentration threshold. If so, mark the area as an area of ​​abnormally concentrated blood flow.

8. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 7, characterized in that, The calculation of the concentration of blood flow in each blood flow region includes: calculating the concentration of blood flow in the blood flow region based on one or more of the following algorithms: K-Means clustering algorithm, DBSCAN algorithm, Otsu algorithm, Canny edge detection algorithm, U-Net algorithm, and CNN algorithm.

9. The medical image analysis method based on fallopian tube anatomical segmentation according to claim 1, characterized in that, Image segmentation and pathological feature identification of the aforementioned areas of concentrated abnormal blood flow include: Image segmentation technology is used to initially extract the areas of abnormal blood flow, separating the blood areas from other normal tissue areas; Extract the shape and boundary features of the blood region; Using deep learning-based image classification, object detection, and lesion detection methods, the system identifies whether the areas of abnormal blood flow concentration belong to cysts, adhesions, or foreign bodies. Abnormal blood flow areas with pathological characteristics are labeled, and prompts including lesion type, disease diagnosis suggestions, and treatment plan suggestions are generated.

10. A medical image analysis device based on fallopian tube anatomical segmentation, employing the medical image analysis method based on fallopian tube anatomical segmentation as described in claims 1-9, characterized in that, include: The image acquisition and processing module is used to acquire laparoscopic images and perform preprocessing to generate blood flow direction maps and fallopian tube anatomical structure boundary maps. The identification and analysis module is used to analyze the blood flow direction map and the anatomical structure boundary map. Through image segmentation and pathological feature recognition technology, it locates bleeding points, assesses the wound size and bleeding rate of the bleeding points, and determines whether pathological tissue exists. Decision support module: Based on the analysis results of the identification and analysis module, it generates corresponding treatment suggestions; Interactive modules are used to display analysis results in the form of images, charts, or sound.

Images