Pedicle detection method, device and computer equipment

Abstract

The present application relates to a pedicle detection method, apparatus, computer device, and readable storage medium. The method comprises: determining the center point coordinates of each pedicle region center point in the pedicle segmentation image based on a target spine segmentation image; performing curve fitting processing on the center point coordinates of each pedicle region center point to obtain a pedicle center curve equation; obtaining the center point coordinates of missed pedicles based on the pedicle center curve equation and the center point coordinates of each pedicle region center point; and determining the pedicle detection result corresponding to the target spine based on the center point coordinates of each pedicle region center point and the center point coordinates of the missed pedicles. This method can be implemented using a computer program and can obtain the pedicle detection result corresponding to the target spine without manual intervention, thereby reducing detection errors and improving the accuracy of pedicle detection results.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular to a pedicle detection method, device and computer equipment. Background Art

[0002] When the spine rotates in a cross-section, the center point of the pedicle of the spine shifts relative to the center point of the vertebral body. Therefore, to assess the degree of vertebral rotation, the pedicle center point can be detected. The degree of vertebral rotation can be assessed by the amount of displacement of the pedicle center point relative to the vertebral body center point. Related technologies can use methods based on snake models or machine learning, combined with manual interaction, to detect the pedicle area and determine the pedicle center point.

[0003] However, the detection method in the related art may result in inaccurate detection of the pedicle center point. Summary of the Invention

[0004] Based on this, it is necessary to provide a pedicle detection method, device and computer equipment to address the above technical problems.

[0005] A pedicle detection method, the method comprising:

[0006] According to the pedicle segmentation image of the target spine, the center point coordinates of the center points of each pedicle region in the pedicle segmentation image are determined;

[0007] Perform curve fitting on the coordinates of the center point of each pedicle area to obtain the pedicle center curve equation;

[0008] According to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, the center point coordinates of the missed pedicle center points are obtained;

[0009] The pedicle detection result corresponding to the target spine is determined according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0010] In one embodiment, determining the center point coordinates of the center points of each pedicle region in the pedicle segmentation image includes:

[0011] Obtain the total number of pixels and the coordinates of each pixel in each pedicle area;

[0012] According to the total number of pixel points in each pedicle area and the coordinates of each pixel point, the center point coordinates of the center point of each pedicle area are obtained.

[0013] In one embodiment, obtaining the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region includes:

[0014] According to the first preset direction, the center point coordinates of the first missed pedicle center point are determined based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle region;

[0015] Determining the center point coordinates of the second missed pedicle center point according to the pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point according to the second preset direction; either one of the first preset direction and the second preset direction is a direction from the head end to the tail end of the target spine, and the other preset direction is a direction from the tail end to the head end of the target spine;

[0016] The center point coordinates of the missed pedicle center point are determined based on the center point coordinates of the first missed pedicle center point and the center point coordinates of the second missed pedicle center point.

[0017] In one embodiment, determining the center point coordinates of the first missed pedicle center point according to the first preset direction and the pedicle center curve equation and the center point coordinates of the center points of each pedicle region includes:

[0018] Acquire a plurality of first center point combinations according to the positional relationship between the first preset direction and each pedicle region; each first center point combination includes three adjacent center points;

[0019] Obtaining a first distance evaluation value for each first center point combination based on the center point coordinates of three adjacent center points in each first center point combination; the first distance evaluation value represents a degree of deviation in the distance between any two adjacent center points in each first center point combination;

[0020] If the first distance evaluation value is greater than the first preset threshold, it is determined that there is a first missed pedicle center point in the current first center point combination, and the center point coordinates of the first missed pedicle center point are determined based on the center point coordinates of the pedicle area center points in the current first center point combination and the pedicle center curve equation.

[0021] In one embodiment, determining the center point coordinates of the second missed pedicle center point according to the second preset direction based on the pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point includes:

[0022] Acquire multiple second center point combinations according to the second preset direction, the positional relationship between each pedicle region, and the positional relationship between the first missed pedicle center points; each second center point combination includes three adjacent center points among the center points of all pedicle regions and all the first missed pedicle center points;

[0023] Obtaining a second distance evaluation value for each second center point combination based on the center point coordinates of three adjacent center points in each second center point combination; the second distance evaluation value represents a degree of deviation in the distance between any two adjacent center points in each second center point combination;

[0024] If the second distance evaluation value is greater than the second preset threshold, it is determined that there is a second missed pedicle center point in the current second center point combination, and the center point coordinates of the second missed pedicle center point are determined based on the center point coordinates of the pedicle area center points in the current second center point combination and the pedicle center curve equation.

[0025] In one embodiment, obtaining a first distance evaluation value of each first center point combination according to center point coordinates of three adjacent center points in each first center point combination includes:

[0026] For each first center point combination, obtaining, based on the coordinates of three adjacent center points in the first center point combination, a first distance between first two adjacent center points in the three adjacent center points, and a second distance between last two adjacent center points in the three adjacent center points;

[0027] The ratio between the first distance and the second distance is determined as a first distance evaluation value of the first center point combination.

[0028] In one embodiment, if the first distance evaluation value is greater than a first preset threshold, determining that there is a missed pedicle center point in the current first center point combination includes:

[0029] If the first distance evaluation value is greater than a first preset threshold, it is determined that a missed pedicle center point exists between the first two adjacent center points of the three adjacent center points of the current first center point combination.

[0030] In one embodiment, determining the center point coordinates of the first missed pedicle center point according to the center point coordinates of the pedicle region center point in the current first center point combination and the pedicle center curve equation includes:

[0031] Determine the average of the first coordinates of the first two adjacent center points in the current first center point combination as the first coordinate of the first missed pedicle center point;

[0032] Substituting the average of the first coordinates of the first two adjacent center points in the current first center point combination into the pedicle center curve equation to obtain the second coordinate of the first missed pedicle center point, and determining the center point coordinate of the first missed pedicle center point through the first coordinate and the second coordinate of the first missed pedicle center point;

[0033] Among them, one of the first coordinate and the second coordinate is the horizontal coordinate, and the other is the vertical coordinate.

[0034] In one embodiment, the method further includes:

[0035] obtaining medical images of the target spine;

[0036] The pedicle segmentation model is used to segment the pedicles in the medical image of the target spine to obtain a pedicle segmentation image.

[0037] A pedicle detection device, comprising:

[0038] A first coordinate determination module is used to determine the center point coordinates of the center points of each pedicle region in the pedicle segmentation image according to the pedicle segmentation image of the target spine;

[0039] A fitting processing module is used to perform curve fitting processing on the center point coordinates of the center points of each pedicle area to obtain the pedicle center curve equation;

[0040] The second coordinate determination module is used to obtain the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area;

[0041] The detection result determination module is used to determine the pedicle detection result corresponding to the target spine according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0042] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed:

[0043] According to the pedicle segmentation image of the target spine, the center point coordinates of the center points of each pedicle region in the pedicle segmentation image are determined;

[0044] Perform curve fitting on the coordinates of the center point of each pedicle area to obtain the pedicle center curve equation;

[0045] According to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, the center point coordinates of the missed pedicle center points are obtained;

[0046] The pedicle detection result corresponding to the target spine is determined according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0047] The above-mentioned pedicle detection method, device and computer equipment include: determining the center point coordinates of the center points of each pedicle area in the pedicle segmentation image based on the pedicle segmentation image of the target spine, performing curve fitting processing on the center point coordinates of the center points of each pedicle area to obtain the pedicle center curve equation, obtaining the center point coordinates of the missed pedicle center points based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, and determining the pedicle detection result corresponding to the target spine based on the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points; this method can be implemented through a set of computer programs, and the pedicle detection result corresponding to the target spine can be obtained without human intervention, thereby reducing detection errors and improving the accuracy of pedicle detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A diagram showing an application environment of a pedicle detection method in one embodiment;

[0049] Figure 2 Schematic diagram of a pedicle detection method according to an embodiment;

[0050] Figure 3 FIG1 is a flow chart of a step of determining the center point coordinates of the center points of each pedicle region in a pedicle segmentation image in one embodiment;

[0051] Figure 4 A schematic flow chart of a method for obtaining the coordinates of the center points of missed pedicles in another embodiment;

[0052] Figure 5 is a schematic diagram of a pedicle segmentation image in another embodiment;

[0053] Figure 6 for Figure 5 Schematic diagram of pedicle segmentation image corresponding to the embodiment;

[0054] Figure 7 A schematic flow chart of a method for obtaining the coordinates of the center point of a first missed pedicle in another embodiment;

[0055] Figure 8 This is a flow chart of a method for obtaining the coordinates of the center point of a second missed pedicle in another embodiment;

[0056] Figure 9 A schematic flow chart of a method for obtaining first distance evaluation values ​​for each first center point combination in another embodiment;

[0057] Figure 10 A schematic flow chart of a method for determining the coordinates of the center point of a first missed pedicle in another embodiment;

[0058] Figure 11 for Figure 5 Schematic diagram of pedicle segmentation image corresponding to the embodiment;

[0059] Figure 12 for Figure 11 Schematic diagram of the center points of the pedicle area and all missed pedicle centers corresponding to the embodiment;

[0060] Figure 13 A schematic flow chart of a method for obtaining a pedicle segmentation image in another embodiment;

[0061] Figure 14 2 is a schematic structural diagram of a UNet-based segmentation model in another embodiment;

[0062] Figure 15 is a structural block diagram of a pedicle detection device in one embodiment;

[0063] Figure 16 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0065] The pedicle detection method provided in this application can be applied to Figure 1 The pedicle detection system shown can be applied to scenarios where the degree of vertebral rotation of an imaging subject is evaluated. The pedicle detection system includes a medical scanning device and a computer device. The computer device and the medical scanning device can be connected for communication, and the communication method can be Wi-Fi, mobile network or Bluetooth connection, etc. The medical scanning device can be a computer X-ray system or a direct digital X-ray system, etc., and can also be other medical photography systems that can capture X-ray images; the computer device can be various personal computers, laptops, smart phones, tablet computers and portable wearable devices, but are not limited to these. The pedicle detection method can perform a series of analyses on the pedicle segmentation image of the target spine of the imaging subject to determine the pedicle detection results corresponding to the target spine. Furthermore, the offset of the pedicles in the spine relative to the vertebral center can be analyzed based on the pedicle detection results corresponding to the target spine, thereby obtaining the degree of vertebral rotation.

[0066] In one embodiment, Figure 2 As shown, a pedicle detection method is provided, which is applied to Figure 1 The computer device in the example is used to illustrate the process, including the following steps:

[0067] S100 , determining the center point coordinates of the center points of each pedicle region in the pedicle segmentation image according to the pedicle segmentation image of the target spine.

[0068] Specifically, the spine is a bone structure composed of vertebral bodies, pedicles, lamina, transverse processes, articular processes, and spinous processes. Generally, the spine can be divided into five parts: cervical vertebrae, thoracic vertebrae, lumbar vertebrae, sacral vertebrae, and coccygeal vertebrae. The target spine can be at least one of the five parts of the spine of the imaging subject. The pedicle segmentation image of the target spine includes images of the regions where the multiple pedicles of the target spine are located.

[0069] It is understood that the computer device can obtain a pedicle segmentation image of the target spine of the imaging subject and perform arithmetic operations, analysis, comparison, and / or coordinate conversion on the pedicle segmentation image to obtain the center point coordinates of the center points of each pedicle region in the pedicle segmentation image. Optionally, the arithmetic operations can be addition, subtraction, division, and / or multiplication operations; the analysis can be understood as the process of extracting significant feature points of the pedicle region image in the pedicle segmentation image; the comparison can be understood as the process of comparing the pedicle segmentation image with a standard pedicle image of the imaging subject; and the coordinate conversion can be understood as the process of adding an offset to the coordinates of each point on the edge of the pedicle region in the pedicle segmentation image to obtain the center point coordinates of the pedicle region center point, wherein the offset can be a two-dimensional coordinate, and the offset can be a coordinate (x', y'). For example, if the coordinates of a point on the edge of the pedicle region are (x, y), then the coordinates of the center point of the pedicle region obtained by performing coordinate transformation processing on the point are (x+x', y+y').

[0070] The segmented image of the pedicles of the target spine can be an image generated directly by a computer device based on data scanned by a medical scanning device, or can be an image generated by a computer device after preprocessing the image generated by the medical scanning device. The preprocessing can include denoising, data conversion, cropping, and / or analysis. Optionally, there can be multiple pedicle region center points, and the total number of pedicle region center points can be determined based on the total number of pedicles in the target spine.

[0071] S200 , performing curve fitting processing on the center point coordinates of the center points of each pedicle region to obtain a pedicle center curve equation.

[0072] Specifically, the computer device can use a curve fitting algorithm to perform curve fitting on the center point coordinates of all pedicle region centers of the target spine to obtain the pedicle center curve equation. Optionally, there can be one or two pedicle center curves. Optionally, the curve fitting algorithm can be a method of approximating discrete data using an analytical expression, a least squares method, etc. Of course, other curve fitting methods are also possible, and this embodiment does not limit this.

[0073] In this embodiment, the computer device can perform curve fitting according to the following formula (1) to obtain the pedicle center curve equation Y, that is:

[0074]

[0075] Where N represents the order of the polynomial, K represents the polynomial parameter to be determined, and X and Y represent the abscissa and ordinate of the point on the fitted curve, respectively. N can be determined based on the overall distribution of each pedicle in the pedicle segmentation image and is an empirical value. Preferably, in practical applications, N can be 3 or 5. In this case, the determined pedicle center curve can better fit the overall shape of the center point of each pedicle region.

[0076] S300 , obtaining the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region.

[0077] Specifically, the computer device can calculate the average value of the center point coordinates of the pedicle region center points, i.e., the average value coordinate, using the center point coordinates of all or some of the pedicle region center points, and then substitute the abscissa or ordinate of the average value coordinate into the pedicle center curve equation to obtain the ordinate or abscissa, and then combine the ordinate or abscissa with the abscissa or ordinate in the average value coordinate to obtain the center point coordinates of the missed pedicle center points. Optionally, the computer device can also calculate the median value of the center point coordinates of the pedicle region center points, i.e., the median coordinate, using the center point coordinates of all or some of the pedicle region center points, and then substitute the abscissa or ordinate of the median coordinate into the pedicle center curve equation to obtain the ordinate or abscissa, and then combine the ordinate or abscissa with the abscissa or ordinate in the median coordinate to obtain the center point coordinates of the missed pedicle center points. Alternatively, the computer device may also perform coordinate conversion on the center point coordinates of all or part of the pedicle area center points to obtain the converted coordinates, then average the converted coordinates to obtain the average value coordinates, then substitute the abscissa or ordinate of the average value coordinates into the pedicle center curve equation to obtain the ordinate or abscissa, and then combine the ordinate or abscissa with the abscissa or ordinate in the average value coordinates to obtain the center point coordinates of the missed pedicle center points; of course, the center point coordinates of the missed pedicle center points may also be obtained by other methods.

[0078] S400 , determining a pedicle detection result corresponding to the target spine according to the center point coordinates of the center points of each pedicle region and the center point coordinates of the missed pedicle center points.

[0079] Specifically, the computer device may combine the center point coordinates of the center points of each pedicle region and the center point coordinates of the missed pedicle center points to obtain the pedicle detection result corresponding to the target spine. In other words, the pedicle detection result may be the center point coordinates of the center points of all pedicle regions and the missed pedicle center points.

[0080] In addition, during the pedicle detection process, if the center point coordinates of the missed pedicle center points are not detected, the center point coordinates of the center points of each pedicle region can be directly used as the pedicle detection results corresponding to the target spine.

[0081] In the above-mentioned pedicle detection method, the computer equipment can determine the center point coordinates of the center points of each pedicle region in the pedicle segmentation image based on the pedicle segmentation image of the target spine, perform curve fitting processing on the center point coordinates of the center points of the pedicle region to obtain the pedicle center curve equation, obtain the center point coordinates of the missed pedicle based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle region, and determine the pedicle detection result corresponding to the target spine based on the center point coordinates of the center points of each pedicle region and the center point coordinates of the missed pedicle center points; this method can be implemented by a set of computer programs and does not require human intervention The pedicle detection results corresponding to the target spine can be obtained, thereby reducing detection errors and improving the accuracy of pedicle detection results. The detection method can also save human resources and improve detection efficiency. Further, medical staff can accurately evaluate the degree of vertebral rotation based on the obtained pedicle detection results, avoiding the problem that medical staff cannot adopt the best diagnosis and treatment plan to treat the scoliosis of the imaging subject in time; at the same time, the method can also obtain the coordinates of the center points of missed pedicles to avoid the situation where the center points of the pedicle area are missed, thereby further improving the accuracy of the pedicle detection results corresponding to the target spine.

[0082] As one example, Figure 3 As shown, the step of determining the center point coordinates of the center points of the pedicle regions in the pedicle segmentation image in S100 can be achieved by the following steps:

[0083] S110 , obtaining the total number of pixels in each pedicle region and the coordinates of each pixel.

[0084] Specifically, the computer device can count the total number of pixels contained in each pedicle region in the pedicle segmentation image of the target spine. At the same time, it can also establish a two-dimensional rectangular coordinate system xoy in the pedicle segmentation image and determine the coordinates of each pixel contained in each pedicle region. Optionally, the x-axis and y-axis in the two-dimensional rectangular coordinate system xoy can also be any two vertical coordinate axes established with any pixel in the pedicle segmentation image as the position of the origin o. For example, the pixel corresponding to any corner point in the pedicle segmentation image is used as the position of the origin o, and the two vertical edge segments adjacent to the origin o in the pedicle segmentation image are respectively used as the x-axis and y-axis in the two-dimensional rectangular coordinate system xoy. Wherein, if the two-dimensional rectangular coordinate systems xoy established are different, the coordinates of each pixel in each pedicle region will also be different.

[0085] It should be noted that in the pedicle segmentation image of the target spinal column of the imaging subject, the total number of pixels contained in different pedicle regions may be equal or unequal. Optionally, the coordinates of the different pixels contained in each pedicle region in the pedicle segmentation image are different. The coordinates of the aforementioned pixel points can be understood as the specific location of the pixel point in the corresponding pedicle region in the pedicle segmentation image; in this embodiment, the coordinates of the pixel points can be two-dimensional coordinates.

[0086] S120 , obtaining the center point coordinates of the center point of each pedicle region according to the total number of pixel points in each pedicle region and the coordinates of each pixel point.

[0087] Specifically, the computer device can perform arithmetic operations and / or coordinate conversions based on the total number of pixels in each pedicle region and the coordinates of each pixel to obtain the center point coordinates of each pedicle region in the pedicle segmentation image of the target spine of the imaging subject. Optionally, the arithmetic operations can include addition, subtraction, division, multiplication, logarithmic operations, and / or exponential operations.

[0088] In this embodiment, the computer device can calculate the center point coordinates (x, y) of the center point of each pedicle region according to the following formula (2), such as:

[0089]

[0090] Among them, x k Indicates the horizontal coordinate of the pixel point in the pedicle area, y k represents the vertical coordinate of the pixel point in the pedicle area, n represents the total number of all pixels in each pedicle area, k represents the position index of the pixel point in the pedicle area, x represents the horizontal coordinate of the center point of the pedicle area, and y represents the vertical coordinate of the center point of the pedicle area.

[0091] The above-mentioned pedicle detection method can obtain the total number of pixel points in each pedicle area and the coordinates of each pixel point, and obtain the center point coordinates of the center point of each pedicle area based on the total number of pixel points in each pedicle area and the coordinates of each pixel point; this method can be implemented by a set of computer programs, and does not require human intervention to determine the center point coordinates of the center point of each pedicle area, thereby reducing detection errors, improving the accuracy of detection results, and also saving human resources and improving detection efficiency.

[0092] As one example, Figure 4 As shown, the step of obtaining the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region in S300 can be achieved by the following steps:

[0093] S310 , determining the center point coordinates of the first missed pedicle center point according to the first preset direction, the pedicle center curve equation, and the center point coordinates of the center points of each pedicle region.

[0094] The center point coordinates of the center points of the pedicle regions include the center point coordinates of the center point of the pedicle region on the left side of the target spine and the center point coordinates of the center point of the pedicle region on the right side of the target spine.

[0095] Specifically, the pedicles in the spine are generally divided into left and right pedicles. Therefore, each pedicle region in the pedicle segmentation image of the target spine includes a left pedicle region and a right pedicle region. Correspondingly, the center point coordinates of the center point of each pedicle region can include the center point coordinates of the center point of the pedicle region on the left side of the target spine and the center point coordinates of the center point of the pedicle region on the right side of the target spine. In this embodiment, the pedicle segmentation image of the target spine includes images of at least two pedicle regions, and the images of the at least two pedicle regions include an image of at least one left pedicle region and a corresponding right pedicle region.

[0096] It should be noted that the first preset direction can be determined according to the distribution direction of each pedicle region in the pedicle segmentation image. Optionally, the distribution direction of each pedicle region in the pedicle segmentation image can be understood as the direction from the head to the tail of the target spine. Figure 5 The figure shows a pedicle segmentation image of a target spine of an imaging subject. The "S"-shaped and wide area in the pedicle segmentation image represents the target spine region. Multiple irregular black small areas are displayed on the left and right sides of the target spine region, respectively. These represent the center points of different pedicle regions in the target spine. Figure 5 The target spines in are all the spines of the imaging object. Figure 5 The distribution direction of each pedicle area is from top to bottom or from bottom to top, which is determined according to the head and tail of the target spine. In this case, the first preset direction can be from top to bottom or from bottom to top. Figure 5 Example, Figure 6 shows a pedicle segmentation image, and Figure 6 and Figure 5 The pedicle segmentation images in are the same as those in Figure 6 The left pedicle center curve formed by fitting the center points of the left pedicle region and the right pedicle center curve formed by fitting the center points of the right pedicle region are marked in the middle.

[0097] In addition, if the distribution direction of each pedicle region in the pedicle segmentation image is from left to right or from right to left, specifically determined based on the head and tail of the target spine, in this case, the first preset direction can be determined as the direction from left to right or from right to left. In addition, the distribution direction and the first preset direction can also be other directions, which are not limited to this embodiment. Optionally, the distribution direction of each pedicle region in the pedicle segmentation image can also be determined by direction, in addition to being determined by up, down, left, and right directions. Such as east, south, west, north, southeast, northeast, southwest, northwest, etc.

[0098] In this embodiment, the first preset direction may be a direction from the head end to the tail end of the target spine or a direction from the tail end to the head end of the target spine.

[0099] It is understandable that the computer device can perform arithmetic operations, analysis, comparison and / or coordinate conversion on the center point coordinates of the center points of each pedicle area in accordance with the first preset direction to obtain the processed center point coordinates, and determine the center point coordinates of the first missed pedicle center point through the processed center point coordinates and the pedicle center curve equation.

[0100] S320: Determine the center point coordinates of the second missed pedicle center point according to the second preset direction based on the pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point. Wherein, either the first preset direction or the second preset direction is a direction from the cranial end to the caudal end of the target spine, and the other preset direction is a direction from the caudal end to the cranial end of the target spine.

[0101] Specifically, the second preset direction may be opposite to the first preset direction. The second preset direction may also be a direction from the left side to the right side of the target spine, or from the right side to the left side of the target spine. However, in this embodiment, if the first preset direction is a direction from the head end to the tail end of the target spine, the second preset direction is a direction from the tail end to the head end of the target spine; if the first preset direction is a direction from the tail end to the head end of the target spine, the second preset direction is a direction from the head end to the tail end of the target spine.

[0102] It should be noted that the center point coordinates of the center points of each pedicle area are combined with the center point coordinates of the first missed pedicle center point that has been determined. Further, the computer device can perform arithmetic operations, analysis, comparison and / or coordinate conversion on the center point coordinates of the center points of each pedicle area and the center point coordinates of the first missed pedicle center point in accordance with a second preset direction to obtain the processed center point coordinates, and determine the center point coordinates of the second missed pedicle center point through the processed center point coordinates and the pedicle center curve equation.

[0103] S330 : Determine the center point coordinates of the missed pedicle center point based on the center point coordinates of the first missed pedicle center point and the center point coordinates of the second missed pedicle center point.

[0104] In this embodiment, there are overlapping coordinates in the determined center point coordinates of the first missed pedicle center point and the center point coordinates of the second missed pedicle center point. Therefore, the computer device can retain only one of the overlapping coordinates in the center point coordinates of the first missed pedicle center point and the second missed pedicle center point, delete one of the overlapping coordinates, and determine all the remaining center point coordinates as the center point coordinates of the missed pedicle center point.

[0105] For example, the center point coordinates of the first missed pedicle center point determined according to the first preset direction are: coordinate 1 (1, 2), coordinate 2 (2, 3), coordinate 3 (1, 2), and the center point coordinates of the second missed pedicle center point determined according to the second preset direction are: coordinate 4 (2, 3), coordinate 5 (3, 2.4), coordinate 6 (2, 3.5), among which coordinate 1 and coordinate 3 overlap, and coordinate 2 and coordinate 4 overlap. Therefore, only coordinate 1 or coordinate 3, and coordinate 2 or coordinate 4 can be retained, and the center point coordinates of all missed pedicle center points finally obtained are: coordinate 1 or coordinate 3, coordinate 2 or coordinate 4, coordinate 5, coordinate 6.

[0106] The above-mentioned pedicle detection method can obtain the coordinates of the center points of missed pedicles based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, thereby avoiding the situation where the center points of the pedicle area are missed during the pedicle detection process, and further improving the accuracy of the pedicle detection results corresponding to the target spine.

[0107] The above steps S310 and S320 will be described in detail below.

[0108] In one embodiment, if Figure 7 As shown, the step of determining the center point coordinates of the first missed pedicle center point in the above S310 according to the first preset direction, based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle region, may include:

[0109] S311. Acquire multiple first center point combinations according to the positional relationship between the first preset direction and each pedicle region; each first center point combination includes three adjacent center points.

[0110] Specifically, the positional relationship of each pedicle region can be understood as the positional arrangement order of each pedicle region determined according to the distribution direction of each pedicle region. In this embodiment, the pedicle segmentation image is a labeled segmentation image. Therefore, the computer device can determine each left pedicle region and each right pedicle region in the pedicle segmentation image, and then further obtain the center point coordinates of the center point of each left pedicle region and the center point coordinates of the center point of each right pedicle region in the pedicle segmentation image. Based on the center point coordinates of each left pedicle region and the center point coordinates of each right pedicle region, the left pedicle region and the right pedicle region in the pedicle segmentation image are simultaneously detected, ultimately obtaining the left pedicle detection result and the right pedicle detection result.

[0111] Exemplarily, the target spinal region in a pedicle segmentation image includes left pedicle region 1, left pedicle region 2, left pedicle region 3, left pedicle region 4, and left pedicle region 5. If the distribution direction of each left pedicle region in the pedicle segmentation image is from top to bottom, left pedicle region 1 is the topmost pedicle region, and they are sorted in sequence from top to bottom, adjacent to left pedicle region 1 is left pedicle region 3, adjacent to left pedicle region 3 is left pedicle region 4, adjacent to left pedicle region 4 is left pedicle region 2, and adjacent to left pedicle region 2 is left pedicle region 5, that is, the position arrangement order of each left pedicle region in the pedicle segmentation image is left pedicle region 1, left pedicle region 3, left pedicle region 4, left pedicle region 2, and left pedicle region 5. In this embodiment, if the left pedicle regions in the pedicle segmentation image correspond to other distribution directions, the positional relationship of the right pedicle regions is similar, which will not be described in detail.

[0112] It can also be understood that the computer device can obtain multiple first center point combinations based on the positional relationship between the first preset direction and each pedicle area; each first center point combination includes the center points of three adjacent pedicle areas.

[0113] Continue to see Figure 5For example, according to the distribution direction from top to bottom, the positional relationship of each right pedicle region is left pedicle region 1, left pedicle region 3, left pedicle region 4, left pedicle region 2, and left pedicle region 5. The computer device can obtain all three adjacent left pedicle region combinations according to the positional relationship, namely region combination 1: left pedicle region 1, left pedicle region 3, and left pedicle region 4, region combination 2: left pedicle region 3, left pedicle region 4, and left pedicle region 2, and region combination 3: left pedicle region 4, left pedicle region 2, and left pedicle region 5. Further, the corresponding first center point combination is obtained based on the region combination. Optionally, the first center point combination can be a combination of the center points corresponding to each region in each region combination. For example, the first center point combination 1 corresponding to regional combination 1 is: left pedicle region center point 1, left pedicle region center point 3, and left pedicle region center point 4; the first center point combination 2 corresponding to regional combination 2 is: left pedicle region center point 3, left pedicle region center point 4, and left pedicle region center point 2; the first center point combination 3 corresponding to regional combination 3 is: left pedicle region center point 4, left pedicle region center point 2, and left pedicle region center point 5.

[0114] S312. Obtain a first distance evaluation value for each first center point combination based on the center point coordinates of three adjacent center points in each first center point combination; the first distance evaluation value represents the degree of deviation of the distance between two adjacent pedicle region center points in each first center point combination.

[0115] Specifically, the computer device may perform screening, coordinate conversion, comparison and / or arithmetic operations on the center point coordinates of each pedicle region center point in each first center point combination to obtain a distance evaluation value of each first center point combination.

[0116] Alternatively, the computer device may further perform processing such as screening, coordinate conversion, comparison, and / or arithmetic operations on the center point coordinates of the two adjacent pedicle region center points in each first center point combination, and then perform arithmetic operations on the processing results corresponding to the two adjacent pedicle region center points to obtain the first distance evaluation value for each first center point combination. The above screening process can be understood as the process of screening the center point coordinates of some pedicle region center points from the center point coordinates of multiple pedicle region center points. The above comparison process can be understood as the process of comparing the center point coordinates of the pedicle region center points with a threshold value to determine the first distance evaluation value corresponding to the first center point combination based on the comparison result.

[0117] S313. If the first distance evaluation value is greater than the first preset threshold, it is determined that there is a first missed pedicle center point in the current first center point combination, and the center point coordinates of the first missed pedicle center point are determined based on the center point coordinates of the pedicle area center points in the current first center point combination and the pedicle center curve equation.

[0118] It is understood that the computer device can determine whether the first distance evaluation value of the current first center point combination is greater than a first preset threshold. If it is determined that the first distance evaluation value of the current first center point combination is greater than the first preset threshold, it is determined that a first missed pedicle center point exists in the current first center point combination, that is, a first missed pedicle region exists between the three adjacent pedicle regions included in the current first center point combination. Optionally, the center point coordinates of the first missed pedicle center point can include the center point coordinates of the first missed pedicle center point on the left side and the center point coordinates of the first missed pedicle center point on the right side of the pedicle region.

[0119] Optionally, the first missed pedicle center point may be all missed pedicle center points corresponding to the pedicle region, or may be some missed pedicle center points corresponding to the pedicle region. If the first missed pedicle center point is all missed pedicle center points corresponding to the pedicle region, then the second missed pedicle center point actually does not exist; if the first missed pedicle center point is some missed pedicle center points corresponding to the pedicle region, then the second missed pedicle center point exists, and all first missed pedicle center points and all second missed pedicle center points are determined as all missed pedicle center points corresponding to the pedicle region.

[0120] In this embodiment, the pedicle center curve equations may be a left pedicle center curve equation and a right pedicle center curve equation. The left pedicle center curve equation is determined by the center point coordinates of the center point of the left pedicle region in the pedicle segmentation image, and the right pedicle center curve equation is determined by the center point coordinates of the center point of the right pedicle region in the pedicle segmentation image. The computer device may obtain the first left center point combination and the first right center point combination, respectively, and then determine the center point coordinates of the first missed pedicle center point on the left and the first missed pedicle center point on the right using the left pedicle center curve equation, the right pedicle center curve equation, the first left center point combination, and the first right center point combination.

[0121] The above-mentioned pedicle detection method can obtain the coordinates of the center point of the first missed pedicle based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, thereby avoiding the situation where the center point of the pedicle area is missed during the pedicle detection process, thereby further improving the accuracy of the pedicle detection results corresponding to the target spine.

[0122] In another embodiment, Figure 8 As shown, the step of determining the center point coordinates of the second missed pedicle center point in the above S320 according to the second preset direction based on the pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point may include:

[0123] S321: Acquire multiple second center point combinations based on the second preset direction, the positional relationship between each pedicle region, and the positional relationship between the first missed pedicle region, wherein each second center point combination includes three adjacent center points among the center points of all pedicle regions and the center point of the first missed pedicle region.

[0124] Specifically, the positional relationships of the first missed pedicle regions can be understood as a positional arrangement order of the first missed pedicle regions determined based on the distribution directions of the first missed pedicle regions. The computer device can comprehensively arrange the positions of all pedicle region center points and all first missed pedicle center points based on the positional relationships of the pedicle regions and the positional relationships of the first missed pedicle regions. Furthermore, the computer device can divide all pedicle region center points and all first missed pedicle center points into a plurality of second center point combinations based on a second preset direction and the comprehensive arrangement order.

[0125] It should be noted that each second center point combination may include a pedicle region center point and a first missed pedicle center point, or may include only three pedicle region center points.

[0126] Among them, the specific division process of step S321 is similar to that of the above step S311, and will not be repeated here.

[0127] S322. Obtain a second distance evaluation value for each second center point combination based on the center point coordinates of three adjacent center points in each second center point combination; the second distance evaluation value represents the degree of deviation of the distance between any two adjacent center points in each second center point combination.

[0128] Specifically, the computer device can perform screening, coordinate conversion, comparison and / or arithmetic operations on the center point coordinates of each pedicle region center point in each second center point combination to obtain the second distance evaluation value of each second center point combination.

[0129] Alternatively, the computer device may also perform screening, coordinate conversion, comparison and / or arithmetic operations on the center point coordinates of the two adjacent pedicle area center points in each second center point combination, and then perform arithmetic operations on the processing results corresponding to the two adjacent pedicle area center points to obtain the second distance evaluation value of each second center point combination.

[0130] S323. If the second distance evaluation value is greater than the second preset threshold, determine that there is a second missed pedicle center point in the current second center point combination, and determine the center point coordinates of the second missed pedicle center point based on the center point coordinates of the pedicle area center points in the current second center point combination and the pedicle center curve equation.

[0131] It is understandable that the computer device can determine whether the second distance evaluation value of the current second center point combination is greater than the second preset threshold value. If it is determined that the second distance evaluation value of the current second center point combination is greater than the second preset threshold value, it is determined that a second missed pedicle center point exists in the current second center point combination, that is, a second missed pedicle region exists between the three adjacent pedicle regions included in the current second center point combination. Optionally, the center point coordinates of the second missed pedicle center point may include the center point coordinates of the corresponding left second missed pedicle center point and the center point coordinates of the right second missed pedicle center point in the pedicle region and the first missed pedicle region. Optionally, the second preset threshold value may be equal to or different from the first preset threshold value, and this embodiment does not limit this.

[0132] In this embodiment, the computer device can obtain the second center point combination on the left and the second center point combination on the right respectively, and then determine the center point coordinates of the second missed pedicle center point on the left and the center point coordinates of the second missed pedicle center point on the right through the left pedicle center curve equation, the right pedicle center curve equation, the left second center point combination and the right second center point combination.

[0133] The above-mentioned pedicle detection method can obtain the coordinates of the second missed pedicle center point based on the pedicle center curve equation, the center point coordinates of the center points of each pedicle area and the center point coordinates of the first missed pedicle center point, thereby avoiding the situation where the center point of the pedicle area is missed during the pedicle detection process, thereby further improving the accuracy of the pedicle detection results corresponding to the target spine.

[0134] As one example, Figure 9 As shown, the step of obtaining the first distance evaluation value of each first center point combination according to the center point coordinates of three adjacent center points in each first center point combination in S312 can be implemented by the following steps:

[0135] S3121. For each first center point combination, according to the coordinates of the three adjacent center points in the first center point combination, obtain a first distance between the first two adjacent center points in the three adjacent center points, and a second distance between the last two adjacent center points in the three adjacent center points.

[0136] Specifically, for each first center point combination, the computer device may calculate the distances between the center points of each pair of adjacent pedicle regions in each first center point combination using a Euclidean distance method or a trigonometric function method based on the center point coordinates of the three adjacent center points in each first center point combination, thereby obtaining a first distance L1 between the first two adjacent center points of the three adjacent center points in each first center point combination, and a second distance L2 between the last two adjacent center points of the three adjacent center points. Optionally, the Euclidean distance method may be a Manhattan distance calculation method, a Euclidean distance calculation method, a cosine similarity calculation method, or the like.

[0137] S3122: Determine the ratio between the first distance and the second distance as a first distance evaluation value of the first center point combination.

[0138] Specifically, the computer device may perform an arithmetic operation on the first distance L1 and the second distance L2 corresponding to each first center point combination to obtain a first distance evaluation value for each first center point combination. Optionally, the arithmetic operation may be an addition operation, a subtraction operation, a division operation, a multiplication operation, an exponential operation, and / or a logarithmic operation, etc. However, in this embodiment, the computer device may calculate the ratio between the first distance L1 and the second distance L2 corresponding to each first center point combination (i.e., L1 / L2) to obtain the first distance evaluation value for each first center point combination.

[0139] Further, in the above S313, if the first distance evaluation value is greater than the first preset threshold, the step of determining that there is a missed pedicle center point in the current first center point combination may include: if the first distance evaluation value is greater than the first preset threshold, determining that there is a missed pedicle center point between the first two adjacent center points of the three adjacent center points of the current first center point combination.

[0140] In this embodiment, if the computer device determines that the first distance evaluation value of the current first center point combination is greater than the first preset threshold, it can be determined that there is a missed pedicle center point between the first two adjacent center points of the three adjacent center points of the current first center point combination, that is, there is no missed pedicle center point between the last two adjacent center points of the three adjacent center points of the current first center point combination.

[0141] Meanwhile, the specific implementation process of the above-mentioned step S322 is similar to that of step S312, which will not be described in detail in this embodiment.

[0142] The above-mentioned pedicle detection method can obtain the first distance evaluation value of each first center point combination, and further determine the center point coordinates corresponding to the missed pedicle center points in each first center point combination based on the first distance evaluation value of each first center point combination, thereby avoiding the situation where the center points of the pedicle area are missed during the pedicle detection process, and further improving the accuracy of the pedicle detection results corresponding to the target spine.

[0143] In some scenarios, in order to reduce the amount of calculation, such as Figure 10 As shown, in one embodiment, the step of determining the coordinates of the first missed pedicle center point in the above S313 based on the center point coordinates of the pedicle region center point in the current first center point combination and the pedicle center curve equation may specifically include:

[0144] S3131. Determine the average of the first coordinates of the first two adjacent center points in the current first center point combination as the first coordinate of the first missed pedicle center point.

[0145] Specifically, the computer device can calculate the average of the first coordinates of the first two adjacent pedicle region center points in each first center point combination, and use the calculated average of the first coordinates as the first coordinate of the first missed pedicle center point. Optionally, the first coordinate can be a horizontal coordinate or a vertical coordinate.

[0146] In this embodiment, the first two adjacent center points and the last two adjacent center points in each first center point combination can be determined based on the positional relationship of each pedicle region. Figure 5 For example, the corresponding first center point combination 1 obtained is the left pedicle region center point 1, the left pedicle region center point 3 and the left pedicle region center point 4. Correspondingly, the first two adjacent center points in the first center point combination 1 are the left pedicle region center point 1 and the left pedicle region center point 3, and the last two adjacent center points in the first center point combination 1 are the left pedicle region center point 3 and the left pedicle region center point 4; the method for determining the first two adjacent center points and the last two adjacent center points from the three adjacent pedicle region center points in other first center point combinations is similar and will not be repeated here.

[0147] S3132. Substitute the average of the first coordinates of the first two adjacent center points in the current first center point combination into the pedicle center curve equation to obtain the second coordinate of the first missed pedicle center point, and determine the center point coordinate of the first missed pedicle center point through the first coordinate and second coordinate of the first missed pedicle center point.

[0148] Among them, one of the first coordinate and the second coordinate is the horizontal coordinate, and the other is the vertical coordinate.

[0149] It is understood that if the first coordinate is a horizontal coordinate, the second coordinate can be a vertical coordinate; if the first coordinate is a vertical coordinate, the second coordinate can be a horizontal coordinate. Optionally, the computer device can calculate the average of the second coordinates of the center points of the first two adjacent pedicle regions in each first center point combination, and use the calculated average of the second coordinates as the second coordinate of the center point of the first missed pedicle. Furthermore, the first coordinate and the second coordinate of the center point of the first missed pedicle are combined to obtain the center point coordinate of the first missed pedicle.

[0150] Continue to see Figure 5 Example, Figure 11 shows a pedicle segmentation image, and Figure 11 and Figure 5 The pedicle segmentation images in are the same as those in Figure 11 The left pedicle area, the left missed pedicle area, the right pedicle area, and the right missed pedicle area are marked in the figure. These areas are represented by irregular black small areas in the pedicle segmentation image. Figure 11 , Figure 12 The corresponding display in the pedicle segmentation image shows all the pedicle area center points and all the missed pedicle center points (i.e. the first missed pedicle center point and the second missed pedicle center point). Figure 12 The small black circle on the target vertebra indicates the center point of the pedicle area or the center point of the missed pedicle.

[0151] In order to ensure the uniqueness of the coordinates of the missed pedicle center points, in practical applications, if the distribution direction of the pedicle area in the pedicle segmentation image is parallel to the length direction of the pedicle segmentation image, the first coordinate can be the vertical coordinate and the second coordinate can be the horizontal coordinate; if the distribution direction of the pedicle area in the pedicle segmentation image is parallel to the width direction of the pedicle segmentation image, the first coordinate can be the horizontal coordinate and the second coordinate can be the vertical coordinate.

[0152] The above pedicle detection method can obtain the center point coordinates of the second missed pedicle center point, thereby avoiding the situation where the center point of the pedicle area is missed during the pedicle detection process, and further improving the accuracy of the pedicle detection results corresponding to the target spine.

[0153] As one embodiment, before executing the above S100, Figure 13 As shown, the above pedicle detection method may further include:

[0154] S500: Acquire a medical image of the target spine.

[0155] In this embodiment, the medical scanning device can perform real-time scanning of the target spine of the imaging object to obtain target spine scanning data. In this embodiment, the medical scanning device can set the pixel interval resampling of the medical image to a target value (that is, the interval of the data scanned in the x-axis and y-axis directions is normalized to the target value, such as 0.7mm). If the target spine scanning data obtained by the medical scanning device does not meet the pixel interval resampling parameters, the computer device needs to perform interval normalization processing on the target spine scanning data, and then reconstruct the interval normalization processing results to obtain at least one frame of medical image of the target spine. Optionally, the interval normalization process can be to first calculate the average value of the grayscale values ​​of all pixels in the medical image to obtain the grayscale average value, and simultaneously calculate the standard deviation of the grayscale values ​​of the medical image to obtain the grayscale standard deviation, and then, subtract the grayscale average value from the grayscale values ​​of all pixels in the medical image and divide it by the grayscale standard deviation to obtain the normalized result.

[0156] Alternatively, the computer device may obtain at least one frame of medical images of the target spine from the cloud or locally. If the computer device obtains multiple frames of medical images of the target spine, the computer device may select a frame of medical images of the target spine with the highest resolution from the multiple frames of medical images of the target spine.

[0157] It should be noted that the medical image may be a computed tomography image, a magnetic resonance image, etc., but in this embodiment, the medical image is an X-ray image, and is an anteroposterior X-ray image of the target spine.

[0158] S600 , segmenting the pedicles in the medical image of the target spine using a pedicle segmentation model to obtain a pedicle segmentation image.

[0159] Specifically, the pedicle segmentation model can be a pre-trained segmentation network model. The computer device can input the medical image of the target spine into the pedicle segmentation model. The pedicle segmentation model segments the pedicles in the medical image of the target spine to obtain a pedicle mask image. The pedicle mask image is then mapped to the medical image to obtain a pedicle segmentation image. Optionally, the pedicle mask image is the same size as the medical image and can be understood as a labeled binary image. The pixel value of the pedicle area in the pedicle mask image is 1, and the pixel values ​​of other areas are all 0.

[0160] Among them, the pedicle segmentation image can include at least two pedicle regions (i.e., at least any one left pedicle region and the corresponding right pedicle region) and a background region image, that is, the pedicle segmentation image does not include images of the vertebral body, lamina, transverse process, articular process and spinous process regions of the target spine. In this case, the pedicle segmentation image of the target spine can be understood as a noise-free segmentation image.

[0161] However, in this embodiment, the pedicle segmentation image of the target spine may include, in addition to images of at least two pedicle regions and a background region image, images of other bone structures surrounding the pedicle regions of the target spine. In this case, the pedicle segmentation image of the target spine may be understood as a noisy segmentation image. The background region image may be understood as an image corresponding to the region between different bone structures. Figure 5 The pedicle segmentation image in is a noisy segmentation image.

[0162] It can be understood that the above-mentioned pedicle segmentation model can be composed of at least one of a convolutional neural network model, a recurrent neural network model, and an adversarial neural network model; wherein, the computer device can perform network model training on the initial pedicle segmentation model through an image set consisting of medical images of different spines of different imaging objects to obtain a pedicle segmentation model (i.e., a pre-trained segmentation network model).

[0163] Specifically, the computer device can input medical images of different spines in the image set into the initial pedicle segmentation model to obtain a pedicle segmentation prediction result, calculate the prediction error value between the pedicle segmentation prediction result and the standard pedicle segmentation result through the loss function, and update the initial network parameters in the initial pedicle segmentation model based on the prediction error value. The above training steps are continuously iterated until the prediction error value meets the preset error threshold or the number of iterations reaches the preset iteration threshold, thereby obtaining the pedicle segmentation model. The above standard pedicle segmentation result can be an idealized pedicle segmentation result. In particular, during the training process, larger-sized medical images need to be input into the initial pedicle segmentation model for training to obtain a larger receptive field.

[0164] Since medical images are two-dimensionally distributed information, such as information distributed on the x-axis and y-axis, the medical scanning device can perform multiple downsampling in the x-axis and y-axis directions during scanning. In this embodiment, the above-mentioned pedicle segmentation model is composed of a multi-layer deep convolutional neural network structure. For example, the multi-layer deep convolutional neural network structure includes a UNet network structure as an example, that is, the pedicle segmentation model is a UNet segmentation model. The UNet segmentation model includes Figure 14 Structure and Figure 14 The convolutional layer after the structure.

[0165] During training, the medical images in the image set input to the UNet segmentation model are of size 640×512. The first layer of the UNet segmentation model consists of two convolutional layers, which perform two convolutions on the input 640×512 medical images to produce 16 640×512 feature maps. The number 16 indicates the number of convolution kernels used in the two convolutions. Furthermore, the 16 640×512 feature maps are downsampled by the second layer and then convolved once to produce 32 320×256 feature maps. This convolution layer uses 32 convolution kernels, and this process continues until 512 10×8 feature maps are obtained. These 512 10×8 feature maps are then upsampled and convolved with the 512 20×16 feature maps obtained by the penultimate downsampling and convolution, resulting in 1024 20×16 feature maps. Furthermore, the 1024 20×16 feature maps are upsampled and convolved with the 256 40×32 feature maps obtained after the third downsampling and convolution, resulting in 512 40×32 feature maps. Similarly, the remaining layers are similarly constructed, ultimately resulting in 32 640×512 feature maps. Furthermore, the final convolutional layer in the UNet segmentation model processes these 32 640×512 feature maps and outputs a pedicle segmentation mask image.

[0166] Optional, Figure 14 The dotted line in the figure represents the concatenation of the previous output with the current layer's downsampling and convolution results. The rightward triangle represents the convolution layer, and the downward arrow represents the downsampling process. The convolution kernel is 3×3, the stride of the convolution layer is 1×1, the stride of the downsampling is 2×2, and normalization is performed before each convolution. The upsampling kernel is 2×2.

[0167] In addition, the segmentation of the pedicles in the medical image of the target spine using the pedicle segmentation model can also be understood as a process of performing three-category segmentation on the medical image. The pedicle segmentation model can include a three-channel output for outputting three-channel images, which are respectively an image of the left pedicle region, an image of the right pedicle region, and an image of a combination of the background region and other bone structure regions in the pedicle segmentation image, wherein each of these three regions can be a labeled image. The pedicle segmentation image can be a labeled image formed by combining the three images of the left pedicle region, the right pedicle region, and the background region and other bone structure regions.

[0168] The above-mentioned pedicle detection method can obtain a medical image of the target spine, and through the pedicle segmentation model, segment the pedicles in the medical image of the target spine to obtain a pedicle segmentation image, thereby narrowing the image detection range and only detecting the pedicle area in the pedicle segmentation image to obtain the pedicle detection result, reducing the algorithm's computational complexity, thereby shortening the pedicle detection time and improving the detection efficiency; at the same time, the center point of the pedicle area can also be detected within the effective area, which can improve the detection accuracy.

[0169] It should be understood that although Figure 2-4 、 Figure 7-10 and Figure 13 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 2-4 、 Figure 7-10 and Figure 13 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The order of execution of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0170] In one embodiment, Figure 15 As shown, a pedicle detection device is provided, comprising: a first coordinate determination module 11, a fitting processing module 12, a second coordinate determination module 13 and a detection result determination module 14, wherein:

[0171] A first coordinate determination module 11 is configured to determine the center point coordinates of the center points of the pedicle regions in the pedicle segmentation image according to the pedicle segmentation image of the target spine;

[0172] A fitting processing module 12 is used to perform curve fitting processing on the center point coordinates of the center points of each pedicle region to obtain a pedicle center curve equation;

[0173] The second coordinate determination module 13 is used to obtain the coordinates of the center points of the missed pedicles according to the pedicle center curve equation and the center point coordinates of the center points of the pedicle regions;

[0174] The detection result determination module 14 is used to determine the pedicle detection result corresponding to the target spine according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0175] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0176] In one embodiment, the first coordinate determination module 11 includes: an information acquisition unit and a first coordinate acquisition unit, wherein:

[0177] An information acquisition unit, configured to acquire the total number of pixels in each pedicle region and the coordinates of each pixel;

[0178] The first coordinate acquisition unit is used to obtain the center point coordinates of the center point of each pedicle area according to the total number of pixel points in each pedicle area and the coordinates of each pixel point.

[0179] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0180] In one embodiment, the second coordinate determination module 13 includes: a first center point coordinate acquisition unit, a second center point coordinate acquisition unit, and a third center point coordinate acquisition unit, wherein:

[0181] A first center point coordinate acquiring unit is configured to determine the center point coordinates of the first missed pedicle center point according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region in accordance with the first preset direction;

[0182] a second center point coordinate acquisition unit, configured to determine the center point coordinates of the second missed pedicle center point according to a pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point in accordance with a second preset direction; wherein either one of the first preset direction and the second preset direction is a direction from the head end to the tail end of the target spine, and the other preset direction is a direction from the tail end to the head end of the target spine;

[0183] The third center point coordinate acquiring unit is used to determine the center point coordinate of the missed pedicle center point according to the center point coordinate of the first missed pedicle center point and the center point coordinate of the second missed pedicle center point.

[0184] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0185] In one embodiment, the first center point coordinate acquisition unit includes: a first acquisition subunit, a second acquisition subunit and a third acquisition subunit, wherein:

[0186] A first acquisition subunit is configured to acquire a plurality of first center point combinations according to a positional relationship between the first preset direction and each pedicle region; each first center point combination includes three adjacent center points;

[0187] The second acquisition subunit is configured to acquire a first distance evaluation value for each first center point combination based on the center point coordinates of three adjacent center points in each first center point combination; the first distance evaluation value represents a degree of deviation of the distances between any two adjacent center points in each first center point combination;

[0188] The third acquisition subunit is used to determine that there is a first missed pedicle center point in the current first center point combination when the first distance evaluation value is greater than the first preset threshold, and determine the center point coordinates of the first missed pedicle center point based on the center point coordinates of the pedicle area center point in the current first center point combination and the pedicle center curve equation.

[0189] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0190] In one embodiment, the second center point coordinate acquisition unit includes: a fourth acquisition subunit, a fifth acquisition subunit and a sixth acquisition subunit, wherein:

[0191] a fourth acquisition subunit, configured to acquire a plurality of second center point combinations based on the second preset direction, the positional relationship between the pedicle regions, and the positional relationship between the first missed pedicle regions; each second center point combination includes three adjacent center points among the center points of all pedicle regions and the center points of all the first missed pedicle regions;

[0192] a fifth acquisition subunit, configured to acquire a second distance evaluation value for each second center point combination based on the center point coordinates of three adjacent center points in each second center point combination; the second distance evaluation value represents a degree of deviation in the distance between any two adjacent center points in each second center point combination;

[0193] The sixth acquisition subunit is used to determine that there is a second missed pedicle center point in the current second center point combination when the second distance evaluation value is greater than the second preset threshold, and determine the center point coordinates of the second missed pedicle center point based on the center point coordinates of the pedicle area center point in the current second center point combination and the pedicle center curve equation.

[0194] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0195] In one embodiment, the second acquisition subunit is specifically used to obtain, for each first center point combination, a first distance between the first two adjacent center points in the three adjacent center points, and a second distance between the last two adjacent center points in the three adjacent center points based on the coordinates of the three adjacent center points in the first center point combination, and determine the ratio between the first distance and the second distance as the first distance evaluation value of the first center point combination.

[0196] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0197] In one embodiment, the third acquisition subunit includes a missed detection point determination subunit, wherein:

[0198] The missed point determination subunit is configured to determine, when the first distance evaluation value is greater than a first preset threshold, that there is a missed pedicle center point between the first two adjacent center points of the three adjacent center points of the current first center point combination.

[0199] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0200] In one embodiment, the third acquisition subunit further includes: a vertical coordinate acquisition subunit and a horizontal coordinate acquisition subunit, wherein:

[0201] The vertical coordinate acquisition subunit is used to determine the average of the first coordinates of the first two adjacent center points in the current first center point combination as the first coordinate of the first missed pedicle center point;

[0202] a horizontal coordinate acquisition subunit, configured to substitute the average of the first coordinates of the first two adjacent center points in the current first center point combination into the pedicle center curve equation to obtain the second coordinate of the first missed pedicle center point, and determine the center point coordinate of the first missed pedicle center point through the first coordinate and the second coordinate of the first missed pedicle center point;

[0203] Among them, one of the first coordinate and the second coordinate is the horizontal coordinate, and the other is the vertical coordinate.

[0204] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0205] In one embodiment, the pedicle detection device further includes: a medical image acquisition module and a segmentation module, wherein:

[0206] Medical image acquisition: Acquire medical images of the target spine;

[0207] The segmentation module is used to segment the pedicles in the medical image of the target spine through the pedicle segmentation model to obtain a pedicle segmentation image.

[0208] The pedicle detection device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0209] The specific limitations of the pedicle detection device can be found in the limitations of the pedicle detection method described above and will not be further elaborated here. Each module in the aforementioned pedicle detection device can be implemented in whole or in part via software, hardware, or a combination thereof. Each of these modules can be embedded in or independent of a processor within a computer device in hardware form, or stored in a computer device memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0210] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 16 As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store medical images and pedicle segmentation images. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a pedicle detection method is implemented.

[0211] Those skilled in the art will understand that Figure 16 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0212] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0213] According to the pedicle segmentation image of the target spine, the center point coordinates of the center points of each pedicle region in the pedicle segmentation image are determined;

[0214] Perform curve fitting on the coordinates of the center point of each pedicle area to obtain the pedicle center curve equation;

[0215] According to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, the center point coordinates of the missed pedicle center points are obtained;

[0216] The pedicle detection result corresponding to the target spine is determined according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0217] In one embodiment, a readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0218] According to the pedicle segmentation image of the target spine, the center point coordinates of the center points of each pedicle region in the pedicle segmentation image are determined;

[0219] Perform curve fitting on the coordinates of the center point of each pedicle area to obtain the pedicle center curve equation;

[0220] According to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, the center point coordinates of the missed pedicle center points are obtained;

[0221] The pedicle detection result corresponding to the target spine is determined according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0222] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0223] According to the pedicle segmentation image of the target spine, the center point coordinates of the center points of each pedicle region in the pedicle segmentation image are determined;

[0224] Perform curve fitting on the coordinates of the center point of each pedicle area to obtain the pedicle center curve equation;

[0225] According to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area, the center point coordinates of the missed pedicle center points are obtained;

[0226] The pedicle detection result corresponding to the target spine is determined according to the center point coordinates of the center points of each pedicle area and the center point coordinates of the missed pedicle center points.

[0227] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0228] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0229] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A pedicle detection method, characterized in that: The method comprises: Determining the center point coordinates of the center points of each pedicle region in the pedicle segmentation image according to the pedicle segmentation image of the target spine; Performing curve fitting processing on the center point coordinates of the center points of the pedicle regions to obtain a pedicle center curve equation; Obtaining the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region; Determining a pedicle detection result corresponding to the target spine according to the center point coordinates of the center points of each pedicle region and the center point coordinates of the missed pedicle center points; Wherein, obtaining the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region includes: Determining reference coordinates based on the center point coordinates of at least two pedicle region center points; Substitute the horizontal coordinate in the reference coordinate into the pedicle center curve equation to obtain the vertical coordinate to be processed, and combine the vertical coordinate to be processed with the horizontal coordinate in the reference coordinate to obtain the center point coordinate of the missed pedicle center point; or substitute the vertical coordinate in the reference coordinate into the pedicle center curve equation to obtain the horizontal coordinate to be processed, and combine the horizontal coordinate to be processed with the vertical coordinate in the reference coordinate to obtain the center point coordinate of the missed pedicle center point.

2. The method according to claim 1, characterized in that The step of obtaining the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of the pedicle regions comprises: According to the first preset direction, the center point coordinates of the first missed pedicle center point are determined based on the pedicle center curve equation and the center point coordinates of the center points of each pedicle region; Determining the center point coordinates of the second missed pedicle center point according to the pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point according to the second preset direction; either one of the first preset direction and the second preset direction is a direction from the head end to the tail end of the target spine, and the other preset direction is a direction from the tail end to the head end of the target spine; The center point coordinates of the missed pedicle center point are determined based on the center point coordinates of the first missed pedicle center point and the center point coordinates of the second missed pedicle center point.

3. The method according to claim 2, characterized in that Determining the center point coordinates of the first missed pedicle center point according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle region in accordance with the first preset direction includes: Acquire a plurality of first center point combinations according to the positional relationship between the first preset direction and each of the pedicle regions; each first center point combination includes three adjacent center points; Obtaining a first distance evaluation value for each of the first center point combinations based on the center point coordinates of three adjacent center points in each of the first center point combinations; the first distance evaluation value represents a degree of deviation in the distances between any two adjacent center points in each of the first center point combinations; If the first distance evaluation value is greater than a first preset threshold, it is determined that there is a first missed pedicle center point in the current first center point combination, and the center point coordinates of the first missed pedicle center point are determined based on the center point coordinates of the pedicle area center points in the current first center point combination and the pedicle center curve equation.

4. The method according to claim 3, characterized in that Determining the center point coordinates of the second missed pedicle center point according to the pedicle center curve equation, the center point coordinates of the center points of each pedicle region, and the center point coordinates of the first missed pedicle center point in accordance with the second preset direction includes: Acquire multiple second center point combinations according to the second preset direction, the positional relationship between each of the pedicle regions, and the positional relationship between the first missed pedicle center points; each second center point combination includes three adjacent center points among the center points of all pedicle regions and all the first missed pedicle center points; Obtaining a second distance evaluation value for each of the second center point combinations based on the center point coordinates of three adjacent center points in each of the second center point combinations; the second distance evaluation value represents a degree of deviation in the distances between any two adjacent center points in each of the second center point combinations; If the second distance evaluation value is greater than the second preset threshold, the second missed pedicle center point in the current second center point combination is determined, and the center point coordinates of the second missed pedicle center point are determined based on the center point coordinates of the pedicle area center point in the current second center point combination and the pedicle center curve equation.

5. The method according to claim 4, characterized in that The obtaining, according to the center point coordinates of three adjacent center points in each of the first center point combinations, a first distance evaluation value of each of the first center point combinations includes: For each of the first center point combinations, obtaining, based on the coordinates of three adjacent center points in the first center point combination, a first distance between first two adjacent center points of the three adjacent center points, and a second distance between last two adjacent center points of the three adjacent center points; The ratio between the first distance and the second distance is determined as a first distance evaluation value of the first center point combination.

6. The method according to claim 5, characterized in that If the first distance evaluation value is greater than a first preset threshold, determining that there is a missed pedicle center point in the current first center point combination includes: If the first distance evaluation value is greater than the first preset threshold, it is determined that there is a missed pedicle center point between the first two adjacent center points of the three adjacent center points of the current first center point combination.

7. The method according to claim 6, characterized in that Determining the center point coordinates of the first missed pedicle center point based on the center point coordinates of the pedicle region center point in the current first center point combination and the pedicle center curve equation includes: Determine the average of the first coordinates of the first two adjacent center points in the current first center point combination as the first coordinate of the first missed pedicle center point; Substitute the average of the first coordinates of the first two adjacent center points in the current first center point combination into the pedicle center curve equation to obtain the second coordinate of the first missed pedicle center point, and determine the center point coordinate of the first missed pedicle center point through the first coordinate and the second coordinate of the first missed pedicle center point; one of the first coordinate and the second coordinate is the horizontal coordinate, and the other is the vertical coordinate.

8. The method according to claim 1, characterized in that The method further comprises: acquiring a medical image of the target spine; The pedicles in the medical image of the target spine are segmented using the pedicle segmentation model to obtain the pedicle segmentation image.

9. A pedicle detection device, characterized in that: The device comprises: A first coordinate determination module is used to determine the center point coordinates of the center points of each pedicle region in the pedicle segmentation image according to the pedicle segmentation image of the target spine; A fitting processing module is used to perform curve fitting processing on the center point coordinates of the center points of each pedicle area to obtain a pedicle center curve equation; A second coordinate determination module is used to obtain the center point coordinates of the missed pedicle center points according to the pedicle center curve equation and the center point coordinates of the center points of each pedicle area; a detection result determination module, configured to determine a pedicle detection result corresponding to the target spine according to the center point coordinates of the center points of the pedicle regions and the center point coordinates of the missed pedicle center points; Wherein, the second coordinate determination module is specifically used for: Determining reference coordinates based on the center point coordinates of at least two pedicle region center points; Substituting the horizontal coordinate in the reference coordinate into the pedicle center curve equation to obtain the vertical coordinate to be processed; or substituting the vertical coordinate in the reference coordinate into the pedicle center curve equation to obtain the horizontal coordinate to be processed; The to-be-processed ordinate is combined with the abscissa in the reference coordinate to obtain the center coordinate of the missed pedicle center point; or the to-be-processed abscissa is combined with the ordinate in the reference coordinate to obtain the center coordinate of the missed pedicle center point.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images