An imaging-assisted assessment system for orthopedic spine surgery

Abstract

The present invention relates to the field of medical care information technology, and more specifically to an imaging-assisted assessment system for orthopedic spinal surgery. The system comprises a region acquisition module, a region correction module, a callus growth coefficient acquisition module, and a network training module. The system obtains the initial vertebral fusion region from spinal images of historical patients after spinal fusion surgery and corrects the initial vertebral fusion region to obtain the true vertebral fusion region; determines a number of first intervertebral gradient regions and second intervertebral gradient regions in the true vertebral fusion region, and determines the callus growth coefficient of the true vertebral fusion region based on the position distribution and pixel grayscale of the first intervertebral gradient region and the second intervertebral gradient region; constructs a training set using the callus growth coefficients of all historical patients and the postoperative risk classification calibrated with the initial vertebral fusion region, and trains a postoperative risk classification network. The present invention effectively improves the classification accuracy of the postoperative risk classification network.

Description

Technical Field

[0001] The present invention relates to the field of medical care information technology, and in particular to an imaging-assisted evaluation system for orthopedic spinal surgery. Background Art

[0002] When a patient's spinal pathology causes pain or functional impairment, spinal fusion surgery is often performed to fuse two or more spinal bones together to limit motion, relieve pain, restore spinal stability, or improve function. A range of complications may occur after orthopedic spinal surgery, such as loosening or displacement of metal devices within the spine, spinal fractures, and bone infections. To ensure the success of the surgery, a postoperative evaluation is often required.

[0003] In the prior art, in order to evaluate the status of a patient's spinal fusion after surgery, partial images of the patient's spine are typically collected for machine vision-assisted evaluation. Specifically, images of the vertebral fusion area in the partial images of the patient's spine are input into a classification model, which then performs classification to obtain an auxiliary evaluation result. However, in the process of training the classification model, since patients typically undergo surgery with implanted fusion devices, the constant changes in bone quality during spinal fusion can be confused with the grayscale differences in fusion produced by the fusion device in the intervertebral space. Furthermore, different fusion devices exhibit different intervertebral imaging performances, making it difficult for the classification model to accurately extract features of the intervertebral fusion area of ​​the patient's spine, resulting in poor classification accuracy. Summary of the Invention

[0004] In order to solve the technical problem that, during the machine vision-assisted evaluation of the patient's postoperative spinal images, the classification model has difficulty accurately extracting the features of the patient's intervertebral fusion area, resulting in poor classification accuracy, the present invention aims to provide an imaging-assisted evaluation system for orthopedic spine surgery. The technical solutions adopted are as follows:

[0005] In a first aspect, the present invention provides an imaging-assisted evaluation system for orthopedic spine surgery, the system comprising:

[0006] a region acquisition module for acquiring an initial vertebral fusion region in a spinal image of a historical patient after spinal fusion surgery;

[0007] a region correction module, configured to correct the initial vertebral fusion region according to the gradient and position distribution of pixel points in the low-variation region of the initial vertebral fusion region to obtain a true vertebral fusion region;

[0008] a callus growth coefficient acquisition module, configured to determine a plurality of first intervertebral gradient regions and second intervertebral gradient regions in the actual vertebral fusion region, and determine the callus growth coefficient of the actual vertebral fusion region based on the position distribution and grayscale values ​​of the pixels in the first intervertebral gradient regions and the second intervertebral gradient regions, wherein the gradient of the pixels in the first intervertebral gradient region is higher than the gradient of the pixels in the second intervertebral gradient region;

[0009] The network training module is used to form a training set using the callus growth coefficients of all historical patients and the postoperative risk classification calibrated by the initial vertebral fusion area, and to train the postoperative risk classification network using the training set to obtain a trained postoperative risk classification network.

[0010] In conjunction with the first aspect above, in some possible implementations, the region correction module includes:

[0011] an intervertebral space coefficient obtaining unit, configured to determine the intervertebral space coefficient of the low-variation area according to the gradient and position distribution of the pixel points in the low-variation area;

[0012] The region correction unit is used to screen out the gap regions in all the low-variation regions according to the intervertebral space coefficient, remove the gap regions from the initial intervertebral fusion region, and obtain a true intervertebral fusion region, wherein the intervertebral space coefficient of the gap regions is greater than a set coefficient threshold.

[0013] In combination with the first aspect above, in some possible implementations, the intervertebral space coefficient obtaining unit includes:

[0014] a pixel cluster determining unit, configured to cluster the pixels in the low-variation area according to the gradients of the pixels in the low-variation area, and divide the low-variation area into a plurality of pixel clusters;

[0015] a sequence distance determining unit, configured to determine a sequence of pixels in the pixel cluster in the direction of vertebrae arrangement, and a first distance from the pixel cluster to the initial vertebrae fusion region in the direction perpendicular to the vertebrae arrangement;

[0016] The intervertebral space coefficient determining unit is used to determine the intervertebral space coefficient of the low-variation area according to the number of pixels in the pixel sequence in the pixel cluster, the first distance corresponding to the pixel cluster, and the gradient of the pixels in the pixel cluster.

[0017] In combination with the first aspect above, in some possible implementations, the intervertebral space coefficient determining unit includes:

[0018] A first arrangement determining unit is configured to arrange all the pixel clusters in the low-variability area in ascending order of the first distances to obtain a first arrangement;

[0019] a second arrangement determining unit, configured to determine an average gradient of pixels in the pixel clusters, and arrange all the pixel clusters in the low-variance area in ascending order of the average gradient to obtain a second arrangement;

[0020] a first arrangement difference determining unit, configured to determine a total number of pixel clusters having arrangement differences between the first arrangement and the second arrangement;

[0021] A third arrangement determining unit, configured to arrange the number of pixels in the pixel sequence in the pixel cluster in ascending order to obtain a third arrangement;

[0022] a second arrangement difference determining unit, configured to determine a total number of pixel point sequences having an arrangement order that is different from an original arrangement of the pixel point sequences in the pixel point cluster;

[0023] The intervertebral space coefficient calculation unit is used to determine the slope of the increase in the number of pixels based on the third arrangement, and determine the intervertebral space coefficient of the low-variation area based on the total number of pixel clusters with arrangement differences corresponding to the low-variation area, the slope of the increase in the number of pixels and the total number of pixel sequences.

[0024] In combination with the first aspect above, in some possible implementations, the intervertebral space coefficient calculation unit includes:

[0025] a first calculation unit, configured to determine an average value of a first product of a slope of increase in the number of pixels in all pixel clusters and a total number of pixel sequences with arrangement differences, to obtain a pixel cluster change index;

[0026] The second calculation unit determines a second product of the total number of pixel clusters with arrangement differences corresponding to the low-variation area and the pixel cluster change index, and normalizes the second product to obtain the intervertebral space coefficient of the low-variation area.

[0027] In combination with the first aspect above, in some possible implementations, the region correction module further includes:

[0028] The low-variation region acquisition unit is configured to perform Otsu threshold segmentation on the grayscale values ​​of the pixels in the initial vertebral fusion region to obtain a plurality of first grayscale value regions and second grayscale value regions, and to use the second grayscale value regions as the low-variation regions of the initial vertebral fusion region, wherein the grayscale values ​​of the first grayscale value regions are higher than the grayscale values ​​of the second grayscale value regions.

[0029] In conjunction with the first aspect above, in some possible implementations, the callus growth coefficient acquisition module includes:

[0030] an information value acquiring unit, configured to determine a second distance in a direction perpendicular to the vertebral arrangement direction from the first intervertebral gradient region to the actual vertebral fusion region, and determine a distance distribution difference information value based on distribution differences of all the second distances;

[0031] a grayscale difference value acquisition unit, configured to determine a second intervertebral gradient region adjacent to the first intervertebral gradient region in a target direction, and to determine a grayscale difference value between the first intervertebral gradient region and the second intervertebral gradient region adjacent to the first intervertebral gradient region, wherein the target direction refers to a direction in which the average grayscale value of each pixel sequence in the first intervertebral gradient region in the vertebral arrangement direction gradually decreases in a direction perpendicular to the vertebral arrangement direction;

[0032] an adjacent region number acquisition unit, configured to determine the number of adjacent first intervertebral gradient regions of the adjacent second intervertebral gradient region of the first intervertebral gradient region;

[0033] The callus growth coefficient determining unit is used to determine the callus growth coefficient of the actual vertebral fusion area according to the distance distribution difference information value, the grayscale difference values ​​corresponding to all the first intervertebral gradient areas, and the number of adjacent first intervertebral gradient areas.

[0034] In combination with the first aspect above, in some possible implementations, the callus growth coefficient determination unit includes:

[0035] a third calculating unit, configured to determine an average value of a third product of the grayscale difference values ​​corresponding to all the first intervertebral gradient regions and the number of adjacent first intervertebral gradient regions, to obtain a callus growth index;

[0036] The fourth calculation unit is used to determine a fourth product of the distance distribution difference information value and the callus growth index, and normalize the fourth product to obtain the callus growth coefficient of the actual vertebral fusion area.

[0037] In combination with the first aspect above, in some possible implementations, the information value acquiring unit includes:

[0038] The information value calculation unit is configured to determine the information entropy of the second distances of all the first intervertebral gradient regions, and use the information entropy as the distance distribution difference information value.

[0039] In conjunction with the first aspect above, in some possible implementations, the callus growth coefficient acquisition module includes:

[0040] The threshold segmentation region acquisition unit is used to perform Otsu threshold segmentation on the gradient of the pixel points in the real vertebral fusion area to obtain a plurality of first intervertebral gradient regions and second intervertebral gradient regions, wherein the gradient of the first intervertebral gradient region is higher than the gradient of the second intervertebral gradient region.

[0041] In a second aspect, the present invention further provides an imaging-assisted evaluation method for orthopedic spine surgery, the method comprising the following steps:

[0042] Obtaining the initial vertebral fusion area in historical patient spinal images after spinal fusion surgery;

[0043] Correcting the initial vertebral fusion region according to the gradient and position distribution of pixel points in the low-variation region of the initial vertebral fusion region to obtain a true vertebral fusion region;

[0044] determining a plurality of first intervertebral gradient regions and second intervertebral gradient regions in the actual vertebral fusion region, and determining a callus growth coefficient of the actual vertebral fusion region based on the position distribution and grayscale values ​​of pixels in the first intervertebral gradient regions and the second intervertebral gradient regions, wherein the gradient of the pixels in the first intervertebral gradient region is higher than the gradient of the pixels in the second intervertebral gradient region;

[0045] A training set is formed using the callus growth coefficients of all historical patients and the postoperative risk classifications calibrated by the initial vertebral fusion area. The training set is used to train a postoperative risk classification network to obtain a trained postoperative risk classification network.

[0046] In a third aspect, the present invention also provides an imaging-assisted evaluation device for orthopedic spinal surgery, comprising a memory, a processor, and a computer program code stored in the memory, wherein the processor is used to call and run the computer program code from the memory so that the device executes the method in the above-mentioned first aspect or any possible implementation of the first aspect.

[0047] In a fourth aspect, the present invention further provides a computer program product, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the above-mentioned first aspect or any possible implementation of the first aspect.

[0048] In a fifth aspect, the present invention also provides a computer-readable storage medium, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the above-mentioned first aspect or any possible implementation of the first aspect.

[0049] The present invention has the following beneficial effects: by obtaining the initial vertebral fusion area in the spinal image of historical patients after spinal fusion surgery and correcting the initial vertebral fusion area, the real vertebral fusion area is obtained to avoid the influence of the intervertebral space area on the evaluation of the fusion condition of the vertebral fusion area. According to the determination of several first intervertebral gradient areas and second intervertebral gradient areas in the real vertebral fusion area, that is, high gradient areas and low gradient areas, the callus growth condition of the real vertebral fusion area is accurately evaluated according to the position distribution of the high gradient area and the low gradient area and the grayscale value of the pixel points, and the callus growth coefficient is determined; finally, the callus growth coefficient of all historical patients is used to train the postoperative risk classification network. By extracting the callus growth coefficient related to the postoperative risk in the vertebral fusion area to train the postoperative risk classification network, the postoperative risk classification network can more accurately learn the relationship between callus growth and postoperative risk, thereby effectively improving the classification accuracy of the postoperative risk classification network. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 This is a schematic structural diagram of an imaging-assisted evaluation system for orthopedic spine surgery according to an embodiment of the present invention;

[0052] Figure 2 This is a flowchart of the steps of an imaging-assisted evaluation method for orthopedic spine surgery according to an embodiment of the present invention;

[0053] Figure 3 A schematic diagram of a non-fused area in an initial vertebral fusion area according to an embodiment of the present invention;

[0054] Figure 4 Schematic diagram of the structure of an imaging-assisted evaluation device for orthopedic spine surgery according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to clearly illustrate the technical features of this solution, the present invention is described in detail below through specific implementation methods in conjunction with the accompanying drawings.

[0056] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0057] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0058] As used herein, the term "including" and its variations are open-ended, i.e., "including but not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment," the term "another embodiment" means "at least one additional embodiment," and the term "some embodiments" means "at least some embodiments." Other terms are defined in the following description.

[0059] It should be noted that the concepts of "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0060] Although operations or steps are described in a particular order in the drawings in the embodiments of the present invention, this should not be understood as requiring that these operations or steps be performed in the particular order shown or in a serial order, or that all of the operations or steps shown be performed to obtain a desired result. In the embodiments of the present invention, these operations or steps may be performed serially; they may also be performed in parallel; or a portion of these operations or steps may be performed.

[0061] At the same time, it is understood that the data involved in the technical solutions of the present invention (including but not limited to the data itself, the acquisition or use of the data) must comply with the requirements of relevant laws, regulations and relevant provisions. Unless otherwise defined, all technical and scientific terms used in this invention have the same meanings as those commonly understood by those skilled in the art to which this invention belongs, and all parameters or indicators in the formulas involved in this invention are normalized values ​​to eliminate dimension effects.

[0062] In order to solve the technical problem that in the process of machine vision-assisted evaluation of the patient's postoperative spinal images, the classification model has difficulty in accurately extracting the features of the patient's spinal intervertebral fusion area, resulting in poor classification accuracy of the classification model, an embodiment of the present invention provides an imaging-assisted evaluation system for orthopedic spine surgery. The system is essentially a software system, which is composed of modules that realize corresponding functions. The corresponding structural diagram is shown in FIG. Figure 1 The core of the system is to implement an imaging-assisted evaluation method for orthopedic spine surgery. Each module in the system corresponds to each step in the method. The corresponding flowchart of the method is shown in Figure 2 The following is a detailed introduction to each module of the system in combination with the specific steps in this method.

[0063] The region acquisition module 101 is used to acquire the initial vertebral fusion region in the spinal image of a historical patient after spinal fusion surgery.

[0064] Specifically, in order to perform imaging-assisted evaluation of orthopedic spinal surgery, X-ray equipment was used to take X-ray images of the spinal part after surgery for several historical patients who had undergone orthopedic spinal fusion surgery. When taking X-ray images, the patient needs to adjust the posture according to the shooting requirements. When the patient is standing or lying down, make sure that the patient's back is aligned with the center of the X-ray machine, and when the patient is lying on the side, make sure that the spine is in a straight state. At the same time, the equipment parameters of the X-ray equipment (such as exposure) are reasonably set to ensure that the X-ray image taken is clear and not overexposed. The X-ray image is used as the spinal image of the historical patient after spinal fusion surgery, and the taken spinal image is uploaded to the auxiliary evaluation system.

[0065] The region acquisition module 101 in the auxiliary assessment system uses a neural network to identify and segment the vertebral fusion region in spinal images of historical patients after spinal fusion surgery, thereby obtaining the initial vertebral fusion region in the spinal image. During intervertebral disc fusion surgery, patients use intervertebral fusion cages for vertebral fusion. The cages clearly appear in the images, marking the fusion site. Therefore, the neural network can be used to identify and segment the vertebral fusion region in the spinal image, thereby determining the initial vertebral fusion region in the spinal image. The initial vertebral fusion region refers to the entire fusion site between the intervertebral spaces where fusion is achieved using the intervertebral fusion cage.

[0066] The region correction module 102 is configured to correct the initial vertebral fusion region according to the gradient and position distribution of the pixel points in the low-variation region of the initial vertebral fusion region to obtain a true vertebral fusion region.

[0067] Specifically, the above images of the initial vertebral fusion area of ​​the spinal column obtained from the patient after spinal fusion surgery. Since the fusion effect between the vertebrae may not be ideal due to bone sclerosis around the implant after surgery, it is necessary to analyze the fusion status of the vertebral fusion area. However, since the intervertebral width of the vertebral fusion area is not completely filled by growth during the fusion process, there will be some unfused areas between the vertebrae in the initial vertebral fusion area obtained by the above segmentation. Figure 3 As shown, region a is a non-fused region within an initial vertebral fusion region. To prevent the non-fused region from affecting the analysis of fusion within the vertebral fusion region, the initial vertebral fusion region is modified to remove the non-fused region, thereby obtaining a true vertebral fusion region.

[0068] Furthermore, in some possible implementations, the region correction module 102 includes: an intervertebral space coefficient acquisition unit for determining an intervertebral space coefficient for the low-variance region based on the gradient and position distribution of pixels in the low-variance region; and a region correction unit for screening out intervertebral space regions from all low-variance regions based on the intervertebral space coefficient, removing the intervertebral space regions from the initial intervertebral fusion region to obtain a true intervertebral fusion region, wherein the intervertebral space coefficient of the intervertebral space region is greater than a set coefficient threshold. The region correction module 102 further includes: a low-variance region acquisition unit for performing Otsu threshold segmentation on the grayscale values ​​of the pixels in the initial intervertebral fusion region to obtain a plurality of first grayscale value regions and a second grayscale value region, wherein the second grayscale value region is defined as the low-variance region of the initial intervertebral fusion region, wherein the grayscale value of the first grayscale value region is higher than the grayscale value of the second grayscale value region.

[0069] Specifically, the grayscale values ​​of the pixels in the initial vertebral fusion region were segmented using the Otsu threshold method, and several high-threshold and low-threshold regions within the initial vertebral fusion region were obtained. These two regions are referred to as the first grayscale value region and the second grayscale value region, respectively. The low-threshold region contains several pixel clusters, and the low-threshold region of the initial vertebral fusion region includes non-fused regions and partially fused regions with low bone density. The low-threshold region is designated as a low-variance region, and the Canny edge detection algorithm is used to obtain the gradients of the pixels in each low-variance region. Because the pixel gradients and pixel positions of the non-fused regions and partially fused regions with low bone density in the low-variance region exhibit different variation characteristics, the gradient and position distribution of the pixels in the low-variance region can be analyzed to determine the intervertebral space coefficient of the low-variance region, which is used to characterize the likelihood that the corresponding low-variance region belongs to the non-fused region in the intervertebral space, thereby enabling the identification and distinction of non-fused regions and partially fused regions with low bone density in the low-variance region.

[0070] Furthermore, in some possible implementations, the above-mentioned intervertebral space coefficient acquisition unit includes: a pixel cluster determination unit, used to cluster the pixels in the low-variation area according to the gradient of the pixel points in the low-variation area, and divide the low-variation area into a plurality of pixel clusters; a sequence distance determination unit, used to determine each pixel sequence in the pixel cluster in the direction of vertebrae arrangement, and a first distance from the pixel cluster to the initial vertebrae fusion area in the direction perpendicular to the vertebrae arrangement; an intervertebral space coefficient determination unit, used to determine the intervertebral space coefficient of the low-variation area according to the number of pixels in the pixel sequence in the pixel cluster, the first distance corresponding to the pixel cluster, and the gradient of the pixel cluster.

[0071] Specifically, the pixels in each low-variance area are clustered according to their gradients, and each low-variance area is divided into several pixel clusters. The gradients of the pixels in the same pixel cluster are similar, while the gradients of the pixels in different pixel clusters are quite different.

[0072] Since the gradients of different pixel clusters in the low-variance area are different, a portion of the low-variance area may be due to some low-density callus during the intervertebral growth process. Determine each pixel sequence in the pixel clusters in the acquired low-variance area in the vertebrae arrangement direction, and determine the number of pixels in each pixel sequence. In this embodiment, since the vertebrae arrangement direction is the vertical direction, each column of pixels in the pixel clusters in the vertical direction is determined, and the number of pixels in each column of pixels is determined. At the same time, the Euclidean distance between each pixel cluster and the center of mass of the initial vertebrae fusion area in the direction perpendicular to the vertebrae arrangement direction is obtained, and this distance is recorded as the first distance. In this embodiment, the vertical direction of the vertebrae arrangement direction refers to the horizontal direction. Based on the number of pixels in each pixel sequence in the pixel clusters in each low-variance area, the first distance corresponding to the pixel clusters, and the gradient of the pixels in the pixel clusters, the intervertebral space coefficient of each low-variance area is determined.

[0073] Furthermore, in some possible implementations, the intervertebral space coefficient determining unit includes: a first arrangement determining unit, configured to arrange all the pixel clusters in the low-variation area in ascending order of the first distance to obtain a first arrangement; a second arrangement determining unit, configured to determine the average gradient of the pixels in the pixel clusters, and arrange all the pixel clusters in the low-variation area in ascending order of the average gradient to obtain a second arrangement; a first arrangement difference determining unit, configured to determine the total number of pixel clusters having arrangement differences between the first arrangement and the second arrangement; and a third arrangement determining unit. A unit is used to arrange the number of pixels in the pixel sequence in the pixel cluster in ascending order to obtain a third arrangement; a second arrangement difference determination unit is used to determine the total number of pixel sequences that have arrangement differences between the arrangement order of the pixel sequence corresponding to the third arrangement and the original arrangement of the pixel sequence in the pixel cluster; an intervertebral space coefficient calculation unit is used to determine the slope of the increase in the number of pixels according to the third arrangement, and determine the intervertebral space coefficient of the low-variation area according to the total number of pixel clusters with arrangement differences, the slope of the increase in the number of pixels, and the total number of pixel sequences corresponding to the low-variation area.

[0074] Specifically, the pixel clusters in each low-variance region are arranged in ascending order of their first distances from the centroid of the initial vertebral fusion region to obtain a first arrangement. The pixel sequences of the pixel clusters are arranged in ascending order of their pixel count to obtain a third arrangement. The rate of increase in the number of pixels in the third arrangement is determined, i.e., the slope corresponding to the number of pixels in the third arrangement is determined and recorded as the pixel count increase slope. The difference in the number of pixel sequences corresponding to the third arrangement and the number of pixel sequences in the original arrangement of the pixel clusters is obtained and recorded as the total number of pixel sequences with arrangement differences. For example, if the pixel sequences in the original arrangement of the pixel clusters are abcde and the pixel sequences corresponding to the third arrangement are abdec, then the total number of pixel sequences with arrangement differences is 3. Simultaneously, the pixel clusters in the low-variance region are arranged in ascending order of the average gradient of the pixels within the clusters to obtain a second arrangement. This second arrangement is compared with the first arrangement of the low-variance region to obtain the total number of pixel clusters with arrangement differences. For example, in the first arrangement, the pixel clusters are numbered 123456, and in the second arrangement, the pixel clusters are numbered 134256. Then, the total number of pixel clusters with arrangement differences is 3.

[0075] The intervertebral disc coefficient of the low-variance region is then determined based on the total number of pixel clusters with different arrangement, the slope of the pixel number increase, and the total number of pixel sequences with different arrangement. When both the slope of the pixel number increase and the total number of pixel sequences with different arrangement are large, it indicates that the pixel clusters in the low-variance region change more rapidly in the direction perpendicular to the vertebral arrangement and their change pattern is less consistent with normal growth. This is because faster changes indicate faster changes in the intervertebral disc width, which is less consistent with the slow changes during normal growth and the gradual change in the number of pixels in the pixel sequences of the pixel clusters during normal growth. Furthermore, a larger total number of pixel clusters with different arrangement in the low-variance region indicates that the distribution of the pixel clusters in the low-variance region is less consistent with the gradual increase in bone density. Therefore, the low-variance region is more likely to be an intervertebral disc and its corresponding intervertebral disc coefficient should be larger.

[0076] Furthermore, in some possible implementations, the above-mentioned intervertebral space coefficient calculation unit includes: a first calculation unit, used to determine the average value of the first product of the increasing slope of the number of pixels in all the pixel clusters and the total number of pixel sequences with arrangement differences, thereby obtaining a pixel cluster change index; a second calculation unit: determining the second product of the total number of pixel clusters with arrangement differences corresponding to the low-change area and the pixel cluster change index, normalizing the second product, and obtaining the intervertebral space coefficient of the low-change area.

[0077] In this embodiment, the intervertebral space coefficient of the low-variability region is determined by the following formula:

[0078] ;

[0079] Where: Indicates the The intervertebral space coefficient of the low-variability area; Indicates the The total number of pixel clusters with arrangement differences corresponding to the low-variation areas; Indicates the The pixel clustering change index of the low-change area, that is, The average value of the first product of the slope of increase in the number of pixels in all pixel clusters in the low-variation area and the total number of pixel sequences with arrangement differences; Represents a hyperbolic function, used for Perform normalization processing.

[0080] The intervertebral space coefficients of all low-variation areas can be obtained using the above method. An appropriate threshold for the coefficient is pre-set. In this embodiment, the threshold is set to 0.5. Low-variation areas corresponding to intervertebral space coefficients greater than 0.5 are defined as intervertebral space areas. All intervertebral space areas are removed from the initial fusion area, and the remaining area is the actual fusion area.

[0081] The callus growth coefficient acquisition module 103 is used to determine a plurality of first intervertebral gradient regions and second intervertebral gradient regions in the actual vertebral fusion region, and determine the callus growth coefficient of the actual vertebral fusion region based on the position distribution of the first intervertebral gradient regions and the grayscale values ​​of the pixels in the second intervertebral gradient regions, wherein the gradient of the pixels in the first intervertebral gradient region is higher than the gradient of the pixels in the second intervertebral gradient region.

[0082] Specifically, because the actual vertebral fusion area may not be fully fused and may still be interfered with by the fusion device, further analysis of the callus growth is required. Considering that when vertebrae are not successfully fused, subtle callus will appear. This subtle callus often leads to irregular vertebral edge morphology. Furthermore, since bone tissue density is lower than that of normal bone, the grayscale of this subtle callus is also relatively low. Based on this characteristic, the callus growth in the actual vertebral fusion area can be evaluated.

[0083] Furthermore, in some possible implementations, the above-mentioned callus growth coefficient acquisition module 103 includes: a threshold segmentation region acquisition unit, used to perform Otsu threshold segmentation on the gradient of the pixel points in the real vertebral fusion area to obtain a plurality of first intervertebral gradient areas and second intervertebral gradient areas, wherein the gradient of the first intervertebral gradient area is higher than the gradient of the second intervertebral gradient area; an information value acquisition unit, used to determine a second distance in the vertical direction from the first intervertebral gradient area to the vertebral arrangement direction of the real vertebral fusion area, and determine a distribution difference information value based on the distribution difference of all the second distances; a grayscale difference value acquisition unit, used to determine the adjacent second distances of the first intervertebral gradient area in the target direction. intervertebral gradient regions, and determining the grayscale difference value between the first intervertebral gradient region and its adjacent second intervertebral gradient region, the target direction refers to the direction in which the average grayscale value of each pixel point sequence in the first intervertebral gradient region in the vertebral arrangement direction gradually decreases in the vertical direction of the vertebral arrangement direction; an adjacent region number acquisition unit, used to determine the number of adjacent first intervertebral gradient regions of the adjacent second intervertebral gradient regions of the first intervertebral gradient region; a callus growth coefficient determination unit, used to determine the callus growth coefficient of the real vertebral fusion region according to the distribution difference information value, the grayscale difference values ​​corresponding to all the first intervertebral gradient regions and the number of adjacent first intervertebral gradient regions.

[0084] Specifically, the pixels in the true vertebral fusion region are segmented using their gradients using the Otsu threshold method. This segmentation method divides the true vertebral fusion region into several high-gradient regions and low-gradient regions, which are referred to as the first intervertebral gradient region and the second intervertebral gradient region, respectively. Because the callus produced by incomplete vertebral fusion is uneven, the high-gradient and low-gradient regions in the true vertebral fusion region are irregularly arranged. Therefore, by analyzing the distribution of these high-gradient and low-gradient regions, the fusion status of the vertebral fusion region can be determined.

[0085] The center of mass of the high-gradient region (i.e., the first intervertebral gradient region) and the true vertebral fusion region are obtained. The Euclidean distance between the center of mass of the first intervertebral gradient region and the center of mass of the true vertebral fusion region in a direction perpendicular to the vertebral arrangement is determined, and this Euclidean distance is recorded as the second distance. The information entropy of the second distances of all first intervertebral gradient regions is obtained, and this information entropy is recorded as the distance distribution difference information value.

[0086] Based on the arrangement of the vertebrae, pixel sequences are obtained in the first intervertebral gradient region. Specifically, pixel sequences are obtained in the first intervertebral gradient region in the direction of the vertebrae arrangement. The two directions of each pixel sequence perpendicular to the vertebrae arrangement are denoted as a first direction and a second direction. In this embodiment, since the vertebrae arrangement direction is vertical, each pixel sequence is a column of pixels in the first intervertebral gradient region in the vertical direction. The first direction and the second direction refer to the directions of the columns of pixels in the first intervertebral gradient region from left to right and from right to left, respectively.

[0087] Obtain the average grayscale value of each pixel point sequence in the first intervertebral gradient region. If the average grayscale value of each pixel point sequence gradually decreases in the first direction as a whole, the first direction is used as the target direction of the first intervertebral gradient region. If the average grayscale value of each pixel point sequence gradually decreases in the second direction as a whole, the second direction is used as the target direction of the first intervertebral gradient region.

[0088] Since callus growth in the vertebral fusion area causes bone density to gradually approach normal vertebral density, the intervertebral gradient area in the target direction of the first intervertebral gradient area should be an area with lower grayscale. The adjacent second intervertebral gradient area of ​​each first intervertebral gradient area in its target direction is obtained, and the grayscale mean difference between each first intervertebral gradient area and its adjacent second intervertebral gradient area is obtained, and the grayscale difference value is determined based on the grayscale mean difference. In this embodiment, the grayscale mean difference is used as the exponent of an exponential function with a natural constant e as the base, and the function value of the exponential function is the grayscale difference value. At the same time, the number of adjacent first intervertebral gradient areas of the adjacent second intervertebral gradient area of ​​each first intervertebral gradient area is obtained. For example, for the adjacent second intervertebral gradient area B1 of a first intervertebral gradient area A1, if the adjacent second intervertebral gradient area B1 is the adjacent second intervertebral gradient area of ​​the other three first intervertebral gradient areas, then the number of adjacent first intervertebral gradient areas corresponding to the first intervertebral gradient area A1 is 3.

[0089] Then, the callus growth coefficient of the true vertebral fusion region is determined based on the distance distribution difference information value, the grayscale difference values ​​corresponding to all first intervertebral gradient regions, and the number of adjacent first intervertebral gradient regions. A larger distance distribution difference information value, as well as a larger grayscale difference value corresponding to all first intervertebral gradient regions and the number of adjacent first intervertebral gradient regions, indicates a more dispersed distribution of first intervertebral gradient regions (i.e., high-gradient regions) in the true vertebral fusion region and a greater number of second intervertebral gradient regions (i.e., low-gradient regions) surrounding the high-gradient regions. This indicates better callus growth, resulting in a greater callus density and, therefore, a larger callus growth coefficient.

[0090] Furthermore, in some possible implementations, the above-mentioned callus growth coefficient determination unit includes: a third calculation unit, used to determine the average value of the third product of the grayscale difference values ​​corresponding to all the first intervertebral gradient regions and the number of adjacent first intervertebral gradient regions to obtain a callus growth index; a fourth calculation unit, used to determine the fourth product of the distance distribution difference information value and the callus growth index, normalize the fourth product, and obtain the callus growth coefficient of the real vertebral fusion area.

[0091] In this embodiment, the callus growth coefficient of the actual vertebral fusion area is determined by the following formula:

[0092] ;

[0093] Where: The callus growth coefficient represents the true vertebral fusion area; Indicates the distribution difference information value; represents the callus growth index, i.e., the average value of the third product of the grayscale difference values ​​corresponding to all first intervertebral gradient regions and the number of adjacent first intervertebral gradient regions; Represents a hyperbolic function, used for Perform normalization processing.

[0094] According to the above method, the callus growth coefficient of the vertebral fusion area in the spinal image of the historical patient after spinal fusion surgery can be determined. The callus growth coefficient reflects the callus growth in the vertebral fusion area of ​​the historical patient after spinal fusion surgery. The better the callus growth, the larger the corresponding callus growth coefficient value, thereby realizing the extraction of callus growth characteristics in the vertebral fusion area.

[0095] The network training module 104 is used to construct a training set using the callus growth coefficients of all historical patients and the postoperative risk classifications calibrated by the initial vertebral fusion area, and to train a postoperative risk classification network using the training set to obtain a trained postoperative risk classification network.

[0096] Specifically, a postoperative risk classification network was constructed, using an Encoder-FC architecture. Postoperative risk classification was manually assigned to the vertebral fusion region in spinal images of all patients undergoing spinal fusion surgery. The postoperative risk classification range was [0, 1], with larger values ​​indicating higher postoperative risk. A training set was constructed using the callus growth coefficients of all patients and the calibrated postoperative risk classifications of the initial vertebral fusion region. This training set was then used to train the constructed postoperative risk classification network, resulting in a trained postoperative risk classification network. During training, the input of the postoperative risk classification network was the callus growth coefficients of the patients, and the output was the postoperative risk classification. The network used a cross-entropy function as the loss function, and trained using gradient descent until the loss function converged. Training the network was completed by extracting the callus growth coefficients associated with postoperative risk in the vertebral fusion region. This allows the network to more accurately learn the relationship between callus growth and postoperative risk, effectively improving its classification accuracy.

[0097] When it is necessary to perform an imaging-assisted assessment based on the real-time patient's spinal image after spinal fusion surgery, the callus growth coefficient corresponding to the real-time patient's spinal image after spinal fusion surgery is extracted, and the extracted callus growth coefficient is input into the trained postoperative risk classification network. The postoperative risk classification network outputs the real-time patient's postoperative risk grade, and the postoperative risk grade is provided to the doctor as auxiliary assessment data for reference. Finally, the doctor determines the real-time patient's postoperative risk situation. It should be emphasized that the real-time patient's postoperative risk grade output by the postoperative risk classification network is only used as an auxiliary assessment reference data, and it does not represent the patient's final postoperative risk. The doctor needs to conduct a joint assessment based on other characteristics of the patient to determine the final postoperative risk situation.

[0098] Based on the same inventive concept, Figure 2 As shown, an embodiment of the present invention further provides an imaging-assisted evaluation method after orthopedic spine surgery, the method comprising the following steps:

[0099] Obtaining the initial vertebral fusion area in historical patient spinal images after spinal fusion surgery;

[0100] Correcting the initial vertebral fusion region according to the gradient and position distribution of pixel points in the low-variation region of the initial vertebral fusion region to obtain a true vertebral fusion region;

[0101] determining a plurality of first intervertebral gradient regions and second intervertebral gradient regions in the actual vertebral fusion region, and determining a callus growth coefficient of the actual vertebral fusion region based on the position distribution and grayscale values ​​of pixels in the first intervertebral gradient regions and the second intervertebral gradient regions, wherein the gradient of the pixels in the first intervertebral gradient region is higher than the gradient of the pixels in the second intervertebral gradient region;

[0102] A training set is formed using the callus growth coefficients of all historical patients and the postoperative risk classifications calibrated by the initial vertebral fusion area. The training set is used to train a postoperative risk classification network to obtain a trained postoperative risk classification network.

[0103] Based on the same inventive concept, the embodiment of the present invention also provides an imaging-assisted evaluation device for orthopedic spine surgery, such as Figure 4 As shown, the device includes: a memory 401, a processor 402, and a computer program code 403 stored in the memory 401 and running on the processor 402, wherein when the processor 402 executes the computer program code 403, the device can execute the module implementation steps in any of the imaging-assisted evaluation systems for orthopedic spine surgery introduced above.

[0104] In embodiments of the present invention, the functional modules of the device can be divided according to the exemplary module implementation steps in the above-mentioned system. For example, these modules can correspond to individual functional modules, or two or more functions can be integrated into a single processing module. The integrated modules can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and represents only a logical functional division. In actual implementation, other division methods may be used.

[0105] Based on the same inventive concept, an embodiment of the present invention also provides a computer program product, which includes: computer program code, which, when running on a computer, enables the computer to execute the module implementation steps in any of the imaging-assisted evaluation systems for orthopedic spine surgery described above.

[0106] Based on the same inventive concept, an embodiment of the present invention also provides a computer-readable storage medium, which stores computer program code. When the computer program code runs on a computer, the computer executes the module implementation steps in any of the imaging-assisted evaluation systems for orthopedic spine surgery described above.

[0107] It should be noted that the above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the various embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. An imaging-assisted evaluation system for orthopedic spine surgery, characterized in that: The system comprises: a region acquisition module for acquiring an initial vertebral fusion region in a spinal image of a historical patient after spinal fusion surgery; a region correction module, configured to correct the initial vertebral fusion region according to the gradient and position distribution of pixel points in the low-variation region of the initial vertebral fusion region to obtain a true vertebral fusion region; a callus growth coefficient acquisition module, configured to determine a plurality of first intervertebral gradient regions and second intervertebral gradient regions in the actual vertebral fusion region, and determine the callus growth coefficient of the actual vertebral fusion region based on the position distribution and grayscale values ​​of the pixels in the first intervertebral gradient regions and the second intervertebral gradient regions, wherein the gradient of the pixels in the first intervertebral gradient region is higher than the gradient of the pixels in the second intervertebral gradient region; a network training module, configured to form a training set using callus growth coefficients and postoperative risk classifications calibrated by initial vertebral fusion areas of all historical patients, and to train a postoperative risk classification network using the training set to obtain a trained postoperative risk classification network; The area correction module includes: an intervertebral space coefficient obtaining unit, configured to determine the intervertebral space coefficient of the low-variation area according to the gradient and position distribution of the pixel points in the low-variation area; a region correction unit, configured to screen out gap regions in all the low-variation regions based on the intervertebral space coefficient, remove the gap regions from the initial intervertebral fusion region, and obtain a true intervertebral fusion region, wherein the intervertebral space coefficient of the gap regions is greater than a set coefficient threshold; The region correction module further includes: a low-variation region acquisition unit, configured to perform Otsu threshold segmentation on the grayscale values ​​of the pixels in the initial vertebral fusion region to obtain a plurality of first grayscale value regions and second grayscale value regions, wherein the second grayscale value regions are used as low-variation regions of the initial vertebral fusion region, and the grayscale values ​​of the first grayscale value regions are higher than the grayscale values ​​of the second grayscale value regions; The callus growth coefficient acquisition module includes: a threshold segmentation region acquisition unit, configured to perform Otsu threshold segmentation on the gradient of pixel points in the real vertebral fusion region to obtain a plurality of first intervertebral gradient regions and second intervertebral gradient regions, wherein the gradient of the first intervertebral gradient region is higher than the gradient of the second intervertebral gradient region; The callus growth coefficient acquisition module includes: an information value acquiring unit, configured to determine a second distance in a direction perpendicular to the vertebral arrangement direction from the first intervertebral gradient region to the actual vertebral fusion region, and determine a distance distribution difference information value based on distribution differences of all the second distances; a grayscale difference value acquisition unit, configured to determine a second intervertebral gradient region adjacent to the first intervertebral gradient region in a target direction, and to determine a grayscale difference value between the first intervertebral gradient region and the second intervertebral gradient region adjacent to the first intervertebral gradient region, wherein the target direction refers to a direction in which the average grayscale value of each pixel sequence in the first intervertebral gradient region in the vertebral arrangement direction gradually decreases in a direction perpendicular to the vertebral arrangement direction; an adjacent region number acquisition unit, configured to determine the number of adjacent first intervertebral gradient regions of the adjacent second intervertebral gradient region of the first intervertebral gradient region; The callus growth coefficient determining unit is used to determine the callus growth coefficient of the actual vertebral fusion area according to the distance distribution difference information value, the grayscale difference values ​​corresponding to all the first intervertebral gradient areas, and the number of adjacent first intervertebral gradient areas.

2. The imaging-assisted evaluation system for orthopedic spine surgery according to claim 1, characterized in that: The intervertebral space coefficient acquisition unit includes: a pixel cluster determining unit, configured to cluster the pixels in the low-variation area according to the gradients of the pixels in the low-variation area, and divide the low-variation area into a plurality of pixel clusters; a sequence distance determining unit, configured to determine a sequence of pixels in the pixel cluster in the direction of vertebrae arrangement, and a first distance from the pixel cluster to the initial vertebrae fusion region in the direction perpendicular to the vertebrae arrangement; The intervertebral space coefficient determining unit is used to determine the intervertebral space coefficient of the low-variation area according to the number of pixels in the pixel sequence in the pixel cluster, the first distance corresponding to the pixel cluster, and the gradient of the pixels in the pixel cluster.

3. The imaging-assisted evaluation system for orthopedic spine surgery according to claim 2, characterized in that: The intervertebral space coefficient determining unit includes: A first arrangement determining unit is configured to arrange all the pixel clusters in the low-variability area in ascending order of the first distances to obtain a first arrangement; a second arrangement determining unit, configured to determine an average gradient of pixels in the pixel clusters, and arrange all the pixel clusters in the low-variance area in ascending order of the average gradient to obtain a second arrangement; a first arrangement difference determining unit, configured to determine a total number of pixel clusters having arrangement differences between the first arrangement and the second arrangement; A third arrangement determining unit, configured to arrange the number of pixels in the pixel sequence in the pixel cluster in ascending order to obtain a third arrangement; a second arrangement difference determining unit, configured to determine a total number of pixel point sequences having an arrangement order that is different from an original arrangement of the pixel point sequences in the pixel point cluster; The intervertebral space coefficient calculation unit is used to determine the slope of the increase in the number of pixels based on the third arrangement, and determine the intervertebral space coefficient of the low-variation area based on the total number of pixel clusters with arrangement differences corresponding to the low-variation area, the slope of the increase in the number of pixels and the total number of pixel sequences.

4. The imaging-assisted evaluation system for orthopedic spine surgery according to claim 3, characterized in that: The intervertebral space coefficient calculation unit includes: a first calculation unit, configured to determine an average value of a first product of a slope of increase in the number of pixels in all pixel clusters and a total number of pixel sequences with arrangement differences, to obtain a pixel cluster change index; A second calculation unit is configured to determine a second product of the total number of pixel clusters with arrangement differences corresponding to the low-variation area and the pixel cluster change index, and to normalize the second product to obtain an intervertebral space coefficient of the low-variation area; The number of adjacent intervertebral gradient areas determines the callus growth coefficient of the actual vertebral fusion area.

5. The imaging-assisted evaluation system for orthopedic spine surgery according to claim 1, characterized in that: The callus growth coefficient determining unit includes: a third calculating unit, configured to determine an average value of a third product of the grayscale difference values ​​corresponding to all the first intervertebral gradient regions and the number of adjacent first intervertebral gradient regions, to obtain a callus growth index; The fourth calculation unit is used to determine a fourth product of the distance distribution difference information value and the callus growth index, and normalize the fourth product to obtain the callus growth coefficient of the actual vertebral fusion area.

6. The imaging-assisted evaluation system for orthopedic spine surgery according to claim 1, characterized in that: The information value acquisition unit includes: The information value calculation unit is configured to determine the information entropy of the second distances of all the first intervertebral gradient regions, and use the information entropy as the distance distribution difference information value.

Images