A 3D Reconstruction System for Laparoscopic Ovarian Tissue Wound Analysis

Abstract

This invention relates to the field of image data processing technology, specifically to a three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis. The system includes a data acquisition module, a feature analysis module, a heterogeneity feature analysis module, and a three-dimensional reconstruction module. These modules are used to acquire video frame images and tissue wound regions under laparoscopy, respectively. A focus index is obtained based on the shape and density distribution of each tissue wound region. The degree of deformation fluctuation is analyzed based on the deviation between the pixel distribution of each tissue wound region in consecutive video frame images and the overall distribution of each tissue wound region. A heterogeneity index is obtained by combining the balance of deformation fluctuations across different tissue wound regions. Based on the heterogeneity index and focus index of each tissue wound region, the tissue wound regions are labeled for three-dimensional reconstruction of the tissue wound. This invention achieves accurate quantification and differentiated modeling of wound features.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology, specifically to a three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis. Background Technology

[0002] Medical image processing and computer-aided surgical techniques have become important supports for laparoscopic gynecological surgery. Laparoscopic surgery, with its advantages of minimal trauma and rapid postoperative recovery, has become the mainstream surgical method for ovarian surgeries such as ovarian cyst removal and ovarian repair. The core of preoperative planning for laparoscopic ovarian suturing is to carry out high-precision three-dimensional reconstruction of the ovarian tissue wound, which provides digital model support for quantifying the degree of tissue wound damage and postoperative assessment, and has important clinical significance.

[0003] However, the ovary is a highly elastic soft tissue, prone to dynamic deformation under surgical manipulation, pneumoperitoneum pressure, instrument contact, and physiological pulsation. Furthermore, different wound areas exhibit heterogeneity in physical properties such as elasticity and toughness due to differences in tissue cell composition, vascular and nerve distribution, and collagen fiber density. In continuous video streams of ovarian wounds acquired during laparoscopy, wound features include both static spatial distribution information in a single frame and dynamic deformation response information over time. Existing techniques for ovarian wound feature analysis and 3D reconstruction only focus on single-dimensional static features such as wound area and spatial clustering in a single frame image. They fail to explore the dynamic deformation patterns of wound areas between consecutive video frames, nor do they consider the differences in deformation characteristics due to the heterogeneity of physical properties in different wound areas. This prevents the comprehensive quantification of multi-dimensional, dynamic, and static features of the ovarian wound area. Consequently, the resulting wound feature analysis deviates significantly from the actual clinical tissue characteristics, leading to insufficient accuracy in 3D reconstruction of ovarian tissue wounds. Summary of the Invention

[0004] To address the technical problem of low accuracy in 3D reconstruction of ovarian tissue wounds due to the consideration of single-dimensional static features in existing 3D reconstruction methods, this invention aims to provide a 3D reconstruction system for laparoscopic ovarian tissue wound analysis. The specific technical solution adopted is as follows: The data acquisition module is used to acquire video frame images under laparoscopy and the tissue wound area in each video frame image; The attention feature analysis module is used to obtain the attention index of each tissue wound area based on the shape distribution of each tissue wound area in the video frame image and the dense distribution within the neighborhood of each tissue wound area. The heterogeneity feature analysis module is used to analyze the degree of deformation fluctuation of the tissue wound area based on the deviation between the pixel distribution of each tissue wound area in continuous video frame images and the overall distribution of each tissue wound area. Combined with the balance of the degree of deformation fluctuation of different tissue wound areas, the heterogeneity index of each tissue wound area is obtained. The 3D reconstruction module is used to annotate tissue wound areas based on heterogeneity and attention indicators for 3D reconstruction of tissue wounds.

[0005] Preferably, the step of analyzing the deformation fluctuation degree of the tissue wound area based on the deviation between the pixel distribution of each tissue wound area in consecutive video frame images and the overall distribution of each tissue wound area specifically includes: The feature matching dataset for each tissue wound region is constructed by obtaining the tissue wound regions that match in adjacent video frame images for each tissue wound region. Based on the displacement fluctuation and displacement equilibrium features of each pixel in the target wound area and the corresponding pixel in the adjacent video frame images in the feature matching dataset, the deformation feature factor of the matched wound area in each video frame image is obtained. Wherein, the target wound area is any tissue wound area in the first video frame image, and the matching wound area is the tissue wound area in a single video frame image that matches the target wound area; The degree of deformation fluctuation in the target wound area is obtained by considering the balance of deformation characteristic factors in each matching wound area corresponding to the target wound area, and the relative characteristic distribution of deformation characteristic factors in each matching wound area.

[0006] Preferably, the step of obtaining the deformation feature factor of the matched wound region in each video frame image based on the displacement fluctuation features and displacement equalization features of each pixel in the matched wound region in each video frame image of the feature matching dataset of the target wound region and the corresponding pixels in adjacent video frame images specifically includes: Using the optical flow method, in the feature matching dataset of the target wound area, based on the displacement field between each pixel in the matched wound area in each video frame image and the next adjacent video frame image, the motion vector of each pixel in the matched wound area in each video frame image is obtained. Based on the average magnitude and standard deviation of the motion vectors of all pixels matching the wound area in each video frame image, the deformation feature factor of the matching wound area in each video frame image is determined.

[0007] Preferably, the step of obtaining the deformation fluctuation degree of the target wound region based on the balance of deformation feature factors of each matching wound region corresponding to the target wound region and the relative feature distribution of deformation feature factors of each matching wound region specifically includes: The minimum value of the deformation feature factor of all matching wound regions corresponding to the target wound region is obtained as the deformation reference value; The first feature factor is determined based on the deviation between the deformation feature factor of each matched wound region corresponding to the target wound region and the deformation reference value. The second feature factor is determined based on the mean of the deformation feature factors of all matching wound regions corresponding to the target wound region. The product of the first feature factor and the second feature factor is normalized to obtain the degree of deformation fluctuation in the target wound area.

[0008] Preferably, the heterogeneity index for each tissue wound region is obtained by combining the equilibrium of deformation fluctuations in different tissue wound regions, specifically including: The difference coefficient of the target wound area is obtained based on the deviation between the degree of deformation fluctuation of the target wound area and the equilibrium coefficient of the degree of deformation fluctuation of all tissue wound areas. Heterogeneity indices of the target wound area are determined by multiplying the degree of deformation fluctuation and the coefficient of difference in the target wound area.

[0009] Preferably, the step of obtaining the difference coefficient of the target wound area based on the deviation between the deformation fluctuation degree of the target wound area and the equilibrium coefficient of the deformation fluctuation degree of all tissue wound areas specifically includes: The difference coefficient of the target wound area is determined based on the absolute value of the difference between the deformation fluctuation degree of the target wound area and the mean of the deformation fluctuation degree of all tissue wound areas. The value range of the difference coefficient is [1,2].

[0010] Preferably, the step of obtaining the attention index for each tissue wound region based on the shape distribution of each tissue wound region in the video frame image and the dense distribution within the neighborhood of each tissue wound region specifically includes: In the feature matching dataset of the target wound area, the area of ​​the matching wound area in each video frame image is obtained to obtain the shape distribution coefficient of the matching wound area; The dense distribution weight of the matched wound region is obtained by acquiring the pixel percentage of all tissue wound regions within a preset range of the center pixel of the matched wound region in each video frame image; The product of the shape distribution coefficient and the dense distribution weight is normalized to obtain the wound attention of the matched wound region. The mean of the wound attention of all matched wound regions in the feature matching dataset of the target wound region is used as the attention index of the target wound region.

[0011] Preferably, the step of labeling tissue wound areas based on heterogeneity and attention indicators for each tissue wound area specifically includes: The normalized coefficient of the product between the heterogeneity index and the attention index of each tissue wound region is used as the importance level of each tissue wound region, and each tissue wound region is labeled according to the importance level.

[0012] Preferably, the step of marking each tissue wound area according to its importance specifically includes: Wound areas with an importance level greater than or equal to a preset importance threshold are labeled as high-importance wounds, while wound areas with an importance level less than the preset importance threshold are labeled as low-importance wounds.

[0013] Preferably, the method for obtaining the tissue wound area in each video frame image is as follows: For any video frame image, the gradient value of each pixel in the video frame image is obtained, and the optimal gradient segmentation threshold is obtained using the Otsu threshold segmentation algorithm. Pixels whose gradient values ​​are greater than the optimal gradient segmentation threshold are taken as edge points. Connectivity analysis is performed on all edge points to obtain the tissue wound area in the video frame image.

[0014] The embodiments of the present invention have at least the following beneficial effects: This invention first acquires laparoscopic video frames and locates the tissue wound area, completing the preprocessing of the original images and initial wound screening to remove interference such as optical distortion and noise, providing a clear and spatially consistent image foundation for subsequent wound feature analysis. Then, based on the wound shape distribution and neighborhood density distribution characteristics, a focus index is calculated to intuitively distinguish the clinical repair priority and reconstruction weight of different wounds, clarifying the key focus areas for 3D reconstruction. Furthermore, the degree of deformation fluctuation is analyzed by examining the pixel distribution deviation of consecutive video frames, and a heterogeneity index is obtained by combining the deformation balance between wounds, quantifying the differences in physical properties between ovarian tissue regions and compensating for the deficiency of traditional methods in ignoring dynamic deformation. Finally, the heterogeneity index and the focus index are integrated to label the wound area, guiding the reconstruction algorithm to use high-precision modeling for high-value wounds and conventional modeling for ordinary wounds, ultimately generating a 3D model that closely matches the actual morphology of the ovary, providing reliable support for preoperative suture planning. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis provided by the present invention; Figure 2 This is a schematic diagram of the heterogeneous feature analysis module provided by the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] 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 invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis provided by this invention.

[0020] Please see Figure 1 The diagram shows a schematic of a three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis provided by an embodiment of the present invention. The system includes a data acquisition module 10, a feature analysis module 11, a heterogeneous feature analysis module 12, and a three-dimensional reconstruction module 13.

[0021] The data acquisition module 10 is used to acquire video frame images under laparoscopy and the tissue wound area in each video frame image.

[0022] To achieve accurate 3D reconstruction of the tissue wound before laparoscopic ovarian suturing, the 3D reconstruction process relies on image data acquired via laparoscopy. Identifying the tissue wound area in the image provides the foundation for subsequent feature analysis and model construction, thus achieving high-precision reconstruction. However, the acquisition of video frames via laparoscopy is subject to issues such as lens optical distortion, acquisition noise, and inter-frame displacement. Furthermore, the ovarian wound area exhibits abrupt grayscale changes due to differences in physical properties compared to surrounding normal tissue. Directly using raw, unprocessed images cannot accurately identify the true tissue wound area, leading to biased analysis. This invention addresses the impact of laparoscopic image distortion on the accuracy of wound area identification, emphasizing that only by accurately extracting the tissue wound area can subsequent feature quantification and reconstruction be performed.

[0023] Therefore, in order to obtain accurate wound analysis targets, this embodiment of the invention first standardizes the laparoscopic video frame images, and then identifies and extracts the tissue wound region in each video frame image based on differences in tissue physical properties. Ideally, the processed image should be free of distortion and noise interference, and the tissue wound region should be completely and clearly identifiable. If the image is not preprocessed or the identification method is inappropriate, missed or incorrect detection of wound regions may occur. By combining image preprocessing with precise region extraction, image distortion interference can be eliminated, and the tissue wound region can be accurately located, making subsequent wound feature analysis more accurate.

[0024] As a specific example, video of the abdominal cavity is captured in real time by a laparoscopic camera and transmitted to a computer via a video capture card. Image frames are extracted from the continuous laparoscopic video stream at fixed time intervals at a preset frame rate (such as 30fps), and after grayscale processing, they are saved as ordered high-resolution video frame images.

[0025] In some embodiments, to ensure that the image quality meets the requirements of subsequent 3D reconstruction, preprocessing operations can be performed on the video frame images. Specifically, optical distortion correction can be performed on the images. Based on the intrinsic calibration parameters (focal length, principal point coordinates) and radial and tangential distortion coefficients of the laparoscopic lens, Zhang's calibration method is used to compensate for image distortion, eliminating image stretching and distortion caused by the laparoscopic optical lens and restoring the true spatial morphology of the ovarian wound. Gaussian filtering algorithm is used to remove noise generated during data acquisition and transmission in the video frame images, while preserving tissue edges and texture details, avoiding rough reconstructed surfaces due to noise. Inter-frame image registration can also be performed. Specifically, based on the Scale Invariant Feature Transform (SIFT) algorithm, key feature points of ovarian tissue in continuous image sequences are extracted. Through feature point matching and spatial transformation model solving, registration correction is performed on continuous frame images with slight displacement and rotation, eliminating inter-frame spatial deviations caused by slight shaking during laparoscopic operation and small displacements of ovarian tissue, ensuring spatial consistency between images.

[0026] It should be noted that the methods for grayscale conversion and preprocessing of images are well-known techniques and will not be elaborated here.

[0027] Furthermore, during the three-dimensional reconstruction of the tissue wound planned before laparoscopic ovarian suturing, the physical properties of the ovarian wound area differ from those of the surrounding normal tissue in the laparoscopic images. This results in a large difference in gray values ​​on both sides of the abrupt change (where the tissue wound meets the normal tissue), causing the edge of the tissue wound to exhibit a high gradient value. Based on this, the corresponding tissue wound area can be identified and determined from each video frame image based on the gradient feature differences between tissues, providing standardized basic data for subsequent wound feature analysis and three-dimensional modeling.

[0028] Specifically, for any video frame image, the gradient value of each pixel in the video frame image is obtained. The optimal gradient segmentation threshold is obtained using the Otsu thresholding algorithm. Pixels with gradient values ​​greater than the optimal gradient segmentation threshold are designated as edge points. Connected component analysis is performed on all edge points to obtain the tissue wound region in the video frame image. It should be understood that connected component analysis yields a closed region contour, within which lies a complete tissue wound region.

[0029] More specifically, the Sobel operator can be used to obtain the gradient value of each pixel in the video frame image. This gradient value reflects the rate of gray-level change in the neighborhood of the pixel. At the edge of the wound, due to the abrupt change in tissue properties, the gradient value usually shows a local peak.

[0030] The attention feature analysis module 11 is used to obtain the attention index of each tissue wound area based on the shape distribution of each tissue wound area in the video frame image and the dense distribution within the neighborhood of each tissue wound area.

[0031] To achieve a quantitative assessment of the static characteristics of tissue wound areas, it is necessary to clarify the clinical repair priorities of different wound areas during the three-dimensional reconstruction of the wound. By quantifying the shape distribution and neighborhood density distribution characteristics, the attention index of the wound area can be determined, thereby providing a basis for the allocation of reconstruction accuracy.

[0032] Ovarian tissue wounds vary in size and spatial distribution. The shape and distribution of the wound area directly reflect the extent of the damage, while dense distribution in the vicinity reflects the degree of damage concentration. If only the shape or dense distribution characteristics are considered, the importance of the wound area cannot be fully assessed, which will lead to an unreasonable division of reconstruction priorities.

[0033] This invention considers that the static spatial characteristics of the wound area are the core basis for preoperative suture planning and reconstruction precision allocation, and that a single-dimensional feature cannot fully characterize the clinical importance of the wound. Therefore, this invention, in order to comprehensively quantify the static characteristics of the wound, combines the shape distribution of each tissue wound area with the dense distribution within its neighborhood to obtain a corresponding attention index. Ideally, the attention index of the wound area should be positively correlated with the extent of the injury and the density of the distribution. If only a single-dimensional feature is considered, the attention index may not match actual clinical needs. Integrating both shape and density distribution features comprehensively reflects the static importance of the wound area, making the quantification of the attention index more aligned with clinical reality.

[0034] This invention transforms the shape distribution characteristics and neighborhood density distribution characteristics of tissue wound areas into quantifiable attention indicators. By calculating and normalizing the shape distribution coefficient and density distribution weight, it achieves accurate differentiation of the importance of different tissue wound areas, providing an objective quantitative basis for subsequent differentiated modeling of wound areas and priority treatment of key areas.

[0035] Relying solely on morphological features cannot fully reflect the importance of a tissue wound area, and it is necessary to avoid confusing wound areas that are "irregularly shaped but densely distributed" with those that are "regularly shaped but sparsely distributed." Simple visual observation is insufficient to quantify the priority differences between these two types of wound areas. This step first extracts the area of ​​the matching wound area in each video frame image from the feature matching dataset of the target wound area to obtain the shape distribution coefficient of the matching wound area; then, it obtains the pixel proportion of all tissue wound areas within a preset range of the center pixel of the matching wound area in each video frame image to obtain the dense distribution weight of the matching wound area; subsequently, it normalizes the product between the shape distribution coefficient and the dense distribution weight to obtain the wound attention of the matching wound area; finally, it uses the mean of the wound attention of all matching wound areas in the feature matching dataset of the target wound area as the attention index of the target wound area, achieving a precise transformation from morphological features and distribution features to quantitative indicators, solving the problem of the difficulty in objectively quantifying the importance of wound areas in traditional assessments.

[0036] It should be noted that laparoscopic videos are subject to slight camera shake and minor displacement or rotation of ovarian tissue. The pixel position and contour of the same physical wound may change slightly between adjacent frames. Mismatches can lead to the same wound being misidentified as different regions, compromising the uniqueness of wound identification. Before obtaining the attention index for each tissue wound region, it is necessary to perform frame-by-frame feature matching of tissue wound regions in adjacent video frames based on the contour features, grayscale features, and spatial location features of the wound region. This will allow for obtaining the corresponding matching tissue wound region for each tissue wound region in consecutive video frame images.

[0037] Specifically, the feature matching dataset for each tissue wound region is formed by obtaining the tissue wound regions that match each other in adjacent video frame images.

[0038] More specifically, any tissue wound region in the first video frame image is designated as the first wound region. The tissue wound region with the largest overlap area between the tissue wound region and the first wound region in the second video frame image is selected as the second wound region. Then, the tissue wound region with the largest overlap area between the tissue wound region and the second wound region in the third video frame image is selected as the third wound region. The first wound region is matched with the second wound region, the second wound region is matched with the third wound region, and so on, until the last video frame image. All matching tissue wound regions corresponding to the first wound region are obtained, which constitute the feature matching dataset of the first wound region.

[0039] It should be understood that, taking the tissue wound area in the first video frame as the object, there is a matching tissue wound area in each video frame in chronological order, and there is a matching relationship between each tissue wound area in the feature matching dataset.

[0040] As a concrete example, let's take any tissue wound area as an illustration. We'll denote any tissue wound area in the first video frame as the target wound area. All tissue wound areas that match the target wound area are called matching wound areas; that is, there is one matching wound area in each video frame.

[0041] The first step is to obtain the shape distribution coefficient of the matched wound region by acquiring the area of ​​the matched wound region in each video frame image from the feature matching dataset of the target wound region.

[0042] Since larger wound areas are often the main operating areas for ovarian suturing, the accurate reconstruction of their contours and shapes will directly affect the quantitative recording of the degree of tissue wound damage and the rationality of postoperative assessment. If there is a deviation in the reconstruction of such areas, it will easily lead to insufficient accuracy of the three-dimensional reconstruction results. Therefore, the area size shown by the shape distribution of the tissue wound is quantified.

[0043] Specifically, in the feature matching dataset of the target wound area, the total number of pixels in the matched wound area in each video frame image is obtained as the area of ​​the region, thus obtaining the shape distribution coefficient of the matched wound area. This coefficient reflects the size of the tissue wound area, and its value directly reflects the visual proportion of the wound in the image; the larger the value, the larger the actual damage area of ​​the wound.

[0044] The second step is to obtain the pixel ratio of all tissue wound areas within a preset range of the center pixel of the matching wound area in each video frame image to obtain the dense distribution weight of the matching wound area.

[0045] The dense distribution weight mainly quantifies the spatial aggregation characteristics of ovarian tissue damage, reflecting the concentrated distribution characteristics of tissue wounds. The more concentrated the tissue wound distribution, the more necessary it is to perform high-precision reconstruction operations as a whole as a joint repair area.

[0046] Based on this, for any matching wound area in any video frame image, the ratio of the total number of pixels of all tissue wound areas within a rectangular area of ​​a preset size centered on the center pixel of the matching wound area to the total number of pixels in the rectangular area is obtained. This ratio is used as the density distribution weight corresponding to the matching wound area. The proportion of tissue wound pixels within a preset range around the matching wound area is quantified by this ratio.

[0047] In this embodiment, the pixel corresponding to the centroid of the matched wound area is taken as the center pixel, and the preset range refers to the range centered on the center pixel. The square area, where M can be 50. In other embodiments, the implementer can determine it based on the size of the video frame image. In this embodiment, the value is an empirical value obtained through a large number of experiments.

[0048] It should be understood that the larger the value of the density distribution weight of the matching wound area, the higher the proportion of tissue wounds around the matching wound area and the greater the spatial density. In this case, there may be pixels of other tissue wound areas not only in the preset range, but also in the matching wound area. The ovarian tissue damage range is more concentrated.

[0049] The third step is to normalize the product between the shape distribution coefficient and the dense distribution weight to obtain the wound attention of the matched wound area.

[0050] Considering that large and concentrated wounds are the core operating area for ovarian suturing, area or density alone cannot fully represent its clinical priority. Multiplication can amplify the feature weight of the core area. The normalization process can employ a minimization-max normalization method, where the normalization range can specifically be all matching wound areas corresponding to the target wound area.

[0051] The fourth step is to use the average wound attention score of all matching wound regions in the feature matching dataset of the target wound region as the attention score index of the target wound region.

[0052] The attention index reflects the comprehensive static damage characteristics of the target wound area, which is also the comprehensive quantitative value of the static spatial characteristics of the ovarian wound. The larger the value, the larger and more concentrated the tissue wound is, which is the highest priority area for clinical repair and three-dimensional reconstruction. This aligns with the clinical need for precise treatment of large and dense wounds in laparoscopic ovarian suturing.

[0053] The heterogeneity feature analysis module 12 is used to analyze the degree of deformation fluctuation of the tissue wound area based on the deviation between the pixel distribution of each tissue wound area in continuous video frame images and the overall distribution of each tissue wound area, and to obtain the heterogeneity index of each tissue wound area by combining the balance of the degree of deformation fluctuation of different tissue wound areas.

[0054] To achieve accurate analysis of the dynamic deformation characteristics of the wound area, the dynamic changes of the ovarian elastic soft tissue need to be considered during the 3D reconstruction of the wound. The degree of deformation fluctuation is obtained by analyzing the pixel distribution deviation in consecutive frames, and a heterogeneity index is obtained by combining the deformation balance between regions, thereby restoring the true physical property differences of the wound tissue. As an elastic soft tissue, the ovary undergoes dynamic deformation under factors such as surgical manipulation and physiological pulsation. Different wound areas exhibit different deformation patterns due to differences in tissue composition. If only the static features of a single frame are analyzed, ignoring the dynamic deformation and inter-regional balance in consecutive frames, the heterogeneity of the tissue's physical properties cannot be reflected, leading to deviations between the reconstructed model and the actual intraoperative situation. This embodiment of the invention considers that dynamic deformation is a core feature of ovarian soft tissue, and the deformation balance between regions directly reflects tissue heterogeneity, which is crucial for improving reconstruction accuracy. Therefore, this embodiment of the invention quantifies the dynamic heterogeneity characteristics of the wound by analyzing the degree of deformation fluctuation based on the pixel distribution deviation in consecutive video frames, and obtaining a heterogeneity index by combining the deformation fluctuation balance of different regions. In the ideal state of dynamic feature analysis, heterogeneity indicators should be positively correlated with differences in tissue physical properties. If dynamic deformation or regional equilibrium is ignored, heterogeneity indicators may fail to accurately reflect tissue characteristics. However, by integrating deformation fluctuations and regional equilibrium characteristics, the dynamic heterogeneity of wound tissue can be accurately characterized, making subsequent reconstruction more consistent with the true deformation patterns of soft tissue.

[0055] As a concrete example, such as Figure 2 As shown, the heterogeneous feature analysis module 12 includes a first feature analysis unit 121, a second feature analysis unit 122, and a third feature analysis unit 123.

[0056] The first feature analysis unit 121 is used to obtain the deformation feature factor of the matched wound region in each video frame image based on the displacement fluctuation features and displacement equilibrium features of each pixel in the matched wound region in each video frame image of the target wound region and the corresponding pixel in the adjacent video frame images.

[0057] Because laparoscopic surgery is a minimally invasive procedure, even slight movements of the laparoscopic camera, minor traction of surgical instruments, fluid flow within the abdominal cavity, and even the patient's breathing during the operation can cause slight but continuous dynamic deformation of the ovarian tissue. The wound area will undergo non-rigid deformation and dynamic changes in overall displacement between consecutive video frames. Different wound areas have different tissue physical properties, and their deformation rate, amplitude, and pattern will also show significant differences. This is manifested in the fact that fibrotic areas with poor elasticity deform at a smaller amplitude and slower rate, while the soft, normal cortical areas deform at a larger amplitude and faster rate.

[0058] In order to achieve accurate extraction of pixel-level dynamic deformation of the matching wound area and reliable calculation of deformation feature factors, in the dynamic feature analysis of 3D reconstruction of ovarian wound, it is necessary to obtain pixel displacement information and quantify deformation features through scientific methods to provide accurate data support for the calculation of deformation feature factors.

[0059] In continuous laparoscopic video frames, the pixel displacement field matching the wound area contains core information about soft tissue deformation. Directly observing pixel displacement cannot quickly and accurately extract motion features, and a single displacement amplitude or dispersion cannot comprehensively quantify the degree of regional deformation, making it difficult to support accurate calculation of deformation feature factors. This invention considers that optical flow can efficiently acquire pixel displacement fields between consecutive frames, the average amplitude of motion vectors can reflect the overall deformation intensity of the wound, and the standard deviation of motion vector amplitudes can reflect the degree of deformation non-uniformity within the region. Combining these three factors can accurately support the determination of deformation feature factors.

[0060] Therefore, in order to accurately acquire pixel motion information and determine deformation feature factors, this embodiment of the invention utilizes optical flow to obtain the motion vector of each pixel in the matched wound region in each video frame image based on the displacement field between each pixel in the matched wound region and the next adjacent video frame image in the feature matching dataset of the target wound region. Based on the average amplitude and standard deviation of the motion vector amplitude of all pixels in the matched wound region in each video frame image, the deformation feature factors of the matched wound region in each video frame image are determined.

[0061] As a concrete example, in the feature matching dataset of the target wound area, for each video frame, the Farneback dense optical flow algorithm is used to calculate the pixel-level displacement field between the current frame and the next frame. This yields the displacement field between the matched wound area and the next image in each video frame. The displacement vector of each pixel in the displacement field is then used as the motion vector of each pixel in the matched wound area of ​​each video frame. It should be understood that the method of obtaining the pixel displacement field using optical flow is a well-known technique and will not be described further here. It should be noted that global motion compensation has been performed on adjacent video frames using feature registration techniques (such as the SIFT algorithm) to eliminate the global background displacement caused by the movement of the laparoscopic lens. The retained pixel motion vectors are used to characterize the local relative deformation of the tissue itself.

[0062] Furthermore, in the feature matching dataset of the target wound area, the arithmetic mean of the magnitudes of the motion vectors of all pixels in the matched wound area in each video frame image is calculated as the average magnitude of the motion vectors of the matched wound area in each video frame image. The standard deviation of the magnitudes of the motion vectors of all pixels in the matched wound area in each video frame image is calculated. The product between the average magnitude of the motion vectors of the matched wound area in each video frame image and the standard deviation of the motion vector magnitudes is determined as the deformation feature factor of the matched wound area in each video frame image.

[0063] It should be understood that a deformation feature factor corresponds to the matched wound region in a single video frame image within the feature matching dataset. The deformation feature factor reflects the complexity of the dynamic deformation of the tissue wound in a single frame image. The average magnitude of the displacement vector measures the macroscopic scale of the overall deformation of the matched wound region by calculating the arithmetic mean of the amplitudes of all pixel motion vectors, while the standard deviation measures the microscopic consistency of the internal deformation by analyzing the dispersion of pixel motion amplitudes. By multiplying the average magnitude of the displacement vector by the standard deviation, the dynamic deformation intensity and internal deformation inhomogeneity of the ovarian soft tissue in a single frame image are quantified, reflecting the magnitude of the impact of deformation in this region on the accuracy of 3D reconstruction.

[0064] The second feature analysis unit 122 obtains the degree of deformation fluctuation of the target wound area based on the balance of deformation feature factors of each matching wound area corresponding to the target wound area and the relative feature distribution of deformation feature factors of each matching wound area.

[0065] The target ovarian wound area corresponds to multiple matching wound areas in consecutive laparoscopic video frames. The deformation feature factors of each matching wound area can only reflect the deformation state of a single frame. A single deformation feature factor cannot characterize the overall deformation fluctuation pattern of the target wound area over time. Directly using single-frame data leads to a lack of benchmark for deformation fluctuation quantification, high dispersion of results, and difficulty in truly reflecting the continuous deformation characteristics of the wound. This embodiment of the invention considers that the minimum value of the deformation feature factor can serve as a stable deformation benchmark value to eliminate interference from extreme low values. The deviation of a single factor from the benchmark value can reflect the degree of deformation offset between frames, and the mean of all factors can reflect the overall average deformation level. A single deviation dimension cannot comprehensively characterize the deformation fluctuation amplitude.

[0066] Therefore, in order to accurately quantify the overall deformation fluctuation of the target wound area, this embodiment of the invention obtains the minimum value of the deformation feature factors of all matching wound areas corresponding to the target wound area as the deformation benchmark value; based on the deviation value between the deformation feature factors of each matching wound area corresponding to the target wound area and the deformation benchmark value, a first feature factor is determined; based on the mean value of the deformation feature factors of all matching wound areas corresponding to the target wound area, a second feature factor is determined; and the product of the first feature factor and the second feature factor is normalized to obtain the deformation fluctuation of the target wound area.

[0067] As a concrete example, if the nth tissue wound area in the first video frame is taken as the target wound area, the method for obtaining the deformation fluctuation degree of the target wound area can be expressed by the formula: ; in, This indicates the degree of deformation fluctuation in the target wound area, where n represents the nth tissue wound area (i.e., the target wound area) in the first video frame image. This represents the average value of the deformation feature factors corresponding to the target wound region in all video frame images, which is also the second feature factor; This represents the deformation feature factor of the matched wound region in the k-th video frame image. The minimum value of the deformation feature factor of all matching wound regions in the matching feature dataset of the target wound region is represented, which is also the deformation baseline value; K represents the total number of video frame images. This is a normalization function used to normalize the product result. For example, a minimax normalization algorithm can be used, where the maximum and minimum values ​​can be determined based on all tissue wound areas in the first video frame image.

[0068] As the first feature factor, it reflects the balance of the total deviation of relative deformation between each matched wound area and the baseline in the time dimension. By assessing the balance of instantaneous deformation in a single frame, the deformation characteristics of the overall balanced distribution in the time dimension can be evaluated, which can capture the dynamic change characteristics of tissue.

[0069] Second characteristic factor This reflects the inherent dynamic deformation level of the wound tissue. Areas with higher overall deformation intensity have a more critical impact on reconstruction due to their fluctuations, and therefore require higher weighting. By coupling the average deformation intensity with the cumulative total deviation, the more severe the deformation and the more obvious the fluctuations, the more exponentially the index value is amplified.

[0070] The degree of deformation fluctuation in the target wound area reflects the comprehensive dynamic deformation characteristic value. The larger the value of the deformation fluctuation, the more drastic the changes in the shape and position of the wound area between consecutive frames. The difficulty of inter-frame registration increases exponentially, and traditional reconstruction is prone to model distortion, edge blurring, and point cloud misalignment, necessitating the use of high-precision dynamic modeling. The smaller the value, the more stable the deformation of the target wound area between frames, the better the spatial consistency, and the more accurate the model can be guaranteed by conventional reconstruction with lower computational resource consumption.

[0071] The third feature analysis unit 123 is used to combine the balance of deformation fluctuation in different tissue wound areas to obtain the heterogeneity index of each tissue wound area.

[0072] Due to differences in tissue composition and physical properties, the deformation fluctuations of various ovarian wound areas vary significantly. Relying solely on the deformation fluctuation of the target wound area cannot reflect its deviation from the overall deformation level of all wound areas, making it difficult to accurately distinguish the deformation specificity of each wound area. This leads to a lack of comparative benchmarks for subsequent heterogeneity analysis. This invention considers that a deformation fluctuation equilibrium coefficient across all tissue wound areas can reflect the overall average deformation level. The deviation of the target wound area from this equilibrium coefficient directly reflects its deformation uniqueness. A single deformation fluctuation index cannot quantify the differences between regions.

[0073] Therefore, in order to accurately characterize the deformation differences in the target wound area, this embodiment of the invention obtains the difference coefficient of the target wound area based on the deviation between the deformation fluctuation degree of the target wound area and the equilibrium coefficient of the deformation fluctuation degree of all tissue wound areas. By calculating the deviation between the overall equilibrium coefficient and the deformation fluctuation of the target area, the regional deformation differences can be accurately quantified, making the difference coefficient more consistent with the actual deformation distribution pattern of the ovarian wound area.

[0074] As a specific example, the difference coefficient of the target wound area is determined based on the absolute value of the difference between the degree of deformation fluctuation of the target wound area and the mean of the degree of deformation fluctuation of all tissue wound areas. The value range of the difference coefficient is [1,2].

[0075] In one embodiment of the present invention, taking the target wound area as an example, the method for obtaining the difference coefficient of the target wound area can be expressed by the formula: ; in, This represents the difference coefficient of the target wound area. This indicates the degree of deformation fluctuation in the target wound area, where n represents the nth tissue wound area (i.e., the target wound area) in the first video frame image. This represents the average degree of deformation fluctuation across all tissue wound areas. This is a normalization function used to normalize the absolute value of the difference. For example, you can choose the minimax normalization method. The maximum and minimum values ​​can be determined based on all tissue wound areas. Adding 1 makes the final difference coefficient fall within the range of [1,2].

[0076] It reflects the absolute deviation of the target wound area from the average fluctuation of all wounds, and is the original quantitative value of the difference in deformation patterns between regions. It characterizes the uniqueness of the target wound area from the benchmark of ordinary wounds. The larger the deviation, the more incompatible the tissue physical properties and deformation patterns of the target wound area are with other wounds. Areas with large deviations are mostly wounds with special anatomical structures such as the ovarian cortex (high elasticity), medulla (low elasticity), and fibrotic scars (inelasticity), which are the core focus areas for surgery and reconstruction.

[0077] Furthermore, 1 represents the basic weight of its own fluctuation. When the target wound area is no different from the group, its own basic value of deformation fluctuation should be retained to avoid the difference coefficient being 0, which would cause the heterogeneity calculation to fail.

[0078] The heterogeneity of ovarian wounds is determined by both their own dynamic deformation fluctuations and the relative deformation differences to the overall wound. Using only one indicator, such as the degree of deformation fluctuation or the coefficient of variation, cannot fully characterize the overall heterogeneity of the wound area in terms of tissue properties and deformation patterns. This leads to heterogeneity indicators failing to accurately reflect the differences in soft tissue physical properties. This invention considers that the degree of deformation fluctuation reflects the intensity of deformation within the target wound area itself, while the coefficient of variation reflects its deviation from the overall wound deformation. Combining these two indicators can comprehensively characterize the regional heterogeneity.

[0079] Therefore, in order to accurately determine the heterogeneity index of the target wound area, this embodiment of the invention determines the heterogeneity index of the target wound area based on the product of the deformation fluctuation degree and the difference coefficient of the target wound area.

[0080] As a concrete example, the product of the deformation fluctuation degree and the difference coefficient of the target wound area is used as an indicator of the heterogeneity of the target wound area. This calculation process couples the product of its own deformation characteristics and relative differences to reflect the dynamic differences in the tissue's physical and mechanical properties.

[0081] The heterogeneity index of the target wound area reflects the dynamic deformation characteristics, fundamentally solving the problem of insufficient reconstruction accuracy caused by neglecting the heterogeneity of ovarian soft tissue in traditional methods.

[0082] Thus far, this module has designed core quantitative indicators for the heterogeneity of physical properties of the elastic soft tissue of the ovary (different regions have different elasticity, toughness, tensile strength, etc.). The calculation process progresses from single-frame dynamic deformation quantification to time-series deformation fluctuation quantification, and then to the quantification of differences in deformation patterns between regions. It transforms the differences in the internal tissue physical properties of the ovarian wound area into calculable and comparable external dynamic deformation characteristic indicators, and finally accurately characterizes the degree of difference between a single wound area and the surrounding normal tissue and other wound areas in terms of tissue characteristics and deformation patterns.

[0083] It should be noted that the method for determining the target wound area and the matching wound area in this module is the same as the method described in the feature analysis module. That is, the target wound area is any tissue wound area in the first video frame image, and the matching wound area is the tissue wound area in a single video frame image that matches the target wound area.

[0084] The 3D reconstruction module 13 is used to annotate the tissue wound area based on the heterogeneity index and attention index of each tissue wound area for the 3D reconstruction of the tissue wound.

[0085] To achieve differentiated and high-precision 3D reconstruction of ovarian tissue wounds, the 3D reconstruction process requires hierarchical labeling based on the core features of the wound area. By integrating heterogeneity indicators and attention indicators, appropriate reconstruction strategies are assigned to different wound areas, thereby improving overall reconstruction accuracy and clinical fit. Traditional 3D reconstruction does not differentiate the importance of wound areas and uses a uniform reconstruction precision, which can easily lead to the loss of details in key wound areas and the waste of resources in non-critical areas. This fails to meet the need for accurate assessment of wound morphology in preoperative suturing planning and affects the rationality of the suturing plan. The embodiments of this invention take into account that heterogeneity indicators reflect the dynamic characteristics of tissue and attention indicators reflect static importance. The integration of the two can comprehensively characterize the reconstruction priority of wound areas.

[0086] Therefore, in order to achieve targeted 3D reconstruction, this embodiment of the invention annotates the wound regions based on heterogeneity and attention indicators for each tissue wound area, and applies the annotation results to the 3D reconstruction of the tissue wound. Ideally, in 3D reconstruction, high-importance wound areas should use high-precision reconstruction algorithms, while low-importance areas should use conventional algorithms. Without hierarchical annotation, a mismatch between reconstruction accuracy and wound requirements may occur. However, by using dual-indicator fusion annotation of wound regions, differentiated reconstruction can be achieved, making the 3D reconstruction of ovarian wounds more accurate and better suited to clinical surgical planning needs.

[0087] As a concrete example, the normalized coefficient of the product between the heterogeneity index and the attention index of each tissue wound area is used as the importance of each tissue wound area.

[0088] The normalization coefficient can be processed using the minimax normalization method to calculate the product. Its maximum and minimum values ​​can be determined based on the product calculation results corresponding to all tissue wound areas, which will not be elaborated here.

[0089] Attention index reflects the inherent static damage properties of the wound, providing a static clinical weight for importance; the larger the wound area and the more concentrated the distribution, the higher the static baseline priority. Heterogeneity index reflects the differences in the dynamic mechanical properties of the wound, providing a dynamic technical weight for importance; the stronger the tissue heterogeneity and the more severe the deformation in the wound area, the higher the required reconstruction accuracy.

[0090] The higher the importance of each tissue wound region, the more critical that region is during surgery, and the greater its impact on reconstruction accuracy, quantification of tissue wound damage, and postoperative assessment. Higher resolution modeling methods (such as dense point cloud reconstruction) are required, and the closure and support of the wound region should be prioritized during wound feature quantification and postoperative tissue recovery assessment. The physical meaning of importance is the weighting of subsequent computational resource allocation or the level of feature quantification accuracy.

[0091] The importance of the wound area reflects the ultimate decision-making indicator that deeply integrates static spatial damage characteristics and dynamic tissue heterogeneity characteristics. It is the core hub connecting quantitative wound analysis, 3D reconstruction accuracy allocation, and preoperative planning for laparoscopic suturing. Its calculation process achieves unified quantification of clinical repair priority and 3D reconstruction technology weight, while the selection of high- and low-importance wounds directly determines the modeling accuracy, computational resource allocation, and surgical operation focus, fundamentally solving the core problem of traditional reconstruction ignoring wound differences.

[0092] In one embodiment of the present invention, each tissue wound area is marked according to the degree of importance.

[0093] Specifically, tissue wound areas with an importance level greater than or equal to a preset importance threshold are marked as high-importance tissue wounds, while tissue wound areas with an importance level less than the preset importance threshold are marked as low-importance tissue wounds.

[0094] The important threshold is set to 0.7, which can be set by the implementer according to the specific implementation scenario. This threshold is an empirical value derived from a large number of experiments.

[0095] Furthermore, for wounds of high importance, the focus is on the core repair areas that are large, dense, highly heterogeneous, and undergo dramatic deformation. High-precision reconstruction using dense point clouds is employed to generate pixel-level dense point clouds, improving model accuracy and restoring the subtle morphology and edge features of the original wound. For wounds of low importance, the focus is on smaller, more dispersed, homogeneous, and stable deformation secondary areas, which are reconstructed using conventional methods to save resources. This calculation and screening logic closely aligns with the physical characteristics of ovarian soft tissue and the clinical needs of laparoscopic surgery, fundamentally improving the accuracy and clinical fit of 3D reconstruction and providing a scientific and reliable quantitative basis for preoperative planning of ovarian suturing.

[0096] This invention abandons the crude approach of applying uniform precision to all wounds, and instead achieves graded and targeted reconstruction, improving the overall model accuracy. It automatically marks core operation areas to assist doctors in scientifically quantifying the degree of tissue wound damage and postoperative assessment. It achieves high precision for high-importance wounds and routine reconstruction for low-importance wounds, ensuring reconstruction quality while avoiding waste of computational resources.

[0097] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis, characterized by, The system includes: The data acquisition module is used to acquire video frame images under laparoscopy and the tissue wound area in each video frame image; The attention feature analysis module is used to obtain the attention index of each tissue wound area based on the shape distribution of each tissue wound area in the video frame image and the dense distribution within the neighborhood of each tissue wound area. The heterogeneity feature analysis module is used to analyze the degree of deformation fluctuation of the tissue wound area based on the deviation between the pixel distribution of each tissue wound area in continuous video frame images and the overall distribution of each tissue wound area. Combined with the balance of the degree of deformation fluctuation of different tissue wound areas, the heterogeneity index of each tissue wound area is obtained. The 3D reconstruction module is used to annotate tissue wound areas based on heterogeneity and attention indicators for 3D reconstruction of tissue wounds.

2. The three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 1, characterized in that, The step of analyzing the deformation fluctuation degree of the tissue wound area based on the deviation between the pixel distribution of each tissue wound area in consecutive video frame images and the overall distribution of each tissue wound area specifically includes: The feature matching dataset for each tissue wound region is constructed by obtaining the tissue wound regions that match in adjacent video frame images for each tissue wound region. Based on the displacement fluctuation and displacement equilibrium features of each pixel in the target wound area and the corresponding pixel in the adjacent video frame images in the feature matching dataset, the deformation feature factor of the matched wound area in each video frame image is obtained. Wherein, the target wound area is any tissue wound area in the first video frame image, and the matching wound area is the tissue wound area in a single video frame image that matches the target wound area; The degree of deformation fluctuation in the target wound area is obtained by considering the balance of deformation characteristic factors in each matching wound area corresponding to the target wound area, and the relative characteristic distribution of deformation characteristic factors in each matching wound area.

3. The three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 2, characterized in that, The deformation feature factor of the matched wound region in each video frame image is obtained by matching the displacement fluctuation features and displacement equalization features of each pixel in the wound region in each video frame image of the target wound region with the corresponding pixels in adjacent video frame images, based on the feature matching dataset of the target wound region. Specifically, this includes: Using the optical flow method, in the feature matching dataset of the target wound area, based on the displacement field between each pixel in the matched wound area in each video frame image and the next adjacent video frame image, the motion vector of each pixel in the matched wound area in each video frame image is obtained. Based on the average magnitude and standard deviation of the motion vectors of all pixels matching the wound area in each video frame image, the deformation feature factor of the matching wound area in each video frame image is determined.

4. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 2, characterized in that, The step of obtaining the deformation fluctuation degree of the target wound area based on the balance of deformation feature factors of each matching wound area corresponding to the target wound area and the relative feature distribution of deformation feature factors of each matching wound area specifically includes: The minimum value of the deformation feature factor of all matching wound regions corresponding to the target wound region is obtained as the deformation reference value; The first feature factor is determined based on the deviation between the deformation feature factor of each matched wound region corresponding to the target wound region and the deformation reference value. The second feature factor is determined based on the mean of the deformation feature factors of all matching wound regions corresponding to the target wound region. The product of the first feature factor and the second feature factor is normalized to obtain the degree of deformation fluctuation in the target wound area.

5. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 2, characterized in that, The heterogeneity index for each tissue wound region is obtained by combining the equilibrium of deformation fluctuations in different tissue wound regions, specifically including: The difference coefficient of the target wound area is obtained based on the deviation between the degree of deformation fluctuation of the target wound area and the equilibrium coefficient of the degree of deformation fluctuation of all tissue wound areas. Heterogeneity indices of the target wound area are determined by multiplying the degree of deformation fluctuation and the coefficient of difference in the target wound area.

6. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 5, characterized in that, The difference coefficient of the target wound area is obtained based on the deviation between the deformation fluctuation degree of the target wound area and the equilibrium coefficient of the deformation fluctuation degree of all tissue wound areas. Specifically, it includes: The difference coefficient of the target wound area is determined based on the absolute value of the difference between the deformation fluctuation degree of the target wound area and the mean of the deformation fluctuation degree of all tissue wound areas. The value range of the difference coefficient is [1,2].

7. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 2, characterized in that, The method of obtaining the attention index for each tissue wound region based on the shape distribution of each tissue wound region in the video frame image and the dense distribution within the neighborhood of each tissue wound region specifically includes: In the feature matching dataset of the target wound area, the area of ​​the matching wound area in each video frame image is obtained to obtain the shape distribution coefficient of the matching wound area; The dense distribution weight of the matched wound region is obtained by acquiring the pixel percentage of all tissue wound regions within a preset range of the center pixel of the matched wound region in each video frame image; The product of the shape distribution coefficient and the dense distribution weight is normalized to obtain the wound attention of the matched wound region. The mean of the wound attention of all matched wound regions in the feature matching dataset of the target wound region is used as the attention index of the target wound region.

8. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 1, characterized in that, The process of labeling tissue wound areas based on heterogeneity and attention indicators for each tissue wound area specifically includes: The normalized coefficient of the product between the heterogeneity index and the attention index of each tissue wound region is used as the importance level of each tissue wound region, and each tissue wound region is labeled according to the importance level.

9. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 8, characterized in that, The process of marking each tissue wound area according to its importance includes: Wound areas with an importance level greater than or equal to a preset importance threshold are labeled as high-importance wounds, while wound areas with an importance level less than the preset importance threshold are labeled as low-importance wounds.

10. A three-dimensional reconstruction system for laparoscopic ovarian tissue wound analysis according to claim 1, characterized in that, The method for obtaining the tissue wound area in each video frame image is as follows: For any video frame image, the gradient value of each pixel in the video frame image is obtained, and the optimal gradient segmentation threshold is obtained using the Otsu threshold segmentation algorithm. Pixels whose gradient values ​​are greater than the optimal gradient segmentation threshold are taken as edge points. Connectivity analysis is performed on all edge points to obtain the tissue wound area in the video frame image.

Images