Automatic fetal image parameter extraction system for spinal cord neural tube malformation screening

Abstract

The invention discloses a fetal image parameter automatic extraction system for spinal cord neural tube malformation screening, and relates to the technical field of medical image processing, and the system comprises a fetal spine collection module which is used for collecting fetal spine structure data, and carrying out the preprocessing of the collected data; the fracture structure identification module is used for constructing a fracture boundary point set, carrying out quantitative evaluation on the complexity of a fracture geometric structure and judging the complexity level of fracture boundary points; the fissure position correction module is used for quantifying the deviation degree of the fissure scale relative to the spinal canal size and executing position correction operation in combination with the complex grade; and the structure anomaly analysis module is used for performing quantitative evaluation on the sudden change intensity of the ridge crack geometric structure and generating a ridge crack basic parameter set and a ridge crack image structured analysis data set. The problems that in existing fetal spine image processing, fracture boundaries cannot be accurately extracted, so that fracture structure judgment is unstable, and spinal cord neural tube malformation screening lacks a reliable quantization parameter basis are solved.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to an automatic fetal image parameter extraction system for screening for spinal cord neural tube defects. Background Technology

[0002] With the rapid development of medical imaging technology, image processing technology, feature extraction algorithms, and intelligent analysis methods, intrauterine image analysis of the fetal central nervous system is gradually evolving from traditional manual observation to quantitative, structured, and automated methods. In recent years, research on automatic identification, three-dimensional reconstruction, and structural measurement of fetal MRI, ultrasound, and CT images has been increasing. Related technologies have enabled automated processing and parameter inference of complex anatomical structures such as the fetal brain and spine. Existing research has seen the gradual maturation of various detection and reconstruction technologies based on fetal brain images or brain region structural features. These technologies involve image preprocessing, feature extraction, region segmentation, and three-dimensional modeling, providing an important foundation for the automated assessment of fetal structural abnormalities.

[0003] For example, the invention with announcement number CN117611542B relates to a method and system for detecting fetal intrauterine brain images, belonging to the field of digital image processing technology. This invention discloses a method and system for detecting fetal intrauterine brain images. The method includes: acquiring fetal brain images; removing noise interference from the fetal brain images and enhancing boundaries to obtain sliced ​​medical images; establishing a grayscale image data analysis layer; setting pixel marker vectors; assigning intensity values ​​to pixels in the sliced ​​medical images; and extracting feature vectors from the sliced ​​medical images; setting seed pixels; traversing the grayscale image data analysis layer through the pixel marker vectors to obtain a two-dimensional segmented image; pre-setting a three-dimensional scene; and superimposing and matching the acquired two-dimensional segmented images onto the same coordinate system based on the three-dimensional scene to obtain a three-dimensional visualization model; and calculating the intrauterine brain volume of the fetus based on the three-dimensional visualization model.

[0004] For example, the invention disclosed in CN116468655B, a brain development atlas and image processing system based on fetal magnetic resonance imaging, belongs to the field of medical imaging technology. Specifically, it includes: an image acquisition module for periodically acquiring and preprocessing MRI image data; an image segmentation module for segmenting fetal brain images from MRI, obtaining brain parameters, and identifying brain sub-regions; a three-dimensional reconstruction module for detecting whether the fetal brain image is qualified, using a slice registration algorithm to perform three-dimensional reconstruction of the fetal brain image, segmenting and extracting the three-dimensional brain tissue structure, reconstructing the cortex from the segmentation results, and generating a three-dimensional structure of the fetal brain tissue; and an atlas generation module for acquiring cross-sectional images of the three-dimensional structure of the fetal brain tissue from a preset angle, arranging the cross-sectional images according to time sequence, and generating a fetal brain development atlas.

[0005] While existing technologies can segment, reconstruct, and quantify fetal cranial structures, image analysis of the fetal spine and spinal canal regions still largely relies on manual interpretation. Methods for extracting fissure boundaries, ossified structures, and spinal canal geometric parameters generally suffer from reliance on experience, difficulty in quantification, and insufficient repeatability. Particularly in spinal canal malformation screening scenarios, factors such as fetal posture changes, differences in imaging sections, and weak ossification signals can easily lead to edge structure breaks, discontinuous fissure widths, and posterior margin morphological shifts in the spinal canal. This makes it difficult for existing methods to automatically and accurately acquire fissure boundary points, bony structural parameters, and fissure geometric change characteristics, and also hinders the stable quantification of fissure complexity, scale deviation, and structural abrupt change trends.

[0006] Therefore, in response to the above problems, there is an urgent need for an automatic fetal image parameter extraction system for screening for spinal cord and neural tube defects. Summary of the Invention

[0007] To address the problems of inaccurate extraction of cleft boundaries in existing fetal spinal imaging processing, leading to unstable cleft structure determination and a lack of reliable quantitative parameters for spinal cord neural tube defect screening, this invention provides an automatic fetal image parameter extraction system for spinal cord neural tube defect screening. The technical solution is as follows:

[0008] An automated fetal image parameter extraction system for spinal cord and neural tube defect screening is provided. This system includes: a fetal spine acquisition module for acquiring fetal spinal structure data and sequentially performing spatial alignment, noise suppression, anomaly removal, missing data completion, and scale normalization on the acquired data to generate preprocessed fetal spinal structure data; a cleft structure identification module for constructing a cleft boundary point set based on the preprocessed fetal spinal structure data and quantitatively evaluating the complexity of the cleft geometry, determining the complexity level of the cleft boundary points based on the evaluation results; a cleft location correction module for quantifying the deviation of the cleft scale from the spinal canal size based on the fetal spinal structure data and performing location correction operations on the cleft boundary points in conjunction with the complexity level; and a structural anomaly analysis module for extracting the location-corrected cleft boundary points, identifying three types of mutation points and aggregating them to form key structural points, quantitatively evaluating the mutation intensity of the spina bifida geometry based on the number of the three types of mutation points and key structural points, and generating a basic parameter set for spina bifida and a fetal spina bifida image analysis dataset.

[0009] Further, fetal spinal structure data is collected, and spatial alignment, noise suppression, anomaly removal, missing data completion, and scale normalization are sequentially performed on the collected fetal spinal structure data to generate preprocessed fetal spinal structure data. The specific steps are as follows: Fetal spinal structure data is collected, including the distance between the ossification centers of the vertebral laminae, the distance between the pedicles, the sagittal diameter of the spinal canal, the transverse diameter of the spinal canal, the length of the posterior margin defect of the spinal canal, the width of the spinal cleft, the area of ​​the spinal cleft, the distance from the ossification center of the vertebral laminae to the edge of the cleft, and the vertebral laminar opening angle; the collected fetal spinal structure data is then processed... The obtained fetal spinal structure data were used to select an imaging section containing the long axis of the fetal spine as a reference imaging section. A rigid registration method based on maximizing gray-level mutual information was used to align all imaging sections to the reference imaging section. A density-based spatial clustering algorithm was used to identify local abnormal structural points and remove abnormal data caused by fetal micro-movements and imaging artifacts. An interpolation algorithm based on the inverse distance weighting method was used to complete local missing data caused by imaging occlusion. The fetal spinal structure data were scaled using a min-max normalization algorithm to unify the numerical scale and eliminate dimensional differences.

[0010] Further, the specific steps for constructing the gap boundary point set based on the preprocessed fetal spinal structure data are as follows: Based on the preprocessed fetal spinal structure data, gradient operation is performed in the posterior edge image region of the spinal canal to obtain a gradient magnitude map, and pixels with gradient magnitudes greater than the magnitude threshold are selected to construct an initial edge point set; eight-neighbor connected component analysis is performed on the initial edge point set to automatically aggregate spatially adjacent pixels into continuous edge segments. Isolated points and invalid pixels caused by imaging artifacts are removed from the aggregation results to form effective edge segments; least squares polynomial fitting operation is called on each effective edge segment to generate a fitting curve for the trend of the posterior edge of the spinal canal; the actual edge points are compared with the spinal canal boundary boundary image region. The fitted curve of the trailing edge of the spinal canal is compared point by point, and the offset distance between the two is calculated. Pixels with offset distances exceeding the deviation threshold are identified and marked as suspected defects. Using the suspected defects as seed points, a region growing algorithm is called within the image region of the trailing edge of the spinal canal. Adjacent pixels with gray-level differences from the seed points less than the gray-level threshold and whose edge direction is consistent with the surrounding crack morphology are gradually incorporated into the growing region. When no new pixels satisfying the gray-level threshold constraint and edge continuity constraint appear on the growing boundary, the region growing is terminated, generating a set of crack region pixels. Boundary tracking is performed on the crack region pixel set, and a contour tracking algorithm is called to track pixel by pixel along the outer edge of the region, outputting a set of crack boundary points.

[0011] Further, the specific steps for quantitatively assessing the complexity of the spina bifida geometry are as follows: Based on the fetal spinal structure data corresponding to the boundary points of the spina bifida, the maximum width and minimum width of the spina bifida are extracted; the fourth power of the length of the posterior margin defect of the spinal canal is divided by the square of the area of ​​the spina bifida to obtain the defect area ratio term; the square root of the ratio of the maximum width to the minimum width of the spina bifida is multiplied by the defect area ratio term, and the product is incremented by one and the natural logarithm is taken to obtain the geometric amplitude term; the distance from the ossification center of the lamina to the edge of the spina bifida is divided by the sum of the length of the posterior margin defect of the spinal canal and the maximum width of the spina bifida, and the ratio is incremented by one to obtain the structural proportion term; the geometric amplitude term is multiplied by the structural proportion term, and the product is divided by the sum of the absolute value of the cosine of the lamina opening angle and a constant one to obtain the spina bifida complexity assessment value.

[0012] Furthermore, the specific steps for determining the complexity level of fracture boundary points based on the evaluation results are as follows: Real-time comparison of the ridge fracture complexity evaluation values. and multi-level complex thresholds and Determine the complexity level of the fracture boundary points: when ≤ When, the index sequence of the fracture boundary points is directly written into the first-level field; when < < When, the index sequence of the fracture boundary points is written into the second-level field; when ≥ At that time, the index sequence of the fracture boundary points is written into the third-level field.

[0013] Further, the specific steps for quantifying the deviation of the spinal fissure scale from the spinal canal size based on fetal spinal structure data are as follows: Extract the corresponding fetal spinal structure data; square the ratio of the posterior margin defect length of the spinal canal to the sagittal diameter of the spinal canal, add it to the square of the ratio of the maximum width of the spinal fissure to the transverse diameter of the spinal canal, and add one to the sum to obtain the geometric ratio term; divide the area of ​​the spinal fissure by the product of the sagittal diameter and the transverse diameter of the spinal canal, and multiply the resulting ratio by the square root of the ratio of the maximum width to the minimum width of the spinal fissure to obtain the area-width coupling term; add the geometric ratio term and the area-width coupling term, and take the natural logarithm to obtain the scale coupling term; subtract one from the ratio of the distance between the ossification centers of the vertebral laminae to the distance between the pedicles, take the absolute value, and add one to obtain the structural deviation term; multiply the scale coupling term and the structural deviation term to obtain the spinal fissure scale deviation assessment value.

[0014] Furthermore, the specific steps for performing position correction on the fissure boundary points in conjunction with the complexity level are as follows: The complexity level field of the fissure boundary points is read. Combining the spinal fissure scale deviation assessment value and the complexity level field, position correction is performed on the fissure boundary point sequence: When the complexity level field is Level 1, no position correction is performed; when the complexity level field is Level 2, a proportional consistency adjustment program is invoked to calculate the adjustable offset of the boundary points based on the proportional relationship between the spinal fissure scale deviation assessment value and the transverse diameter of the spinal canal. For fissure boundary points where the adjustable offset exceeds the proportional deviation threshold, position correction is performed; when the complexity level field is Level 3, a progressive proportional adjustment program is invoked to perform multiple rounds of position correction on the fissure boundary point sequence using the spinal fissure scale deviation assessment value as the adjustment control value. In each round of correction, the offset of the boundary points is calculated based on the proportional relationship between the spinal fissure scale deviation assessment value and the sagittal diameter, transverse diameter, width, and area of ​​the spinal fissure. For fissure boundary points where the offset exceeds the proportional deviation threshold, progressive position correction is performed according to the magnitude of the spinal fissure scale deviation assessment value. The area of ​​the spinal fissure is updated after each round of correction.

[0015] Further, the specific steps for extracting the position-corrected fracture boundary points, identifying three types of abrupt change points, and aggregating them to form key structural points are as follows: Extract all position-corrected fracture boundary points and perform equal-step traversal, calculate the rate of change of direction between adjacent fracture boundary points, and generate a direction change sequence; perform threshold filtering on the direction change sequence to identify abrupt change locations, and write the corresponding fracture boundary points into the morphological inflection point set; extract the corresponding spinal fissure width and area, construct a width interpolation model based on a one-dimensional cubic spline interpolation algorithm, and generate a fracture width distribution curve; perform monotonicity detection on the width distribution curve, calculate the adjacent width difference according to the interpolation sampling point order during the detection process, identify the locations where the width difference sign is reversed and the absolute value of the difference is greater than the width abrupt change threshold as locations of discontinuous width changes, and write the corresponding fracture boundary points into the width abrupt change points. The system constructs a proportional coherence matrix based on the length of the posterior margin defect of the spinal canal and the area of ​​the spinal fissure. During construction, discrete segments along the length of the fissure are used as indices. For each segment, a normalized proportional value is calculated between the length of the posterior margin defect of the spinal canal and the area of ​​the spinal fissure. This normalized proportional value is then filled into the corresponding element of the proportional coherence matrix. A coherence check is performed on the proportional coherence matrix. During the check, the difference between adjacent normalized proportional values ​​is calculated along the length of the fissure. Positions where the absolute value of the difference exceeds a coherence threshold are identified as abrupt changes in the proportional structure, and the corresponding fissure boundary points are written into the defect coherence abrupt change point set. Cross-comparison is performed on morphological inflection points, width abrupt changes, and defect coherence abrupt changes to identify fissure boundary points that simultaneously satisfy two or more labels, marking them as key structural points. The labels include: morphological inflection points, width abrupt changes, and defect coherence abrupt changes.

[0016] Furthermore, the specific steps for quantitatively evaluating the mutation intensity of the spinal fissure geometry based on the number of three types of mutation points and key structural points are as follows: Count the number of morphological inflection points, the number of width mutation points, the number of defect continuity mutation points, the number of key structural points, and the total number of fissure boundary points. Add the number of morphological inflection points, the number of width mutation points, and the number of defect continuity mutation points, and divide by the total number of fissure boundary points to obtain the mutation ratio. Add one to the number of key structural points and divide by the sum of the number of morphological inflection points, the number of width mutation points, and the number of defect continuity mutation points plus one to obtain the key point proportion. Divide the length of the posterior margin defect of the spinal canal by the sum of the maximum and minimum widths of the spinal fissure, add one to the resulting ratio, and take the natural logarithm to obtain the defect width coupling term. Multiply the mutation ratio term, the key point proportion term, and the defect width coupling term sequentially to obtain the spinal fissure structure mutation evaluation value.

[0017] Further, the specific steps for generating the basic parameter set and fetal spina bifida image analysis dataset are as follows: Real-time comparison of the spina bifida structural mutation assessment value and mutation threshold to determine the data organization method: When the spina bifida structural mutation assessment value is less than or equal to the mutation threshold, a basic parameter set for spina bifida is generated, including the maximum width of the spina bifida fissure, the minimum width of the spina bifida fissure, the area of ​​the spina bifida fissure, the length of the posterior margin defect of the spinal canal, and the spina bifida structural mutation assessment value; When the spina bifida structural mutation assessment value is greater than the mutation threshold, based on the basic parameter set, and combined with the fissure boundary point sequence, morphological inflection point set, width mutation point set, and defect continuity mutation point set, a fetal spina bifida image analysis dataset is generated; The basic parameter set and fetal spina bifida image analysis dataset are written into the system output buffer and packaged into a spina bifida image parameter result data packet, which is then sent to the doctor's display terminal.

[0018] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: (1) An initial set of edge points is constructed through gradient operation. Then, the trend of the posterior edge of the spinal canal is fitted by combining eight-neighbor connected domain analysis and least squares polynomial. The fitted curve is used as a reference to automatically identify suspected defect points that deviate beyond the threshold. Using these defect points as seeds, the pixel set of the fissure region and the point set of the fissure boundary are automatically generated by the region growing and contour tracking algorithm. This method eliminates the need for manual delineation of the fissure region frame by frame. It can stably delineate the boundary of the spinal fissure in images containing ossification artifacts and local fractures, significantly reducing the influence of human experience on the consistency of fissure identification results.

[0019] (2) The complexity of the spinal fissure geometry is quantified by the spinal fissure complexity assessment value, and the assessment results are combined with the complexity level field to design a graded position correction strategy that includes no correction, single-round proportional consistency correction, and multi-round progressive correction. This geometric quantification and graded correction chain can avoid over-adjustment for fissures with simple structures and implement multi-round fine correction for fissures with complex structures and abnormal proportions, thereby improving the measurement accuracy and repeatability of key parameters such as the length of the posterior margin defect of the spinal canal, the width of the fissure, and the area of ​​the fissure.

[0020] (3) In the structural anomaly analysis module, starting from the fracture boundary points after position correction, morphological inflection points, width abrupt change points, and defect continuity abrupt change points are automatically identified. Key structural points are formed through cross-comparison and aggregation. Furthermore, based on the three types of abrupt change points and the number of key structural points, a ridge fracture structural abrupt change assessment value is constructed. This hierarchical structural analysis framework not only provides global abrupt change intensity quantification results, but also retains the specific coordinates and category labels of key location points, elevating the ridge fracture morphology from a simple description of width, narrowness, length, and shortness to a structured information expression containing local abrupt change features, which facilitates subsequent multi-center comparison and objective evaluation.

[0021] (4) Through the coordinated operation of four modules—fetal spine acquisition, cleft structure identification, cleft location correction, and structural abnormality analysis—the system achieves one-time automatic extraction and unified organization of multidimensional parameters such as the length of the posterior margin defect of the spinal canal, the maximum and minimum width of the cleft, the cleft area, geometric evaluation values, and structural mutation characteristics. This system reduces the workload of manual measurement and subjective recording, significantly improves the efficiency and standardization of image parameter acquisition during the screening of spinal cord neural tube defects, and provides a directly accessible structured data foundation for subsequent intelligent analysis and clinical image interpretation. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0023] Figure 1 This is a structural diagram of the automatic fetal image parameter extraction system for screening spinal cord and neural tube defects provided in this embodiment of the invention; Figure 2 This is a flowchart of a surgical field space reconstruction system based on big data analysis provided in an embodiment of the present invention; Figure 3 It is a diagram for determining the complexity level of fracture boundary points based on the ridge fracture complexity assessment value. Detailed Implementation

[0024] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0025] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0026] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0027] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0028] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0029] This invention provides an automatic fetal image parameter extraction system for screening for spinal cord and neural tube defects, such as... Figure 1 The diagram shows the structure of an automatic fetal image parameter extraction system for spinal cord and neural tube defect screening. The system's processing flow includes the following steps: a fetal spine acquisition module, used to acquire fetal spinal structure data and sequentially perform spatial alignment, noise suppression, anomaly removal, missing data completion, and scale normalization on the acquired data to generate preprocessed fetal spinal structure data; a cleft structure identification module, used to construct a cleft boundary point set based on the preprocessed fetal spinal structure data and quantitatively evaluate the complexity of the cleft geometry, determining the complexity level of the cleft boundary points based on the evaluation results; a cleft position correction module, used to quantify the deviation of the cleft scale from the spinal canal size based on the fetal spinal structure data and perform position correction operations on the cleft boundary points in conjunction with the complexity level; and a structural anomaly analysis module, used to extract the position-corrected cleft boundary points, identify three types of mutation points and aggregate them to form key structural points, quantitatively evaluate the mutation intensity of the spina bifida geometry based on the number of the three types of mutation points and key structural points, and generate a basic parameter set for spina bifida and a fetal spina bifida image analysis dataset. The specific operation flow is as follows: Figure 2 As shown.

[0030] Optionally, fetal spinal structure data is collected, and spatial alignment, noise suppression, anomaly removal, missing data completion, and scale normalization are sequentially performed on the collected fetal spinal structure data to generate preprocessed fetal spinal structure data. The specific steps are as follows: Fetal spinal structure data is collected, including the distance between ossification centers of the vertebral laminae, the distance between pedicles, the sagittal diameter of the spinal canal, the transverse diameter of the spinal canal, the length of the posterior margin defect of the spinal canal, the width of the spinal cleft, the area of ​​the spinal cleft, the distance from the ossification center of the vertebral laminae to the edge of the cleft, and the vertebral laminar opening angle. The distance between ossification centers of the vertebral laminae is obtained by extracting the spatial positions of the left and right ossification centers of the same vertebral body from consecutive slices and calculating the straight-line distance between adjacent ossification centers of the vertebral laminae. The distance between pedicles is obtained by... The spinal canal diameter was obtained by extracting the centroid positions of the left and right pedicle ossification areas at the same level and calculating the distance between them in the transverse direction. The sagittal diameter was obtained by selecting the posterior margin of the vertebral body and the posterior margin of the spinal canal on the central sagittal plane of the spinal canal and measuring the distance between them in the anterior-posterior direction. The transverse diameter was obtained by determining the left and right bony boundary points of the spinal canal on the transverse section and calculating the transverse distance between the two bony boundary points. The length of the posterior margin defect of the spinal canal was obtained by searching for continuous bony signals along the posterior margin of the spinal canal image area, identifying the interrupted bony signal segment, and measuring the length of the interrupted segment along the posterior margin of the spinal canal. The width of the spinal fissure was obtained by measuring the width of the defect along the normal direction of the posterior margin of the spinal canal at the level where the fissure image is most prominent. The direction of measurement is used to measure the local distance between the bony boundaries on both sides of the fissure, and representative locations within the fissure area are selected for measurement. The area of ​​the spinal fissure is obtained by outlining the closed contour of the fissure on a slice containing the fissure, counting the number of effective pixels within the closed area and multiplying it by the physical area corresponding to a single pixel. The distance from the ossification center of the vertebral lamina to the edge of the fissure is obtained by measuring the length of the line connecting the ossification center of the vertebral lamina to the nearest edge of the fissure on the same plane. The lamina opening angle is obtained by connecting the bony outer edge feature points of the left and right vertebral lamina with the midpoint of the posterior edge of the vertebral body as the vertex and calculating the angle between the two connecting lines. During the acquisition process, the imaging parameters are uniformly set so that the distance between the ossification centers of the vertebral lamina and the distance between the bony outer edges of the vertebral lamina obtained at different examination times for the same fetus are consistent. The interpedicle spacing, sagittal diameter of the spinal canal, transverse diameter of the spinal canal, length of the posterior margin defect of the spinal canal, width of the spinal fissure, area of ​​the spinal fissure, distance from the ossification center of the lamina to the edge of the fissure, and the lamina opening angle are all under a unified spatial and grayscale scale. For the acquired fetal spinal structure data, the imaging section containing the long axis of the fetal spine is selected as the reference imaging section. A rigid registration method based on maximizing grayscale mutual information is used to align all imaging sections to the reference imaging section, so that the interpedicle spacing, sagittal diameter of the spinal canal, transverse diameter of the spinal canal, length of the posterior margin defect of the spinal canal, width of the spinal fissure, area of ​​the spinal fissure, distance from the ossification center of the lamina to the edge of the fissure, and lamina opening angle corresponding to the same anatomical position are accurately coincident in space.After spatial alignment, a fusion strategy of bilateral filtering and Gaussian smoothing was employed to suppress noise and preserve edges in the original image data. Bilateral filtering smoothed noise in the fetal spinal structure data by jointly considering spatial neighborhood distance and gray-level differences. Gaussian smoothing suppressed random noise by applying a weighted average to local regions, while maintaining gray-level transitions in edge regions containing the length of the posterior margin defect of the spinal canal and the width of the spinal fissure, thus preserving the true structural contours of the inter-laminar ossification distance, pedicle distance, and sagittal and transverse diameters of the spinal canal. Subsequently, a density-based spatial clustering algorithm was used to identify local abnormal structural points. This algorithm statistically analyzed the point density within the pixel neighborhood corresponding to the length of the posterior margin defect of the spinal canal, the width of the spinal fissure, and the area of ​​the spinal fissure, marking isolated points with a density significantly lower than the neighborhood mean as abnormal points. This eliminated abnormal data caused by fetal micromovement and imaging artifacts, preventing abnormal data from affecting the inter-laminar ossification distance and pedicle distance. Subsequent quantification of interpedicle spacing and lamina opening angle; after anomaly removal, an interpolation algorithm based on inverse distance weighting is used to complete locally missing data caused by imaging occlusion. Effective data points around the missing locations are weighted according to the inverse of the sagittal diameter and transverse diameter of the spinal canal and the spatial distance. The length of the posterior margin defect of the spinal canal, the width of the spinal fissure, the area of ​​the spinal fissure, and the distance from the ossification center of the lamina to the edge of the fissure are interpolated and estimated to ensure a continuous spatial distribution of fetal spinal structure data. Finally, a minimum-maximum normalization algorithm is used to normalize the fetal spinal structure data, mapping the interpedicle spacing, interpedicle spacing, sagittal diameter of the spinal canal, transverse diameter of the spinal canal, length of the posterior margin defect of the spinal canal, width of the spinal fissure, area of ​​the spinal fissure, distance from the ossification center of the lamina to the edge of the fissure, and lamina opening angle to a unified numerical range according to their respective minimum and maximum values. This unifies the numerical scale and eliminates dimensional differences, resulting in preprocessed fetal spinal structure data.

[0031] In this implementation plan, through this complete set of acquisition and preprocessing procedures, the distance between the ossification centers of the vertebral laminae, the distance between the pedicles, the sagittal diameter of the spinal canal, the transverse diameter of the spinal canal, the length of the posterior margin defect of the spinal canal, the width of the spinal fissure, the area of ​​the spinal fissure, the distance from the ossification center of the vertebral laminae to the edge of the fissure, and the vertebral laminae opening angle have been unified to the same spatial scale and numerical range before entering subsequent processing. The discrete gaps caused by noise interference, artifacts, and occlusion are effectively corrected. The fetal spinal structure data are transformed from the original scattered measurement results that relied on human experience into a coherent, regular, and structured input that is easy to process in batches. This fundamentally reduces the sensitivity of subsequent algorithms to differences in imaging conditions and operators, and enhances the adaptability of the fissure identification and geometric quantification process among multiple batches of examinations, multiple devices, and multiple fetal samples. This provides a reliable data foundation for the stable implementation of the entire automatic extraction process in engineering.

[0032] Optionally, the specific steps for constructing the gap boundary point set based on the preprocessed fetal spinal structure data are as follows: Based on the preprocessed fetal spinal structure data, gradient operation is performed in the posterior margin image region of the spinal canal to obtain a gradient amplitude map, and pixels with gradient amplitudes greater than the amplitude threshold are selected to construct an initial edge point set; before performing gradient operation, the posterior margin image region of the spinal canal is precisely demarcated according to the preprocessed image partitions, so that the gradient calculation only acts on the imaging range corresponding to the posterior margin of the spinal canal, thereby avoiding the gradient response of irrelevant tissues from interfering with the selection results; eight-neighbor connected component analysis is performed on the initial edge point set to spatially... Adjacent pixels are automatically aggregated into continuous edge segments. Isolated points and invalid pixels caused by imaging artifacts are removed from the aggregation results, forming valid edge segments. During connected component analysis, connectivity is determined by combining pixel spacing and scanning direction consistency, ensuring the continuous direction of the edge segments aligns with the posterior edge morphology of the spinal canal. Least squares polynomial fitting is applied to each valid edge segment to generate a fitting curve for the posterior edge trend of the spinal canal. In the least squares fitting, the edge segments are input sequentially according to their point order, ensuring the fitting curve maintains consistency with the actual posterior edge trend of the spinal canal during numerical convergence. The actual edge points are then compared with the spinal canal... The fitted curves of the posterior edge trend are compared point by point to calculate the offset distance between them. Pixels with offset distances exceeding the deviation threshold are identified and marked as suspected defects. When comparing offset distances, the fitted curves are interpolated according to the image row order structure to ensure that the offset calculation strictly matches the actual imaging resolution, avoiding deviations caused by inconsistent pixel scales. Using suspected defects as seed points, a region growing algorithm is applied within the posterior edge image region of the spinal canal. Adjacent pixels with grayscale differences less than the grayscale threshold and whose edge direction is consistent with the surrounding crack morphology are gradually incorporated into the growing region. When no more defects appear on the growing boundary... Region growth is terminated when a new pixel satisfies the grayscale threshold constraint and edge continuity constraint, generating a set of pixels for the fissure region. During the region growth process, the growth direction is kept consistent with the natural extension direction of the posterior edge of the spinal canal to ensure that the growth range strictly covers the fissure region and avoids erroneous expansion of the surrounding soft tissue. Boundary tracking is performed on the pixel set of the fissure region by calling the contour tracking algorithm to track the region pixel by pixel and output the fissure boundary point set. During the pixel-by-pixel traversal of the contour tracking, a starting point based on the structural characteristics of the fissure region is set to ensure that the boundary point set remains closed and fully presents the outer edge morphology of the fissure.

[0033] In this implementation scheme, by introducing a demarcation strategy targeting the posterior edge of the spinal canal imaging region, a gradient filtering enhancement strategy, an edge segment coherence constraint strategy, a polynomial fitting stabilization strategy, an offset distance precise quantification strategy, a region growth direction constraint strategy, and a boundary tracking closure maintenance mechanism during the fracture boundary point construction process, the extraction process of fracture boundary points can maintain stable contour recognition capabilities even under noise interference, artifact interference, and tissue boundary blurring. This process achieves complete extraction of the outer edge of the fracture region, providing a reliable input basis for the fracture structure in subsequent geometric quantization, scale deviation analysis, and structural abrupt change identification stages, thereby significantly improving the extraction accuracy and reliability of fracture-related parameters.

[0034] Optionally, further, the specific steps for quantitatively assessing the complexity of the spinal cleft geometry are as follows: Based on the fetal spinal structure data corresponding to the cleft boundary points, the maximum and minimum widths of the spinal cleft are extracted. During the extraction process, row-level scanning is performed on adjacent cleft boundary points to ensure that the width values ​​always originate from the anatomical segments corresponding to the same imaging cross-section. The fourth power of the posterior margin defect length of the spinal canal is divided by the square of the spinal cleft area to obtain the defect area ratio term, which reflects the extent of expansion of the defect length relative to the cleft area. During the calculation of the defect area ratio term, a uniform numerical scale is used for the defect length and the cleft area to avoid proportional distortion caused by dimensional differences. The square root of the ratio of the maximum to minimum width of the spinal cleft is multiplied by the defect area ratio term, and the product is incremented by one and the natural logarithm is taken to obtain the geometric amplitude term, which characterizes the geometric increase of the cleft under the combined effects of width variation and defect scale. For the long amplitude, a threshold protection is applied to the product result before taking the natural logarithm to suppress numerical amplification caused by abnormally small values. The distance from the ossification center of the lamina to the edge of the fissure is divided by the sum of the length of the defect at the posterior edge of the spinal canal and the maximum width of the spinal fissure. The ratio is then incremented by one to obtain the structural proportion term, which describes the relative offset between the fissure location and the surrounding bony structures. When obtaining the structural proportion term, the measurement units of all input quantities are kept completely consistent to ensure the numerical stability of the structural proportion term. The geometric amplitude term is multiplied by the structural proportion term, and the product is divided by the sum of the absolute value of the cosine of the lamina opening angle and a constant one to obtain the spinal fissure complexity assessment value, which is used to quantify the comprehensive complexity of the spinal fissure geometry under the combined influence of defect size, width difference, bony structure offset, and lamina opening trend. Before finally obtaining the spinal fissure complexity assessment value, a lower limit is applied to the cosine term to avoid abnormal amplification caused by the cosine value approaching zero when the lamina opening angle is close to 90 degrees.

[0035] The specific formula for calculating the complexity assessment value of the ridge fissure is as follows: ; In the formula, This represents the complexity assessment value of the ridge fissure. Indicates the length of the defect at the posterior margin of the spinal canal. Indicates the area of ​​the spinal fissure. Indicates the maximum width of the spinal hiatus. This indicates the minimum width of the spina bifida cleft. This indicates the distance from the ossification center of the lamina to the edge of the fissure. Indicates the lamina opening angle.

[0036] In this embodiment, Table 1 is a data table of spina bifida complexity assessment values, listing all parameter data used to quantify the geometric complexity of the five bifida boundary points in this assessment process. Specifically: Boundary point D1: The length of the posterior edge defect of the spinal canal is 1.0, the area of ​​the spina bifida is 3.0, the maximum width of the spina bifida is 0.80, the minimum width of the spina bifida is 0.50, the distance from the ossification center of the lamina to the edge of the bifida is 0.30, the lamina opening angle is 0.50, and the spina bifida complexity assessment value is 0.081. Boundary point D2: The length of the posterior edge defect of the spinal canal is 1.4, the area of ​​the spina bifida is 3.5, the maximum width of the spina bifida is 1.00, the minimum width of the spina bifida is 0.60, the distance from the ossification center of the lamina to the edge of the bifida is 0.40, the lamina opening angle is 0.60, and the spina bifida complexity assessment value is approximately 0.212. Spinal fissure boundary point D3: The length of the posterior margin defect of the spinal canal is 1.8 mm, the area of ​​the spinal fissure is 4.0 mm, the maximum width of the spinal fissure is 1.20 mm, the minimum width of the spinal fissure is 0.70 mm, the distance from the ossification center of the lamina to the edge of the fissure is 0.50 mm, the lamina opening angle is 0.70°, and the spinal fissure complexity assessment value is approximately 0.393. Spinal fissure boundary point D4: The length of the posterior margin defect of the spinal canal is 2.2 mm, the area of ​​the spinal fissure is 4.5 mm, the maximum width of the spinal fissure is 1.40 mm, the minimum width of the spinal fissure is 0.80 mm, the distance from the ossification center of the lamina to the edge of the fissure is 0.60 mm, and the lamina opening angle is 0.80°. The spinal fissure complexity assessment value calculated according to the spinal fissure complexity assessment formula is approximately 0.604. The length of the defect at the posterior margin of the spinal canal is 2.6, the area of ​​the spinal fissure is 5.0, the maximum width of the spinal fissure is 1.60, the minimum width of the spinal fissure is 0.90, the distance from the ossification center of the lamina to the edge of the fissure is 0.70, the lamina opening angle is 0.90, and the complexity assessment value of the spinal fissure is approximately 0.830.

[0037] Table 1. Data on the complexity assessment values ​​of the spina bifida. like Figure 3As shown in the figure, the complexity assessment values ​​of five fracture boundary points are displayed, along with the corresponding complexity level classification results. The bar chart uses different colors to indicate the complexity level of the fracture boundary points: blue bars represent fracture boundary points written into the first-level field, green bars into the second-level field, and red bars into the third-level field. Gray and black dashed lines respectively mark the first-level and second-level complexity thresholds, used to classify the structural complexity of different fracture boundary points. The figure shows that boundary point D1 is classified as a first-level field, corresponding to a relatively simple fracture structure; boundary points D2 and D3 are classified into the second-level field, indicating an increasing trend in fracture geometry complexity compared to the first level; and boundary points D4 and D5 are classified into the third-level field, corresponding to the most complex fracture morphology. Figure 3 This paper intuitively demonstrates the mechanism for determining the complexity level of fissure boundary points based on the complexity assessment value and multi-level complexity thresholds. Through comprehensive quantitative analysis of the geometric features of different boundary points, the paper realizes the hierarchical identification of the complexity of fissure structures, providing a reliable basis for subsequent fissure location correction and automatic extraction of fissure image parameters.

[0038] In this implementation scheme, by introducing the defect scale amplification relationship, fissure width difference characteristics, and bony structure offset characteristics during the construction of the defect area ratio, geometric amplitude, and structural proportion terms, respectively, the multidimensional structural parameters of the fissure in terms of geometric expansion, width unevenness, and bony connectivity can be continuously integrated in the same evaluation process. This allows parameters from different sources, with different dimensions, and at different structural levels to be represented as correlated quantitative results under a unified scale in the comprehensive calculation. This correlated quantitative result can produce a sensitive and stable response when the fissure morphology is significantly different, the boundary changes are complex, the width unevenness is high, and the bony structure position is significantly discrete, so that the overall deformation trend of the complex fissure structure can be fully expressed in mathematical form. This significantly improves the identification accuracy of complex fissure structures and the repeatability of structural classification decisions, providing a continuous, controllable, and comparable quantitative basis for subsequent structural level determination, and enhancing the ability to accurately reconstruct the geometry of complex spina bifida.

[0039] Optionally, the specific steps for determining the complexity level of the fracture boundary point based on the evaluation results are as follows: real-time comparison of the ridge fracture complexity evaluation value F and the multi-level complexity threshold. and Determine the complexity level of the fracture boundary points, maintain the continuous quantification of the ridge fracture complexity assessment value F during the comparison process, and ensure the first-level complexity threshold. With second-level complex threshold The hierarchical boundaries between them are stable and separable, ensuring consistency and repeatability in hierarchical decisions; when F≤ When writing, the index sequence of the fracture boundary points is directly written into the first-level field, maintaining the index order consistent with the original set of fracture boundary points during the writing process, so that the first-level field can completely record the positions corresponding to the fracture boundary points with the lowest complexity; when <F< When F ≥ 1, the index sequence of the fracture boundary points is written into the second-level field, maintaining the continuity of the index sequence during the writing process so that the second-level field can accurately mark the positions of fracture boundary points in the intermediate complexity range; when F ≥ 1 When writing, the index sequence of the fracture boundary points is written into the third-level field. During the writing process, all fracture boundary points that meet the high complexity condition are recorded, so that the third-level field can be used as a dedicated input for high morphological complexity fracture regions in subsequent processing.

[0040] In this implementation scheme, a stable comparison mechanism between the ridge fracture complexity assessment value and multi-level complexity thresholds is introduced during the complexity level determination process of fracture boundary points. This ensures that continuous quantitative indicators maintain boundary accuracy and consistency in discrete level classification, thereby avoiding classification fluctuations caused by subjective human experience. The classification results of fracture boundary points retain stable and repeatable resolution capabilities under different image qualities, fracture morphologies, and structural scales. This determination process maintains the original continuous characteristics of the boundary point sequence during index writing, enabling each level field to clearly correspond to the fracture morphology change region. This provides clear and traceable classification criteria for subsequent fracture location correction and structural anomaly analysis, thereby improving the refinement and stability of the overall fracture geometry analysis.

[0041] Optionally, the specific steps for quantifying the deviation of the fissure scale from the spinal canal size based on fetal spinal structure data are as follows: Extract the corresponding fetal spinal structure data, maintaining the original quantitative relationship of each structural parameter during extraction so that subsequent ratio calculations can be performed based on input values ​​of a unified scale; Square the ratio of the posterior margin defect length of the spinal canal to the sagittal diameter of the spinal canal, add it to the square of the ratio of the maximum width of the spinal fissure to the transverse diameter of the spinal canal, and add one to the sum to obtain a geometric ratio term, used to quantify the joint scale difference between the fissure length and width relative to the overall size of the spinal canal, so that the longitudinal and transverse expansion of the fissure can be expressed in a unified form; Divide the area of ​​the spinal fissure by the product of the sagittal diameter and the transverse diameter of the spinal canal, and multiply the resulting ratio by the square root of the ratio of the maximum width to the minimum width of the spinal fissure to obtain an area-width coupling term, used to reflect the proportion of the fissure area in the overall structure. It also reflects the differences in fissure width distribution, allowing the trends of area growth and width unevenness to simultaneously affect scale assessment. Adding the geometric ratio term to the area-width coupling term and then taking the natural logarithm yields the scale coupling term, used to compress structural differences of different magnitudes to a suitable numerical range, ensuring the fissure scale changes remain continuous and distinguishable in the calculation results. Subtracting one from the ratio of the distance between ossification centers of the vertebral laminae to the distance between the pedicles, taking the absolute value, and then adding one, yields the structural deviation term, used to reflect the degree of proportional abnormality of bony structures near the fissure location, allowing bony development deviations to form a stable quantitative contribution in scale analysis. Multiplying the scale coupling term and the structural deviation term yields the spinal fissure scale deviation assessment value, used to comprehensively express the degree of deviation of the fissure at multiple structural scales, presenting the overall abnormal structure of the fissure as a single continuous quantitative index, providing a controllable deviation basis for subsequent fissure boundary point location correction.

[0042] The specific formula for calculating the deviation of the spinal fissure scale from the assessment value is as follows: ; In the formula, This indicates that the vertebral fissure scale deviates from the assessed value. Indicates the length of the defect at the posterior margin of the spinal canal. Indicates the sagittal diameter of the vertebral canal. Indicates the transverse diameter of the vertebral canal. Indicates the maximum width of the spinal hiatus. This indicates the minimum width of the spina bifida cleft. Indicates the area of ​​the spinal fissure. Indicates the distance between the centers of ossification of the vertebral laminae. This indicates the distance between the pedicles.

[0043] In this implementation scheme, by sequentially introducing geometric ratio terms, area-width coupling terms, scale coupling terms, and structural deviation terms during the quantification process of fissure scale deviation, the length expansion, width variation, area distribution, and differences in bony structure of the fissure can be jointly influenced in a multidimensional way in the final spinal fissure scale deviation assessment value. This allows structural parameters from different sources to form a continuous expression under a unified quantification framework. This quantification method can maintain a sensitive response even with complex fissure morphology, uneven width distribution, sudden area increases, and abnormal bony structure proportions, ensuring that both local anomalies and overall scale changes of the fissure are accurately reflected in the assessment. By integrating multiple structural features in a multiplicative manner, the spinal fissure scale deviation assessment value can truly reflect the overall degree of deviation of the fissure under the constraints of spinal canal size, providing a directly usable continuous control quantity for subsequent position correction, giving the boundary point position adjustment a clear dimensional basis and directionality, thereby improving the precision and reliability of fissure edge morphology correction.

[0044] Optionally, the specific steps for performing position correction on the fissure boundary points in conjunction with the complexity level are as follows: Read the complexity level field of the fissure boundary points, maintaining the original index order of the boundary points during the reading process, so that subsequent position correction can be performed based on a continuous sequence of boundary points; Perform position correction on the fissure boundary point sequence by combining the spina bifida scale deviation assessment value and the complexity level field, ensuring that the fetal spinal structure data corresponding to each fissure boundary point remains unchanged during the correction process, so that the offset calculation is always based on the actual structural parameters; When the complexity level field is level one, no position correction is performed, maintaining the original morphology of the fissure boundary points, allowing low-complexity fissure regions to participate in the analysis according to their actual structural morphology in subsequent processing; When the complexity level field is level two, call the proportional consistency adjustment program to calculate the adjustable offset of the boundary points based on the proportional relationship between the spina bifida scale deviation assessment value and the transverse diameter of the spinal canal. After the offset calculation, perform a single-round refinement correction operation on the fissure boundary points, extracting continuous edge change trends within the range of adjacent fissure boundary points to create a smooth transition along the boundary direction, and apply a limit on the fissure boundary points when the offset exceeds the proportional deviation threshold. A fixed-amplitude position correction is performed, maintaining the connectivity between boundary points during the correction process to prevent local breaks in the adjusted structure. When the complexity level field is level three, a proportionally progressive adjustment program is invoked. Multiple rounds of position correction are performed on the fissure boundary point sequence using the spinal fissure scale deviation assessment value as the adjustment control. In each round, the offset of the boundary points is calculated based on the proportional relationship between the spinal fissure scale deviation assessment value and the sagittal diameter, transverse diameter, width, and area of ​​the spinal fissure. After the offset calculation, a local refinement correction operation is performed on each fissure boundary point, using the current boundary... The continuous edge change trend is extracted from adjacent points to make the correction offset gradually and smoothly transition within a local range, thereby avoiding edge breakage caused by abrupt offset. For crack boundary points whose offset exceeds the proportional deviation threshold, progressive position correction is performed according to the magnitude of the deviation from the evaluation value of the ridge crack scale. During the progressive correction process, the continuity of the previous correction result is maintained, so that the boundary morphology gradually approaches the real structure in multiple rounds of adjustment. After each round of correction, the ridge crack area is updated. During the area update process, the pixel set of the crack region is simultaneously corrected so that the crack area can accurately reflect the crack structure after refinement correction.

[0045] In this implementation scheme, a graded correction strategy based on complexity level is introduced during the location correction process. This allows for differentiated adjustment of fracture boundary points at varying complexity levels, achieving targeted correction while maintaining structural continuity. The second-level location correction uses a proportional consistency adjustment procedure combined with single-round refinement correction, enabling the correction offset to gradually transition along adjacent fracture boundary points. This avoids local breakage in moderately complex fracture regions after correction, improving the smoothness of the corrected boundary. The third-level location correction executes multiple rounds of location correction through a progressive proportional adjustment procedure. In each round, local refinement adjustments are made based on the proportional relationship between the spinal fissure scale deviation assessment value and multiple spinal geometric parameters. This allows highly complex fracture regions to gradually approach the true structural morphology through multiple progressive corrections, effectively avoiding boundary misalignment caused by large adjustments. Overall, this location correction mechanism achieves flexible and adaptive boundary point adjustment at different complexity levels, achieving a better balance between continuity, smoothness, and realism in the fracture boundary structure, thereby significantly improving the accuracy of subsequent abrupt change point identification and overall structural analysis.

[0046] Optionally, the specific steps for extracting the position-corrected fracture boundary points, identifying three types of abrupt change points, and aggregating them to form key structural points are as follows: Extract all position-corrected fracture boundary points and perform a step-by-step traversal, maintaining a stable index order of the fracture boundary points during the traversal. Calculate the rate of change of direction between adjacent fracture boundary points, expressing the local orientation change of the fracture boundary in numerical form, generating a sequence of direction changes. Perform threshold filtering on the sequence of direction changes, comprehensively judging based on the numerical amplitude and trend of the rate of change of direction, identifying... Identify locations of abrupt changes in direction and add the corresponding fissure boundary points to a set of morphological inflection points, ensuring that areas with significant boundary changes are recorded. Extract the corresponding spinal fissure width and area, ensuring the interpolation input data remains consistent with the fissure geometry. Construct a width interpolation model based on a one-dimensional cubic spline interpolation algorithm, creating a continuous variation sequence of discrete width data along the fissure length, generating a fissure width distribution curve. Perform monotonicity detection on the width distribution curve, calculating adjacent width differences in the order of interpolation sampling points during the detection process. Identify locations where the sign of the width difference is reversed and the absolute value of the difference is greater than the width abrupt change threshold. For locations with discontinuous width changes, the corresponding fissure boundary points are written into a set of width abrupt change points, allowing regions with sudden increases or decreases in width to be independently identified. A proportional consistency matrix is ​​constructed based on the length of the posterior edge defect of the spinal canal and the area of ​​the spinal fissure. During construction, discrete segments along the length direction of the fissure are used as indices. For each segment, a normalized ratio of the posterior edge defect length to the area of ​​the spinal fissure is calculated, ensuring that the defect scale and fissure area form corresponding proportional records on each segment. The normalized ratio values ​​are then filled into the corresponding elements of the proportional consistency matrix. A consistency check is performed on the proportional consistency matrix, calculating adjacent normalized ratios along the fissure length direction during the check. The difference of the example values ​​identifies abrupt changes in the proportional structure at locations where the absolute value of the difference exceeds the coherence threshold. The corresponding crack boundary points are written into the defect coherence abrupt change point set to individually mark abnormal locations in the coupling relationship between defect scale and area. Cross-comparison is performed on morphological inflection points, width abrupt changes, and defect coherence abrupt changes. By analyzing the overlap of the three types of abrupt changes on the crack boundary point index, crack boundary points that simultaneously satisfy two or more labels are identified and marked as key structural points, enabling the aggregation and representation of multiple abrupt change information at key locations. The labels include: morphological inflection points, width abrupt changes, and defect coherence abrupt changes.

[0047] In this implementation scheme, by performing joint processing of orientation change analysis, width change analysis, and proportional coherence analysis on the position-corrected fissure boundary points, the relationship between the width and area of ​​the spinal fissure and the length of the posterior margin defect of the spinal canal is no longer limited to a single parameter level, but is structurally expressed in the form of a set of morphological inflection points, a set of width abrupt change points, and a set of defect coherence abrupt change points. By introducing normalized proportional values ​​into the proportional coherence matrix and combining them with the screening results of key structural points, it is possible to highlight those locations on the fissure boundary that are abnormal in orientation change, width change, and defect proportion simultaneously, improving the responsiveness of spinal fissure geometric abrupt change identification to subtle structural anomalies. This process makes key structural points the focus of subsequent spinal fissure structure quantification and image interpretation, reducing the interference of irrelevant boundary points on the interpretation results, thereby improving the targeting and stability of spinal canal image parameter extraction while ensuring data integrity.

[0048] Optionally, the specific steps for quantitatively evaluating the mutation intensity of the ridge fracture geometry based on the number of three types of mutation points and key structural points are as follows: Count the number of morphological inflection points, width mutation points, defect continuity mutation points, key structural points, and the total number of fracture boundary points. During the statistical process, for each type of mutation point, each item is accumulated based on its spatial distribution position in the fracture boundary point sequence to ensure that the quantity statistics cover all fracture boundary points after position correction. The number of morphological inflection points, width mutation points, and defect continuity mutation points is added together and divided by the total number of fracture boundary points to obtain the mutation ratio, which reflects the overall structural disturbance proportion of the fracture boundary. The number of key structural points is incremented by one and then divided by... The sum of the number of morphological inflection points, the number of width mutation points, and the number of defect continuity mutation points, plus one, yields the key point proportion term, which measures the degree of spatial aggregation of multiple labels among the three types of mutation points. The length of the posterior edge defect of the spinal canal is divided by the sum of the maximum width and the minimum width of the spinal fissure, and the ratio is increased by one and then the natural logarithm is taken to obtain the defect width coupling term, which is used to characterize the matching strength between the defect length and the fissure width scale. The mutation proportion term, the key point proportion term, and the defect width coupling term are multiplied in sequence to obtain the spinal fissure structural mutation assessment value, so that the spinal fissure structural mutation assessment value can comprehensively reflect the combined influence of fissure geometric disturbance, structural aggregation anomaly, and scale offset on the overall mutation intensity.

[0049] The specific formula for calculating the ridge fissure structural mutation assessment value is as follows: ; In the formula, This indicates the assessment value of the abrupt change in the ridge fissure structure. This represents the total number of fracture boundary points. Indicates the number of morphological turning points. Indicates the number of width abrupt change points. Indicates the number of abrupt changes in the continuity of the defect. Indicates the number of critical structural points. Indicates the length of the defect at the posterior margin of the spinal canal. Indicates the maximum width of the spinal hiatus. This indicates the minimum width of the spinal fissure.

[0050] In this implementation scheme, a method for quantitatively evaluating the mutation intensity of spina bifida geometry based on the number of three types of mutation points and key structural points is adopted. This method enables joint modeling of the mutation ratio, key point proportion, and defect width coupling term within the same evaluation process. Consequently, the spina bifida structural mutation evaluation value can simultaneously reflect the combined influence of the distribution of boundary perturbations, the degree of key structural aggregation, and the coupling relationship between defect scale. This joint quantification strategy effectively enhances the sensitivity of mutation intensity discrimination while maintaining the consistency of the physical meaning of variables, ensuring that the evaluation results have a stable response to both local geometric anomalies and overall scale deviations. This provides more accurate mutation intensity input in the subsequent generation stage of the spina bifida baseline parameter set, enabling the spina bifida image structured analysis dataset to truly reflect the trend of morphological changes in the fracture, improving the reliability of fetal spinal cord neural tube imaging structural abnormality identification, and significantly enhancing the overall parameter extraction accuracy.

[0051] Optionally, the specific steps for generating the basic parameter set and fetal spina bifida image analysis dataset are as follows: Real-time comparison of the spina bifida structural mutation assessment value and mutation threshold; using the difference between the assessment value and the threshold as a criterion to determine the data organization method, enabling different data processing flows to be triggered during the data output stage. When the spina bifida structural mutation assessment value is less than or equal to the mutation threshold, the basic parameter processing flow is invoked to generate a basic parameter set for spina bifida, including the maximum width, minimum width, area, posterior margin defect length of the spinal canal, and the assessment value of the structural mutation. During the generation process, a range check is performed on each parameter to ensure that the generated basic parameter set maintains numerical continuity, stability, and comparability. When the assessed value of spina bifida structural mutation exceeds the mutation threshold, a further data aggregation operation is performed based on the basic spina bifida parameter set, combined with the sequence of bifida boundary points, the set of morphological inflection points, the set of width mutation points, and the set of defect continuity mutation points. By integrating and verifying the consistency of these four types of structural information item by item, a fetal spina bifida image analysis dataset is generated. This dataset fully reflects the changes in bifida morphology, width trends, and defect continuity characteristics. After completing the above data organization, the basic spina bifida parameter set and the fetal spina bifida image analysis dataset are written into the output buffer, ensuring the data structure within the buffer meets subsequent reading requirements. The dataset is then packaged into a spina bifida image parameter result data package and sent to the doctor's display terminal, enabling the doctor to directly interpret the images based on the structured data in a unified format.

[0052] In this implementation plan, by comparing the spina bifida structural mutation assessment value with the mutation threshold in real time and triggering a differentiated data organization process accordingly, the output content can automatically adjust the information granularity according to the mutation intensity of the fissure structure. When the mutation level is low, only the basic parameter set of spina bifida is generated, allowing doctors to quickly browse key indicators without redundant information interference. When the mutation level is high, the sequence of fissure boundary points, the set of morphological inflection points, the set of width mutation points, and the set of defect continuity mutation points are further integrated, giving the output content a full descriptive ability at the structural detail level. By writing the two types of data results into the output buffer and packaging them into a unified format spina bifida image parameter result data package, doctors can directly read the data on the display terminal to obtain structured, hierarchical, and quantifiable image analysis results, thereby improving the efficiency of fissure morphology interpretation and enhancing the traceability of parameters and diagnostic auxiliary value in the spina bifida screening process.

[0053] The following points need to be explained: (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.

[0054] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the invention, i.e., these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element or there may be intermediate elements.

[0055] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0056] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An automatic fetal image parameter extraction system for screening for spinal cord neural tube defects, characterized in that, The system includes: The fetal spine acquisition module is used to acquire fetal spine structure data and sequentially perform spatial alignment, noise suppression, anomaly removal, missing data completion and scale normalization on the acquired fetal spine structure data to generate preprocessed fetal spine structure data. The fissure structure identification module is used to construct a set of fissure boundary points based on preprocessed fetal spinal structure data, and to quantitatively evaluate the complexity of the fissure geometry, and to determine the complexity level of the fissure boundary points based on the evaluation results. The gap location correction module is used to quantify the deviation of the gap scale from the spinal canal size based on fetal spinal structure data, and perform position correction operations on the gap boundary points in combination with the complexity level. The structural anomaly analysis module is used to extract the fracture boundary points after position correction, identify three types of mutation points and aggregate them to form key structural points, perform quantitative evaluation of the mutation intensity of the spina bifida geometry based on the number of the three types of mutation points and key structural points, and generate a basic parameter set of spina bifida and a fetal spina bifida image analysis dataset.

2. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 1, characterized in that, The specific steps for collecting fetal spinal structure data and sequentially performing spatial alignment, noise suppression, anomaly removal, missing data completion, and scale normalization on the collected fetal spinal structure data to generate preprocessed fetal spinal structure data are as follows: Fetal spinal structure data were collected, including the distance between the ossification centers of the vertebral laminae, the distance between the pedicles, the sagittal diameter of the spinal canal, the transverse diameter of the spinal canal, the length of the defect at the posterior edge of the spinal canal, the width of the spinal fissure, the area of ​​the spinal fissure, the distance from the ossification center of the vertebral laminae to the edge of the fissure, and the opening angle of the vertebral laminae. For the acquired fetal spinal structure data, an imaging section containing the long axis of the fetal spine was selected as the reference imaging section. A rigid registration method based on maximizing gray-level mutual information was used to align all imaging sections to the reference imaging section. A fusion strategy of bilateral filtering and Gaussian smoothing was used to perform noise suppression and edge preservation processing on the original image data. A density-based spatial clustering algorithm was used to identify local abnormal structural points and remove abnormal data caused by fetal micro-movement and imaging artifacts. An interpolation algorithm based on inverse distance weighting was used to complete local missing data caused by imaging occlusion. The fetal spinal structure data was scaled using a min-max normalization algorithm to unify the numerical scale and eliminate dimensional differences.

3. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 1, characterized in that, The specific steps for constructing the gap boundary point set based on the preprocessed fetal spinal structure data are as follows: Based on the preprocessed fetal spinal structure data, gradient operation is performed in the posterior edge image region of the spinal canal to obtain a gradient amplitude map, and pixels with gradient amplitudes greater than the amplitude threshold are selected to construct an initial edge point set. Eight-neighbor connected component analysis is performed on the initial edge point set to automatically aggregate spatially adjacent pixels into continuous edge segments. Isolated points and invalid pixels caused by imaging artifacts are removed from the aggregation results to form valid edge segments. Least squares polynomial fitting operation is performed on each valid edge segment to generate a fitting curve of the posterior edge of the spinal canal. The actual edge points are compared with the fitting curve of the posterior edge of the spinal canal point by point to calculate the offset distance between them. Pixels with offset distances exceeding the deviation threshold are identified and marked as suspected defects. Using suspected defect points as seed points, a region growing algorithm is called within the posterior edge image region of the spinal canal. Adjacent pixels whose gray-level difference with the seed point is less than the gray-level threshold and whose edge direction is consistent with the surrounding crack morphology are gradually incorporated into the growing region. When no new pixels satisfying the gray-level threshold constraint and edge continuity constraint appear on the growing boundary, the region growing is terminated, generating a crack region pixel set. Boundary tracking is performed on the crack region pixel set, and a contour tracking algorithm is called to track pixel by pixel along the outer edge of the region, outputting the crack boundary point set.

4. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 1, characterized in that, The specific steps for quantitatively assessing the complexity of the crack geometry are as follows: Based on the fetal spinal structure data corresponding to the boundary points of the spina bifida, the maximum and minimum widths of the spina bifida are extracted. The fourth power of the length of the posterior margin defect of the spinal canal is divided by the square of the area of ​​the spina bifida to obtain the defect area ratio term. The square root of the ratio of the maximum to minimum width of the spina bifida is multiplied by the defect area ratio term, and the product is incremented by one and the natural logarithm is taken to obtain the geometric amplitude term. The distance from the ossification center of the lamina to the edge of the spina bifida is divided by the sum of the length of the posterior margin defect of the spinal canal and the maximum width of the spina bifida, and the ratio is incremented by one to obtain the structural proportion term. The geometric amplitude term is multiplied by the structural proportion term, and the product is divided by the sum of the absolute value of the cosine of the lamina opening angle and a constant one to obtain the spina bifida complexity assessment value.

5. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 4, characterized in that, The specific steps for determining the complexity level of the fracture boundary point based on the evaluation results are as follows: Real-time comparison of ridge fissure complexity assessment values and multi-level complex thresholds and Determine the complexity level of the fracture boundary points: when ≤ At that time, the index sequence of the fracture boundary points is directly written into the first-level field; when < < At that time, the index sequence of the fracture boundary points is written into the second-level field; when ≥ At that time, the index sequence of the fracture boundary points is written into the third-level field.

6. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 1, characterized in that, The specific steps for quantifying the deviation of the cleft size relative to the spinal canal size based on fetal spinal structure data are as follows: Extract the corresponding fetal spinal structure data. Squar the ratio of the posterior margin defect length of the spinal canal to the sagittal diameter of the spinal canal, add it to the square of the ratio of the maximum width of the spinal fissure to the transverse diameter of the spinal canal, and add one to the sum to obtain the geometric ratio term. Divide the area of ​​the spinal fissure by the product of the sagittal diameter and the transverse diameter of the spinal canal, and multiply the resulting ratio by the square root of the ratio of the maximum width to the minimum width of the spinal fissure to obtain the area-width coupling term. Add the geometric ratio term and the area-width coupling term, and take the natural logarithm to obtain the scale coupling term. Subtract one from the ratio of the distance between the ossification centers of the vertebral laminae to the distance between the pedicles, take the absolute value, and add one to obtain the structural deviation term. Multiply the scale coupling term and the structural deviation term to obtain the spinal fissure scale deviation assessment value.

7. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 6, characterized in that, The specific steps for performing position correction operation on the fracture boundary points based on the complexity level are as follows: Read the complexity level field of the fracture boundary points, and combine the ridge fracture scale deviation assessment value and the complexity level field to perform a position correction operation on the fracture boundary point sequence: When the complexity level field is level 1, no position correction operation is performed; When the complexity level field is level 2, the proportional consistency adjustment program is called to calculate the adjustable offset of the boundary point based on the proportional relationship between the spinal fissure scale deviation assessment value and the transverse diameter of the spinal canal. Position correction is performed on the fissure boundary point whose adjustable offset exceeds the proportional deviation threshold. When the complexity level field is level 3, a proportional progressive adjustment procedure is invoked to perform multiple rounds of position correction on the sequence of fissure boundary points using the spina bifida scale deviation assessment value as the adjustment control value. In each round of correction, the offset of the boundary points is calculated based on the proportional relationship between the spina bifida scale deviation assessment value and the sagittal diameter, transverse diameter, width, and area of ​​the spina bifida. For fissure boundary points whose offset exceeds the proportional deviation threshold, progressive position correction is performed according to the magnitude of the spina bifida scale deviation assessment value, and the area of ​​the spina bifida is updated after each round of correction.

8. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 1, characterized in that, The specific steps for extracting the fracture boundary points after position correction, identifying three types of abrupt change points, and aggregating them to form key structural points are as follows: Extract all position-corrected fracture boundary points and perform equal-step traversal, calculate the direction change rate between adjacent fracture boundary points, and generate a direction change sequence; perform threshold filtering on the direction change sequence to identify the abrupt change locations of the direction change, and write the corresponding fracture boundary points into the morphological inflection point set. Extract the corresponding spinal fissure width and area, construct a width interpolation model based on a one-dimensional cubic spline interpolation algorithm, and generate a fissure width distribution curve; perform monotonicity detection on the width distribution curve, calculate the adjacent width difference according to the interpolation sampling point order during the detection process, identify the position where the width difference sign is reversed and the absolute value of the difference is greater than the width abrupt change threshold as the position where the width change is discontinuous, and write the corresponding fissure boundary point into the width abrupt change point set; A proportional coherence matrix is ​​constructed based on the length of the posterior margin defect of the spinal canal and the area of ​​the spinal fissure. During the construction process, discrete segments along the length direction of the fissure are used as indices. For each segment, the normalized proportional value of the length of the posterior margin defect of the spinal canal and the area of ​​the spinal fissure is calculated. The normalized proportional value is filled into the corresponding element of the proportional coherence matrix. A coherence check is performed on the proportional coherence matrix. During the check, the difference between adjacent normalized proportional values ​​is calculated along the length direction of the fissure. The position where the absolute value of the difference exceeds the coherence threshold is identified as the abrupt change position in the proportional structure. The corresponding fissure boundary point is written into the defect coherence abrupt change point set. Cross-comparison is performed on morphological inflection points, width abrupt change points, and defect continuity abrupt change points to identify crack boundary points that simultaneously satisfy two or more labels and mark them as key structural points; wherein the labels include: morphological inflection points, width abrupt change points, and defect continuity abrupt change points.

9. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 8, characterized in that, The specific steps for quantitatively evaluating the mutation intensity of the ridge fissure geometry based on the number of three types of mutation points and key structural points are as follows: The number of morphological inflection points, width abrupt change points, defect continuity abrupt change points, key structural points, and the total number of fissure boundary points are statistically analyzed. The abrupt change ratio is obtained by summing the number of morphological inflection points, width abrupt change points, and defect continuity abrupt change points and then dividing by the total number of fissure boundary points. The key point proportion is obtained by adding one to the number of key structural points and then dividing by the sum of the number of morphological inflection points, width abrupt change points, and defect continuity abrupt change points plus one. The defect width coupling term is obtained by dividing the length of the posterior edge defect of the spinal canal by the sum of the maximum and minimum widths of the spinal fissure, adding one to the ratio, and then taking the natural logarithm. The abrupt change ratio, key point proportion, and defect width coupling term are multiplied sequentially to obtain the spinal fissure structural abrupt change assessment value.

10. The automatic fetal image parameter extraction system for screening spinal cord neural tube defects according to claim 9, characterized in that, The specific steps for generating the basic parameter set and fetal spina bifida image analysis dataset are as follows: Real-time comparison of ridge fissure structural mutation assessment values ​​and mutation thresholds to determine data organization methods: When the spinal cleft structural mutation assessment value is less than or equal to the mutation threshold, a basic set of spinal cleft parameters is generated, including the maximum width of the spinal cleft, the minimum width of the spinal cleft, the area of ​​the spinal cleft, the length of the defect at the posterior margin of the spinal canal, and the spinal cleft structural mutation assessment value. When the structural mutation assessment value of spina bifida is greater than the mutation threshold, a fetal spina bifida image analysis dataset is generated based on the basic parameter set of spina bifida, combined with the sequence of rift boundary points, the set of morphological inflection points, the set of width mutation points, and the set of defect continuity mutation points. The basic parameter set of spina bifida and the fetal spina bifida image analysis dataset are written into the system output buffer and packaged into a spina bifida image parameter result data package, which is then sent to the doctor's display terminal.

Images