A method for identifying and positioning a spinal model

Abstract

The present application relates to the technical field of spine image recognition, and particularly relates to a spine model recognition and positioning method, which determines the vertebra sub-contours of each vertebra in the first model image of the target object in a first pose state, and determines whether the recognition reliability of the vertebra sub-contours meets the standard according to the position distribution state of the contour recognition line; the recognition reference contour of the vertebra sub-contour whose recognition reliability does not meet the standard is determined, the pose change data is determined based on the coordinate change of the edge calibration point of the recognition reference contour, the pose change reference coordinates of the edge calibration point in the second model image are calculated, and whether the model image is effective is determined based on the actual coordinate difference of the edge calibration point in the second model image and the pose change reference coordinates, thereby, the verification scheme based on the associated image data and the model effectiveness determination means are provided for the vertebra sub-contours whose recognition reliability does not meet the standard, and the spine model recognition and positioning accuracy and effectiveness are improved.

Description

Technical Field

[0001] This invention relates to the field of spinal image recognition technology, and in particular to a method for recognizing and locating a spinal model. Background Technology

[0002] In the fields of spinal disease diagnosis and rehabilitation assessment, accurate identification and localization of spinal models are crucial for personalized medicine and precise intervention. Currently, acquiring spinal structural information and constructing digital spinal models through medical imaging has become the mainstream technical approach. However, due to the complexity of the spinal structure and significant individual differences, the diverse vertebral morphologies, and the close anatomical relationships between adjacent vertebrae, coupled with image noise and soft tissue interference that may exist during image acquisition, automatic identification of vertebral sub-contours is prone to problems such as blurred edges and misjudgment of feature points, directly affecting the reliability of identification. Identification and localization methods based on single images are difficult to effectively adapt to various interference scenarios, and are prone to problems such as mismatch of correspondences and localization deviations. Therefore, how to achieve high-reliability identification of vertebral sub-contours in scenarios with various interferences has become an important research direction for improving the accuracy of spinal model identification and localization.

[0003] For example, Chinese Patent Publication No. CN117350932A discloses a method, apparatus, and device for segmenting and recognizing spinal images, which can reduce computational resources and improve vertebral segmentation accuracy. This application uses some vertebrae of the spine as key vertebrae to perform key vertebral segmentation and recognition on spinal images, obtaining key vertebral segmentation and recognition results. Based on the position of vertebral positioning points in the spinal image, the key vertebral segmentation results, and the recognition results, the vertebral category corresponding to each vertebral positioning point is determined. One of the vertebral positioning points is used as a clipping reference point. Based on the vertebral positioning points adjacent to the clipping reference point, the clipping boundary is determined, and an image sub-block including a single complete vertebra is clipped from the spinal image. The image sub-block and a Gaussian probability distribution map generated centered on the clipping reference point are input into the vertebral segmentation model to obtain the segmentation result of a single vertebra. Based on the vertebral category corresponding to the vertebral positioning point used as the clipping reference point, the recognition result of a single vertebra is obtained.

[0004] The following problems still exist in the existing technology:

[0005] Existing technologies rely on single images, which are difficult to adapt to interference scenarios and are prone to recognition errors. There is a lack of verification schemes based on associated image data and means to determine the effectiveness of models for vertebral contours that do not meet the recognition reliability standards, which in turn affects the recognition and positioning accuracy and effectiveness of spinal models. Summary of the Invention

[0006] To address these issues, this invention provides a method for identifying and locating a spinal model. This method overcomes the problems of existing technologies, such as difficulty in adapting to interference scenarios, the lack of verification schemes based on associated image data, and the absence of means to determine the validity of vertebral sub-contours with unreliable identification, which affect the accuracy and effectiveness of spinal model identification and location.

[0007] To achieve the above objectives, the present invention provides a method for identifying and locating a spinal model, comprising:

[0008] Acquire a first model image of the target object captured in a first pose state and a second model image captured in a second pose state;

[0009] Within the first model image, the sub-contours of each vertebra are determined, and several edge calibration points of the sub-contours are marked in a pre-set world coordinate system. The recognition reliability of the sub-contours is determined based on the positional distribution of the contour recognition lines.

[0010] Each contour recognition line is determined based on edge calibration points;

[0011] In response to the judgment result that the recognition confidence of the vertebral sub-contours is not up to standard, the vertebral sub-contours of each vertebra in the second model image are extracted, and the correspondence between the vertebral sub-contours of each vertebra in the second model image and the vertebral sub-contours of each vertebra in the first model image is determined according to the position of each vertebral sub-contour in the world coordinate system.

[0012] The vertebral sub-contours that are spatially adjacent to the vertebral sub-contours with unsatisfactory recognition reliability are identified as recognition reference contours. Based on the coordinates of the edge calibration points of the recognition reference contours in the first model image and the second model image, the pose change data are determined.

[0013] Based on the coordinates of the edge calibration points on the vertebral contour that do not meet the reliability standard and the pose change data, the pose change reference coordinates of the edge calibration points in the second model image are calculated. The validity of the model image is determined based on the difference between the pose change reference coordinates and the actual coordinates of the edge calibration points in the second model image.

[0014] Furthermore, the process of calibrating several edge calibration points of the vertebral sub-contour includes:

[0015] Obtain the original coordinate set of the vertebral contour, wherein each coordinate in the original coordinate set corresponds to a contour point on the vertebral contour;

[0016] A rectangular coordinate system is established with the geometric center of the vertebral contour as the origin. All coordinates in the original coordinate set are transformed to the rectangular coordinate system to obtain a standardized coordinate set.

[0017] Calculate the distance from each contour point in the standardized coordinate set to the origin, and determine the contour point corresponding to the maximum distance in each quadrant of the rectangular coordinate system as the edge calibration point;

[0018] The edge calibration points include a first edge calibration point determined in the first quadrant, a second edge calibration point determined in the second quadrant, a third edge calibration point determined in the third quadrant, and a fourth edge calibration point determined in the fourth quadrant.

[0019] Further, the positional distribution of the contour recognition lines is determined, including:

[0020] A first contour recognition line is determined based on the first edge calibration point and the second edge calibration point, and a second contour recognition line is determined based on the third edge calibration point and the fourth edge calibration point;

[0021] Determine the angle between the first contour recognition line and the second contour recognition line.

[0022] Further, determining whether the recognition reliability of the vertebral sub-contour meets the standard includes:

[0023] The included angle is compared with a preset included angle threshold. If the included angle is greater than the included angle threshold, the recognition reliability of the vertebral sub-contour is determined to be substandard.

[0024] Furthermore, the process of determining the correspondence between the vertebral sub-contours of each vertebra in the second model image and the vertebral sub-contours of each vertebra in the first model image includes:

[0025] The vertebral sub-contours of each vertebra in the first model image and the vertebral sub-contours of each vertebra in the second model image are both transformed to the world coordinate system to obtain the position information of each vertebral sub-contour in the world coordinate system.

[0026] Calculate the matching parameters between the vertebral sub-contour of each vertebra in the second model image and the vertebral sub-contour of each vertebra in the first model image;

[0027] Based on the determination result that the matching parameters meet the preset matching conditions, it is determined that the two corresponding vertebral sub-contours have a corresponding relationship.

[0028] Furthermore, the corresponding matching parameter is the straight-line distance between the geometric center coordinates of the vertebral contour in the second model image and the geometric center coordinates of the vertebral contour in the first model image in the world coordinate system.

[0029] The matching condition is that the straight-line distance is less than or equal to a preset distance threshold, which is determined based on the average size of the vertebrae.

[0030] Furthermore, pose change data is determined based on the coordinates of the edge calibration points of the identified reference contour within the first and second model images, respectively, including:

[0031] Determine the x-coordinate, y-coordinate, and ordinate of the edge calibration points of the identification reference contour within the first and second model images, respectively.

[0032] Calculate the changes in the horizontal, vertical, and y coordinates of the edge calibration points within the second model image relative to those within the first model image.

[0033] The set of data consisting of the average changes in the horizontal coordinate, the average changes in the vertical coordinate, and the average changes in the ordinate of several edge calibration points on the reference contour is defined as pose change data.

[0034] Furthermore, the process of calculating the pose change reference coordinates of the edge calibration points within the second model image includes:

[0035] Within the first model image, obtain the coordinates of each edge calibration point on the recognition reference contour for vertebral sub-contours with unsatisfactory recognition reliability.

[0036] The x-coordinate of each edge calibration point is added to the x-coordinate change in the pose change data to determine the reference x-coordinate of the pose change of each edge calibration point;

[0037] The vertical coordinates of each edge calibration point are added to the vertical coordinate change in the pose change data to determine the reference vertical coordinates of the pose change of each edge calibration point;

[0038] The ordinate of each edge calibration point is added to the ordinate change in the pose change data to determine the reference ordinate of the pose change of each edge calibration point.

[0039] Furthermore, the process of determining the actual coordinate differences within the second model image includes:

[0040] Determine the corresponding feature vertebral sub-contours in the second model image for vertebral sub-contours whose recognition confidence in the first model image is insufficient, and obtain the actual coordinates of each edge calibration point on the feature vertebral sub-contours;

[0041] Calculate the difference between the actual coordinates and the pose change reference coordinates, and determine the cumulative difference of the actual coordinates based on the average difference of the horizontal coordinate, the average difference of the vertical coordinate, and the average difference of the ordinate.

[0042] Furthermore, the process of determining whether a model image is valid includes:

[0043] The actual accumulated coordinate difference is compared with the preset accumulated coordinate difference threshold.

[0044] If the actual coordinate difference accumulation is less than or equal to the preset coordinate difference accumulation threshold, the model image is determined to be valid.

[0045] If the accumulated amount of actual coordinate differences is greater than the preset threshold for accumulated coordinate differences, the model image is determined to be invalid.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention determines the vertebral sub-contours of each vertebra in the first model image of the target object in its first pose state, and marks several edge calibration points of the vertebral sub-contours in a pre-set world coordinate system. The reliability of the vertebral sub-contour recognition is determined based on the positional distribution of the contour recognition lines. The correspondence between the vertebral sub-contours in the second model image and the vertebral sub-contours in the first model image is determined based on the position of each vertebral sub-contour in the second model image. By determining the recognition reference contour for vertebral sub-contours with unreliable recognition reliability, pose change data is determined based on the coordinate changes of the edge calibration points of the recognition reference contour. By calculating the pose change reference coordinates of the edge calibration points in the second model image, the validity of the model image is determined based on the difference between the pose change reference coordinates and the actual coordinates of the edge calibration points in the second model image. Thus, the present invention provides a verification scheme based on associated image data and a means of determining the validity of the model for vertebral sub-contours with unreliable recognition reliability, thereby improving the accuracy and effectiveness of spinal model recognition and positioning.

[0047] Furthermore, this invention quantifies the accuracy of contour recognition by comparing the angle between the contour recognition lines and a preset threshold. The normal anatomical structure of a vertebra determines that the shape of its upper and lower boundaries follows a relatively fixed pattern. This pattern can be reflected by the angle between the first and second contour recognition lines corresponding to the upper and lower boundaries. Under normal circumstances, this angle of a single vertebra will be stably within a reasonable range that conforms to its inherent shape. In actual judgment, the angle between the first and second contour recognition lines of the sub-contour of the vertebra to be detected is compared with the preset threshold. If the angle is greater than the preset threshold, it indicates that the shape of the upper and lower boundaries of the currently identified vertebra deviates from the inherent reasonable range of a single vertebra, and there is a problem of edge blurring during the contour recognition process. This makes the extracted contour unable to accurately reflect the true shape of the vertebra itself. Therefore, the recognition reliability of the sub-contour of the vertebra is determined to be substandard, thereby improving the recognition efficiency.

[0048] Furthermore, the significance of determining the correspondence of vertebral sub-contours in this invention lies in utilizing the relative stability of the spatial position of the same vertebra in images taken in different poses, and achieving precise correspondence of vertebrae across images through spatial distance matching of geometric centers. Since the overall morphological changes of the spine mainly stem from uneven intervertebral spaces rather than changes in the morphology of individual vertebrae, the spatial position of the same vertebra in images taken in different poses may exhibit overall changes such as movement, but it still maintains continuity relative to the spatial arrangement of other vertebrae. Based on this, this invention uniformly transforms all vertebral sub-contours in two images to the world coordinate system, determining the straight-line distance between the geometric centers of each vertebral sub-contour in the second model image and each vertebral sub-contour in the first model image. This distance quantifies the degree of proximity of the two vertebrae in space. When the geometric center distance between two vertebral sub-contours meets the requirements, they are determined to be corresponding contours of the same vertebra in different images, thereby achieving accurate matching of vertebrae across images.

[0049] Furthermore, this invention determines pose change data by accurately calculating and averaging the coordinate changes of the reference contour edge calibration points across images. The significance lies in identifying the reference contour as an adjacent vertebral contour that has passed the confidence assessment. Its edge calibration points can stably characterize the geometric features of the vertebra. By extracting the three-dimensional coordinates of these calibration points in the first and second model images respectively, the spatial position differences across poses can be fully captured. The pose change data constructed by the average of the three-dimensional coordinate changes can quantitatively reflect the translation trend of the spine in three-dimensional space. The average of the changes in each axis directly corresponds to the direction and magnitude of movement. Thus, it provides a verification scheme based on associated image data for vertebral sub-contours whose recognition confidence does not meet the standards.

[0050] Furthermore, this invention calculates the pose change reference coordinates within the second model image using the coordinates of the low-confidence contour edge calibration points in the first model image and the pose change data. The significance lies in providing theoretical reference coordinates for the low-confidence vertebral sub-contours based on the true pose transformation law of the spine. It can be understood that the pose change data originates from the average cross-image coordinate change of the identification reference contour, which can truly reflect the three-dimensional spatial translation law of the same spine from the first pose to the second pose. Since the overall pose change of the spine is continuous, and the spatial transformation trends of adjacent vertebrae are similar, the coordinates of the low-confidence contour edge calibration points in the first model image are superimposed with the pose change amount along the corresponding axis to obtain the position of the calibration point in the second model image based on the pose change logic. This data-driven approach achieves objective verification, improving the reliability of spinal model identification and positioning.

[0051] Furthermore, this invention locates the feature vertebral sub-contours in the second model image that correspond to the low-confidence vertebral sub-contours in the first model image, obtains the actual coordinates of their edge calibration points, calculates the difference between the actual coordinates and the pose change reference coordinates, and obtains the average difference in the three-dimensional axis. Finally, this average difference is compared with a preset threshold to determine the validity of the model image. The significance lies in constructing an objective image validity verification mechanism based on quantified differences, solving the problem of effectively distinguishing contour recognition errors in single image recognition. The difference between the actual coordinates and reference coordinates of the corresponding feature vertebral sub-contours is calculated. When the average difference is within the threshold, it indicates that the image acquisition quality is reliable and can be used for subsequent spinal model construction; if it exceeds the threshold, it indicates that the image validity is questionable, avoiding invalid images from affecting subsequent diagnostic and treatment applications. This provides a verification scheme based on associated image data and a means of determining model validity for vertebral sub-contours with unreliable recognition confidence, improving the accuracy and effectiveness of spinal model recognition and positioning. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the steps of the spinal model identification and localization method according to an embodiment of the present invention;

[0053] Figure 2 This is a step diagram illustrating the process of marking several edge calibration points on the vertebral contour according to an embodiment of the present invention;

[0054] Figure 3 This is a step diagram illustrating the process of determining the correspondence between vertebral sub-contours in different model images according to an embodiment of the present invention;

[0055] Figure 4 This is a flowchart illustrating the steps for determining pose change data in an embodiment of the present invention.

[0056] Figure 5 This is a flowchart illustrating the logic for determining whether a model image is valid, as shown in an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0058] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0059] It should be noted that in the description of this invention, the terms "upper," "lower," "inner," "outer," etc., which indicate the direction or positional relationship, are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0060] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation" and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0061] Please see Figure 1 The diagram shows the steps of the spinal model identification and localization method according to an embodiment of the present invention. The spinal model identification and localization method of the present invention includes:

[0062] Step S100: Acquire a first model image of the target object captured in the first pose state and a second model image captured in the second pose state.

[0063] In this invention, the first model image and the second model image can be X-ray images. The first pose state and the second pose state are different in this invention. By controlling the X-ray imaging instrument, the spinal images of the target object are acquired in the first pose state and the second pose state respectively. The two acquired spinal images are the first model image in the first pose state and the second model image in the second pose state. Preferably, the second pose state can be the position after the X-ray imaging instrument moves 3cm along the direction perpendicular to the length of the spine at the position of the first pose state.

[0064] In the implementation of this invention, the target object can be a patient, and this invention does not limit this.

[0065] It should be noted that the present invention does not impose any restrictions on the order in which the first model image taken in the first pose state and the second model image taken in the second pose state are captured.

[0066] Step S200: Determine the vertebral sub-contour of each vertebra within the first model image, and mark several edge calibration points of the vertebral sub-contour in a pre-set world coordinate system. Determine whether the recognition reliability of the vertebral sub-contour meets the standard based on the positional distribution of the contour recognition line.

[0067] Each contour recognition line is determined based on edge calibration points;

[0068] Specifically, this invention does not limit the specific method for determining the sub-contours of each vertebra. The recognition of the sub-contours of the vertebra can be achieved by using a pre-imported threshold segmentation method and edge detection algorithm. Recognizing feature contours in images is a common technical means, which will not be elaborated here.

[0069] Step S300: In response to the judgment result that the recognition confidence of the vertebral sub-contour is not up to standard, extract the vertebral sub-contour of each vertebra in the second model image, and determine the correspondence between the vertebral sub-contour of each vertebra in the second model image and the vertebral sub-contour of each vertebra in the first model image according to the position of each vertebral sub-contour in the world coordinate system.

[0070] Step S400: The vertebral sub-contours that are spatially adjacent to the vertebral sub-contours with unsatisfactory recognition credibility are determined as recognition reference contours. Based on the coordinates of the edge calibration points of the recognition reference contours in the first model image and the second model image, the pose change data are determined.

[0071] In this invention, the invention does not specify the specific method for determining the vertebral sub-outlines of vertebrae that are spatially adjacent to those of vertebrae with unsatisfactory recognition reliability. It is understood that each vertebra is connected end to end to form the spine, and each vertebra has one or two spatially adjacent vertebrae. Preferably, the spatial positional relationship of each vertebra can be used to determine the adjacent vertebrae.

[0072] Step S500: Calculate the pose change reference coordinates of the edge calibration points in the second model image based on the coordinates of the edge calibration points on the vertebral contour with unreliable confidence and the pose change data. Determine whether the model image is valid based on the difference between the pose change reference coordinates and the actual coordinates of the edge calibration points in the second model image.

[0073] Specifically, in this invention, the horizontal axis of the world coordinate system is perpendicular to the length of the spine, the vertical axis is parallel to the length of the spine, and the vertical axis is the depth direction of the image.

[0074] Those skilled in the art will understand that when the reliability of vertebral contour recognition is deemed insufficient, introducing a second model image in a second pose state is significant in constructing a cross-pose verification and correction mechanism to overcome the limitations of single-image recognition. The introduction of the second model image establishes a correspondence with the vertebral contour in the first model image, finding a common reference for low-reliability contours and avoiding the cumulative transmission of recognition errors from a single image. Furthermore, by extracting pose patterns from the coordinate changes of adjacent reliable vertebrae in the two images, the overall spatial transformation characteristics of the same spine under different poses are reflected. This allows for the calculation of theoretical reference coordinates for the edge calibration points of low-reliability contours. The difference between the reference coordinates and the actual coordinates in the second model image objectively determines whether the deviation is caused by image acquisition issues, ultimately achieving accurate verification of the model image's effectiveness.

[0075] Specifically, please refer to Figure 2 The diagram illustrates the steps of marking several edge calibration points on the vertebral contour according to an embodiment of the present invention. The process of marking several edge calibration points on the vertebral contour includes:

[0076] Step S201: Obtain the original coordinate set of the vertebral contour, wherein each coordinate in the original coordinate set corresponds to a contour point on the vertebral contour;

[0077] Step S202: Establish a rectangular coordinate system with the geometric center of the vertebral contour as the origin, and transform all coordinates in the original coordinate set to the rectangular coordinate system to obtain a standardized coordinate set;

[0078] Step S203: Calculate the distance from each contour point in the standardized coordinate set to the origin, and determine the contour point corresponding to the maximum distance in each quadrant of the rectangular coordinate system as the edge calibration point;

[0079] The edge calibration points include a first edge calibration point determined in the first quadrant, a second edge calibration point determined in the second quadrant, a third edge calibration point determined in the third quadrant, and a fourth edge calibration point determined in the fourth quadrant.

[0080] Those skilled in the art will understand that obtaining the original coordinate set of the vertebral contour to acquire basic data of all contour points, then establishing a Cartesian coordinate system with the geometric center of the vertebral contour as the origin, transforming the original coordinates into this local coordinate system to form a standardized coordinate set, and assigning the standardized contour points to four quadrants based on the quadrant division rules of the Cartesian coordinate system. Utilizing the geometric characteristic that the contour point with the largest distance from the origin in each quadrant best reflects the extreme morphological features of the contour in that quadrant direction, the contour point farthest from the origin in each quadrant is selected as the edge calibration point. The resulting edge calibration points corresponding to the four quadrants can completely characterize the overall geometric contour features of the vertebral contour from four vertical directions.

[0081] Specifically, determining the positional distribution of contour recognition lines includes:

[0082] A first contour recognition line is determined based on the first edge calibration point and the second edge calibration point, and a second contour recognition line is determined based on the third edge calibration point and the fourth edge calibration point;

[0083] Determine the angle between the first contour recognition line and the second contour recognition line.

[0084] Those skilled in the art will understand that by determining the distribution of contour recognition lines, the principle lies in constructing contour recognition lines with contour orientation characteristics based on representative edge calibration points of the vertebral contour. In this invention, the first and second edge calibration points correspond to the first and second quadrants of the Cartesian coordinate system, respectively, and the third and fourth edge calibration points correspond to the third and fourth quadrants. Connecting the first and second edge calibration points in the upper half of the Cartesian coordinate system forms the first contour recognition line, and connecting the third and fourth edge calibration points in the lower half of the Cartesian coordinate system forms the second contour recognition line. These two straight lines can respectively characterize the overall boundary morphology of the upper and lower contours of the vertebral contour. By calculating the angle between two contour recognition lines, the geometric features of the contour are transformed into quantifiable angular parameters. These angular parameters can intuitively reflect the overall shape and orientation of the upper and lower boundaries of the vertebral contour. Since the upper and lower boundaries belong to the same vertebra, the interference caused by scoliosis will not have a significant impact on the overall shape of the vertebra. The unevenness of the gaps between vertebrae is the main reason for the change in the overall shape of the spine. If the contour recognition is accurate, the overall shape and orientation will be within a reasonable range that conforms to the normal shape of the vertebra. If the contour has blurred edges, the recognized overall shape and orientation will deviate from the reasonable range, thus improving the efficiency of vertebral sub-contour recognition reliability.

[0085] Specifically, determining whether the recognition reliability of the vertebral sub-contour meets the standard includes:

[0086] The included angle is compared with a preset included angle threshold. If the included angle is greater than the included angle threshold, the recognition reliability of the vertebral sub-contour is determined to be substandard.

[0087] If the included angle is less than or equal to the included angle threshold, then the recognition reliability of the vertebral sub-contour is determined to be up to standard.

[0088] In this invention, the included angle threshold can be set based on a large amount of historical anatomical data of normal vertebrae, and the included angle of normal vertebrae can be statistically analyzed. In this invention, the included angle threshold ranges from [6° to 10°], and preferably, the included angle threshold is 8°.

[0089] Specifically, this invention quantifies the accuracy of contour recognition by comparing the angle between the contour recognition lines and a preset threshold. The normal anatomical structure of a vertebra determines that the shape of its upper and lower boundaries follows a relatively fixed pattern. This pattern can be reflected by the angle between the first and second contour recognition lines corresponding to the upper and lower boundaries. Under normal circumstances, this angle of a single vertebra will be stably within a reasonable range that conforms to its inherent shape. In actual judgment, the angle between the first and second contour recognition lines of the sub-contour of the vertebra to be detected is compared with the preset threshold. If the angle is greater than the preset threshold, it indicates that the shape of the upper and lower boundaries of the currently identified vertebra deviates from the inherent reasonable range of a single vertebra, and there is a problem of edge blurring during the contour recognition process. This makes the extracted contour unable to accurately reflect the true shape of the vertebra itself, and therefore the recognition reliability of the sub-contour of the vertebra is determined to be substandard.

[0090] Specifically, please refer to Figure 3 The diagram illustrates the steps of determining the correspondence between vertebral sub-contours in different model images according to an embodiment of the present invention. The process of determining the correspondence between the vertebral sub-contours in the second model image and the vertebral sub-contours in the first model image includes:

[0091] Step S301: Convert the vertebral sub-contours of each vertebra in the first model image and the vertebral sub-contours of each vertebra in the second model image to the world coordinate system to obtain the position information of each vertebral sub-contour in the world coordinate system.

[0092] Step S302: Calculate the matching parameters between the vertebral sub-contour of each vertebra in the second model image and the vertebral sub-contour of each vertebra in the first model image.

[0093] Step S303: Based on the determination result that the matching parameters meet the preset matching conditions, determine that the corresponding two vertebral sub-contours have a corresponding relationship.

[0094] Specifically, the corresponding matching parameter is the straight-line distance between the geometric center coordinates of the vertebral sub-contour in the second model image and the geometric center coordinates of the vertebral sub-contour in the first model image in the world coordinate system.

[0095] The matching condition is that the straight-line distance is less than or equal to a preset distance threshold, which is determined based on the average size of the vertebrae.

[0096] In this invention, the coordinates of the geometric center can be determined based on the average coordinates of several points on the vertebral contour on different coordinate axes, which will not be elaborated here.

[0097] In the implementation of this invention, the preset distance threshold can be obtained by pre-measurement and calculation. The dimensions of each vertebra along the length of the spine are pre-measured, and the average length of the vertebra along the length of the spine is determined as the preset distance threshold, so as to ensure that the reasonable positional deviation of the same vertebra due to the change of posture will not be misidentified as different vertebrae.

[0098] As will be understood by those skilled in the art, the significance of determining the correspondence of vertebral sub-contours in this invention lies in utilizing the relative stability of the spatial position of the same vertebra in images of different poses, and achieving precise correspondence of vertebrae across images through spatial distance matching of geometric centers. Since changes in the overall morphology of the spine mainly stem from uneven intervertebral spaces rather than changes in the morphology of individual vertebrae, the spatial position of the same vertebra in images taken in different poses may exhibit overall changes such as movement, but it still maintains continuity relative to the spatial arrangement of other vertebrae. Based on this, this invention uniformly transforms all vertebral sub-contours in two images to the world coordinate system, determining the straight-line distance between the geometric centers of each vertebral sub-contour in the second model image and each vertebral sub-contour in the first model image. This distance quantifies the degree of proximity of the two vertebrae in space. When the geometric center distance between two vertebral sub-contours meets the requirements, they are determined to be corresponding contours of the same vertebra in different images, thereby achieving accurate matching of vertebrae across images.

[0099] Specifically, please refer to Figure 4 The diagram illustrates the steps for determining pose change data according to an embodiment of the present invention. The pose change data is determined based on the coordinates of the edge calibration points of the identified reference contour within the first and second model images, including:

[0100] Step S401: Determine the horizontal, vertical, and ordinate of the edge calibration points of the identification reference contour in the first model image and the second model image respectively.

[0101] Step S402: Calculate the changes in the horizontal coordinate, vertical coordinate, and ordinate of the edge calibration points in the second model image relative to the first model image.

[0102] Step S403: The set of data consisting of the average value of the horizontal coordinate change, the average value of the vertical coordinate change, and the average value of the vertical coordinate change of several edge calibration points on the identified reference contour is determined as pose change data.

[0103] Specifically, this invention determines pose change data by accurately calculating and averaging the coordinate changes of the reference contour edge calibration points across images. The significance lies in identifying the reference contour as an adjacent vertebral contour that has passed the confidence assessment. Its edge calibration points can stably characterize the geometric features of the vertebra. Furthermore, by extracting the three-dimensional coordinates of these calibration points in the first and second model images respectively, the spatial position differences across poses can be fully captured. The pose change data constructed by the average of the three-dimensional coordinate changes can quantitatively reflect the translational trend of the spine in three-dimensional space. The average of the changes in each axis directly corresponds to the direction and magnitude of movement. Thus, it provides a verification scheme based on associated image data for vertebral sub-contours whose recognition confidence does not meet the standards.

[0104] Specifically, the process of calculating the pose change reference coordinates of the edge calibration points within the second model image includes:

[0105] Within the first model image, obtain the coordinates of each edge calibration point on the recognition reference contour for vertebral sub-contours with unsatisfactory recognition reliability.

[0106] The x-coordinate of each edge calibration point is added to the x-coordinate change in the pose change data to determine the reference x-coordinate of the pose change of each edge calibration point;

[0107] The vertical coordinates of each edge calibration point are added to the vertical coordinate change in the pose change data to determine the reference vertical coordinates of the pose change of each edge calibration point;

[0108] The ordinate of each edge calibration point is added to the ordinate change in the pose change data to determine the reference ordinate of the pose change of each edge calibration point.

[0109] Specifically, this invention calculates the reference coordinates for pose changes within the second model image using the coordinates of the low-confidence contour edge calibration points in the first model image and pose change data. The significance lies in providing theoretical reference coordinates for the low-confidence vertebral sub-contours based on the true pose transformation laws of the spine. It can be understood that the pose change data originates from the average cross-image coordinate changes of the identification reference contour, truly reflecting the three-dimensional spatial translation law of the same spine from the first pose to the second pose. Since the overall pose change of the spine is continuous, and the spatial transformation trends of adjacent vertebrae are similar, the coordinates of the low-confidence contour edge calibration points in the first model image are superimposed with the pose change amount along the corresponding axis to obtain the position of the calibration point in the second model image based on the pose change logic. This data-driven approach achieves objective verification, improving the reliability of spinal model identification and positioning.

[0110] Specifically, the process of determining the actual coordinate differences within the second model image includes:

[0111] Determine the corresponding feature vertebral sub-contours in the second model image for vertebral sub-contours whose recognition confidence in the first model image is insufficient, and obtain the actual coordinates of each edge calibration point on the feature vertebral sub-contours;

[0112] In this invention, the feature vertebral sub-contours in the second model image are the vertebral sub-contours in the first model image whose recognition reliability is not up to standard in the first pose state.

[0113] Calculate the difference between the actual coordinates and the pose change reference coordinates, and determine the cumulative difference of the actual coordinates based on the average difference of the horizontal coordinate, the average difference of the vertical coordinate, and the average difference of the ordinate.

[0114] In this invention, the sum of the average difference in the horizontal coordinate, the average difference in the vertical coordinate, and the average difference in the ordinate can be determined as the cumulative difference in the actual coordinates. The average difference in the horizontal coordinate is the arithmetic mean of the difference between the actual horizontal coordinates of all edge calibration points and the horizontal coordinates of the pose change reference coordinates. The average difference in the vertical coordinate is the arithmetic mean of the difference between the actual vertical coordinates of all edge calibration points and the vertical coordinates of the pose change reference coordinates. The average difference in the ordinate is the arithmetic mean of the difference between the actual vertical coordinates of all edge calibration points and the vertical coordinates of the pose change reference coordinates.

[0115] Specifically, please refer to Figure 5 As shown, it is a flowchart of the logic for determining whether a model image is valid according to an embodiment of the present invention. The process for determining whether a model image is valid includes:

[0116] The actual accumulated coordinate difference is compared with the preset accumulated coordinate difference threshold.

[0117] If the actual coordinate difference accumulation is less than or equal to the preset coordinate difference accumulation threshold, the model image is determined to be valid.

[0118] If the accumulated amount of actual coordinate differences is greater than the preset threshold for accumulated coordinate differences, the model image is determined to be invalid.

[0119] In the implementation of this invention, the preset coordinate difference accumulation threshold can be set by those skilled in the art based on the difference between the first pose state and the second pose state. Preferably, when the imaging instrument in the second pose state is an X-ray image and is located 3cm after moving in a direction perpendicular to the length of the spine at the position of the first pose state, the coordinate difference accumulation threshold can be set to 6mm.

[0120] Those skilled in the art will understand that by locating the feature vertebral sub-contours in the second model image that correspond to the low-confidence vertebral sub-contours in the first model image, obtaining the actual coordinates of their edge calibration points, calculating the difference between the actual coordinates and the pose change reference coordinates, and obtaining the average difference in the three-dimensional axis, and finally comparing this average with a preset threshold to determine the validity of the model image, the significance lies in constructing an objective image validity verification mechanism based on quantified differences, solving the problem of difficulty in effectively distinguishing contour recognition errors in single image recognition. Calculating the difference between the actual coordinates and reference coordinates of the corresponding feature vertebral sub-contours, if the average difference is within the threshold, indicates that the image acquisition quality is reliable and can be used for subsequent spinal model construction; if it exceeds the threshold, it indicates that the image validity is questionable, avoiding invalid images from affecting subsequent diagnostic and treatment applications. This provides a verification scheme based on associated image data and a means of determining model validity for vertebral sub-contours with unacceptable recognition confidence, improving the accuracy and effectiveness of spinal model recognition and positioning.

[0121] Those skilled in the art will understand that all or part of the processes described above can be implemented by a computer program instructing related hardware, and the computer program described above can be stored in a computer-readable storage medium.

[0122] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0123] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying and locating a spinal model, characterized in that, include: Acquire a first model image of the target object captured in a first pose state and a second model image captured in a second pose state; Within the first model image, the sub-contours of each vertebra are determined, and several edge calibration points of the sub-contours are marked in a pre-set world coordinate system. The recognition reliability of the sub-contours is determined based on the positional distribution of the contour recognition lines. Each contour recognition line is determined based on edge calibration points; In response to the judgment result that the recognition confidence of the vertebral sub-contours is not up to standard, the vertebral sub-contours of each vertebra in the second model image are extracted, and the correspondence between the vertebral sub-contours of each vertebra in the second model image and the vertebral sub-contours of each vertebra in the first model image is determined according to the position of each vertebral sub-contour in the world coordinate system. The vertebral sub-contours that are spatially adjacent to the vertebral sub-contours with unsatisfactory recognition reliability are identified as recognition reference contours. Based on the coordinates of the edge calibration points of the recognition reference contours in the first model image and the second model image, the pose change data are determined. Based on the coordinates of the edge calibration points on the vertebral contour that do not meet the reliability standard and the pose change data, the pose change reference coordinates of the edge calibration points in the second model image are calculated. The validity of the model image is determined based on the difference between the pose change reference coordinates and the actual coordinates of the edge calibration points in the second model image.

2. The method for identifying and locating a spinal model according to claim 1, characterized in that, The process of calibrating several edge calibration points of the vertebral sub-contour includes: Obtain the original coordinate set of the vertebral contour, wherein each coordinate in the original coordinate set corresponds to a contour point on the vertebral contour; A rectangular coordinate system is established with the geometric center of the vertebral contour as the origin. All coordinates in the original coordinate set are transformed to the rectangular coordinate system to obtain a standardized coordinate set. Calculate the distance from each contour point in the standardized coordinate set to the origin, and determine the contour point corresponding to the maximum distance in each quadrant of the rectangular coordinate system as the edge calibration point; The edge calibration points include a first edge calibration point determined in the first quadrant, a second edge calibration point determined in the second quadrant, a third edge calibration point determined in the third quadrant, and a fourth edge calibration point determined in the fourth quadrant.

3. The method for identifying and locating a spinal model according to claim 2, characterized in that, Determine the positional distribution of contour recognition lines, including: A first contour recognition line is determined based on the first edge calibration point and the second edge calibration point, and a second contour recognition line is determined based on the third edge calibration point and the fourth edge calibration point; Determine the angle between the first contour recognition line and the second contour recognition line.

4. The method for identifying and locating a spinal model according to claim 3, characterized in that, Determining whether the recognition reliability of the vertebral sub-contour meets the standard includes: The included angle is compared with a preset included angle threshold. If the included angle is greater than the included angle threshold, the recognition reliability of the vertebral sub-contour is determined to be substandard.

5. The method for identifying and locating a spinal model according to claim 4, characterized in that, The process of determining the correspondence between the sub-contours of each vertebra in the second model image and the sub-contours of each vertebra in the first model image includes: The vertebral sub-contours of each vertebra in the first model image and the vertebral sub-contours of each vertebra in the second model image are both transformed to the world coordinate system to obtain the position information of each vertebral sub-contour in the world coordinate system. Calculate the matching parameters between the vertebral sub-contour of each vertebra in the second model image and the vertebral sub-contour of each vertebra in the first model image; Based on the determination result that the matching parameters meet the preset matching conditions, it is determined that the two corresponding vertebral sub-contours have a corresponding relationship.

6. The method for identifying and locating a spinal model according to claim 5, characterized in that, The corresponding matching parameter is the straight-line distance between the geometric center coordinates of the vertebral contour in the second model image and the geometric center coordinates of the vertebral contour in the first model image in the world coordinate system. The matching condition is that the straight-line distance is less than or equal to a preset distance threshold, which is determined based on the average size of the vertebrae.

7. The method for identifying and locating a spinal model according to claim 5, characterized in that, Based on the coordinates of the edge calibration points of the identified reference contour within the first and second model images, pose change data is determined, including: Determine the x-coordinate, y-coordinate, and ordinate of the edge calibration points of the identification reference contour within the first and second model images, respectively. Calculate the changes in the horizontal, vertical, and y coordinates of the edge calibration points within the second model image relative to the first model image. The set of data consisting of the average changes in the horizontal coordinate, the average changes in the vertical coordinate, and the average changes in the ordinate of several edge calibration points on the reference contour is defined as pose change data.

8. The method for identifying and locating a spinal model according to claim 7, characterized in that, The process of calculating the pose change reference coordinates of edge calibration points within the second model image includes: Within the first model image, obtain the coordinates of each edge calibration point on the recognition reference contour for vertebral sub-contours with unsatisfactory recognition reliability. The x-coordinate of each edge calibration point is added to the x-coordinate change in the pose change data to determine the reference x-coordinate of the pose change of each edge calibration point; The vertical coordinates of each edge calibration point are added to the vertical coordinate change in the pose change data to determine the reference vertical coordinates of the pose change of each edge calibration point; The ordinate of each edge calibration point is added to the ordinate change in the pose change data to determine the reference ordinate of the pose change of each edge calibration point.

9. The method for identifying and locating a spinal model according to claim 8, characterized in that, The process of determining the actual coordinate differences within the second model image includes: Determine the corresponding feature vertebral sub-contours in the second model image for vertebral sub-contours whose recognition confidence in the first model image is insufficient, and obtain the actual coordinates of each edge calibration point on the feature vertebral sub-contours; Calculate the difference between the actual coordinates and the pose change reference coordinates, and determine the cumulative difference of the actual coordinates based on the average difference of the horizontal coordinate, the average difference of the vertical coordinate, and the average difference of the ordinate.

10. The method for identifying and locating a spinal model according to claim 9, characterized in that, The process of determining whether a model image is valid includes: The actual accumulated coordinate difference is compared with the preset accumulated coordinate difference threshold. If the actual coordinate difference accumulation is less than or equal to the preset coordinate difference accumulation threshold, the model image is determined to be valid. If the accumulated amount of actual coordinate differences is greater than the preset threshold for accumulated coordinate differences, the model image is determined to be invalid.

Images