A single vertebra segmentation method, device, equipment and storage medium

Abstract

The application provides a single-segment vertebra segmentation method, device, equipment and storage medium, and relates to the technical field of medical image processing. The single-segment vertebra segmentation method comprises the following steps: obtaining first spine boundary information, wherein the first spine boundary information is boundary information of a spine in a segmentation result obtained by performing image segmentation on an original CT image; obtaining first vertebra position information, wherein the first vertebra position information is position information of a single-segment vertebra in the spine in the original CT image; obtaining an initial image of the single-segment vertebra according to the first spine boundary information and the first vertebra position information; marking the position of the single-segment vertebra in the initial image of the single-segment vertebra to generate an updated image of the single-segment vertebra; and determining the single-segment vertebra in a restored image corresponding to the original CT image according to the updated image of the single-segment vertebra. According to the embodiment of the application, the single-segment segmentation of the spine in the CT image can be performed.

Description

Technical Field

[0001] This application relates to the field of medical image processing technology, and more specifically, to a method, apparatus, device, and storage medium for segmenting a single vertebral segment. Background Technology

[0002] Currently, spinal surgeons generally need to develop surgical plans based on the patient's CT images before surgery. Doctors require clearer and more intuitive spinal CT images to make accurate diagnoses. Therefore, high-precision single-segment segmentation of spinal CT images is of great significance in spinal surgery.

[0003] The inventors of this application have discovered that segmenting the spine into individual segments is more user-friendly than existing methods that separate the entire spine from background information. However, the similarity in strength of adjacent structures within the spine, as well as the differences in shape and size of vertebrae in different parts of the spine, increases the difficulty of segmenting the spine into individual segments. Summary of the Invention

[0004] According to one aspect of this application, a method for segmenting a single vertebra is provided, comprising: acquiring first spinal boundary information, wherein the first spinal boundary information is the boundary information of the spine in the segmentation result obtained after image segmentation of an original CT image; acquiring first vertebral position information, wherein the first vertebral position information is the position information of a single vertebra in the spine in the original CT image; acquiring an initial image of the single vertebra based on the first spinal boundary information and the first vertebral position information; marking the position of the single vertebra in the initial image of the single vertebra to generate an updated image of the single vertebra; and determining the single vertebra in the restored image corresponding to the original CT image based on the updated image of the single vertebra.

[0005] According to some embodiments, obtaining first spinal boundary information includes: constructing a first network model; obtaining segmentation results of the spine, sacrum, and ribs in the original CT image through the first network model to obtain a segmented image of the original CT image; performing morphological dilation on the spine in the segmented image to obtain a morphologically dilated spinal image; converting the morphologically dilated spinal image into a first spinal image in an image coordinate system; and obtaining the first spinal boundary information corresponding to the first spinal image.

[0006] According to some embodiments, obtaining the first vertebral position information includes: constructing a second network model; and obtaining the first vertebral position information from the original CT image through the second network model.

[0007] According to some embodiments, obtaining an initial image of a single vertebra based on the first spinal boundary information and the first vertebral position information includes: determining the boundary information of a single vertebra based on the first spinal boundary information and the first vertebral position information; and obtaining the initial image of the single vertebra from the first spinal image based on the boundary information of the single vertebra.

[0008] According to some embodiments, marking the position of a single vertebra in an initial image of the single vertebra to generate an updated image of the single vertebra includes: determining second spinal boundary information, wherein the second spinal boundary information is the spinal boundary information in the initial image of the single vertebra; updating the position information of the single vertebra in the initial image of the single vertebra based on the second spinal boundary information; obtaining the position marker of the single vertebra in the initial image of the single vertebra based on the updated position information of the single vertebra; converting the initial image of the single vertebra containing the position marker of the single vertebra into a position marker image of the single vertebra in a volume coordinate system; and generating an updated image of the single vertebra based on the initial image of the single vertebra and the position marker image of the single vertebra.

[0009] According to some embodiments, generating an updated image of a single vertebra based on an initial image of the single vertebra and a location marker image of the single vertebra includes: constructing a third network model; converting the initial image of the single vertebra into an initial image of the single vertebra in a volume coordinate system; inputting the initial image of the single vertebra in the volume coordinate system and the location marker image of the single vertebra into the third network model in a preset order; obtaining the updated image of the single vertebra through the third network model in the preset order; and converting the updated image of the single vertebra into an updated image of the single vertebra in an image coordinate system.

[0010] According to some embodiments, determining a single vertebra in the restored image corresponding to the original CT image based on the updated image of the single vertebra includes: converting the original CT image into an original image in an image coordinate system; obtaining a zero-pixel image corresponding to the original image; marking the single vertebra in the zero-pixel image according to the preset order based on the updated image of the single vertebra in the image coordinate system; obtaining the position marker of the single vertebra in the original image based on the zero-pixel image corresponding to the marked single vertebra; updating the boundary information of the single vertebra in the original image of the marked single vertebra based on the updated image of the single vertebra in the image coordinate system to determine the single vertebra in the original image; and determining the original image with the determined single vertebra as the restored image.

[0011] According to one aspect of this application, a segmentation device for a single vertebra is provided, comprising: a data acquisition module for acquiring an original CT image; a data processing module for setting a first network model, a second network model, and a third network model; acquiring segmentation results of the spine, sacrum, and ribs from the original CT image using the first network model; acquiring first vertebral boundary information based on the segmentation results; acquiring first vertebral position information from the original CT image using the second network model; acquiring an initial image of a single vertebra based on the first vertebral boundary information and the first vertebral position information; marking the position of the single vertebra in the initial image of the single vertebra; generating an updated image of the single vertebra using the third network model based on the initial image of the single vertebra and the position-marked image of the single vertebra; determining the single vertebra in a restored image in the image coordinate system corresponding to the original CT image based on the updated image of the single vertebra; and a data output module for outputting the restored image after determining the single vertebra.

[0012] According to one aspect of this application, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs; and, when the one or more programs are executed by the one or more processors, causing the one or more processors to perform the method as described above.

[0013] According to one aspect of this application, a computer-readable storage medium is provided that stores a computer program or instructions thereon, which, when executed by a processor, implement the method as described above.

[0014] According to embodiments of this application, a neural network model can be used to segment the spine in CT images into single segments, dividing the CT image into small blocks corresponding to each single vertebra in sequence. This solves the problem of single-segment spine segmentation that is difficult to achieve in the prior art, shortens the spine segmentation time, improves the spine segmentation accuracy, and makes spine segmentation more efficient and faster.

[0015] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit this application. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application.

[0017] Figure 1 A flowchart illustrating a method for segmenting a single vertebra according to an example embodiment of this application is shown.

[0018] Figure 2A diagram showing the effect of displaying a single vertebra according to an example embodiment of this application.

[0019] Figure 3 A flowchart illustrating the acquisition of an initial image of a single vertebra according to an example embodiment of this application is shown.

[0020] Figure 4 A flowchart illustrating the acquisition of an updated image of a single vertebra according to an example embodiment of this application is shown.

[0021] Figure 5 A flowchart illustrating the training process of a first network model according to an example embodiment of this application is shown.

[0022] Figure 6 This diagram illustrates the segmentation effect of the spine, sacrum, and ribs according to an example embodiment of this application.

[0023] Figure 7 A flowchart illustrating the training process of a second network model according to an example embodiment of this application is shown.

[0024] Figure 8 A rendered diagram showing the position information of a single vertebra in the spine according to an example embodiment of this application.

[0025] Figure 9 A flowchart illustrating the training process of a third network model according to an example embodiment of this application is shown.

[0026] Figure 10 A schematic diagram of a single vertebral segment is shown according to an example embodiment of this application.

[0027] Figure 11 A block diagram of a segmentation device for a single vertebra according to an example embodiment of this application is shown.

[0028] Figure 12 A block diagram of an electronic device according to an example embodiment of this application is shown. Detailed Implementation

[0029] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that this application will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0030] The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a full understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of these specific details, or other methods, components, materials, apparatus, or operations may be employed. In these cases, well-known structures, methods, apparatuses, implementations, materials, or operations will not be shown or described in detail.

[0031] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0032] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0033] This application provides a method, apparatus, device, and storage medium for segmenting a single vertebral segment. It achieves rapid and effective segmentation of a single vertebral segment through a neural network model, thereby improving the segmentation accuracy of a single vertebral segment.

[0034] The following will describe in detail, with reference to the accompanying drawings, a method, apparatus, device, and storage medium for segmenting a single vertebra according to embodiments of this application.

[0035] This application uses the following terms:

[0036] Zero-pixel image: An image obtained by replacing all elements in the 3D matrix corresponding to the coordinates of each point in the image coordinate system with 0. In a zero-pixel image, the pixel label value of each point is 0.

[0037] Figure 1 A flowchart illustrating a method for segmenting a single vertebra according to an example embodiment of this application is shown.

[0038] like Figure 1 As shown, in step S110, the first spinal boundary information is obtained.

[0039] For example, in step S110, the first spinal boundary information is obtained through the first network model, wherein the first spinal boundary information is the boundary information of the spinal image after morphological dilation in the original CT image.

[0040] In a specific embodiment of this application, the specific steps for obtaining the first spinal boundary information are as follows:

[0041] A first network model is constructed, and the original CT image is input into the first network model to obtain the segmentation results of the spine, sacrum and ribs in the original CT image through the first network model, thereby obtaining the segmented image of the original CT image.

[0042] After obtaining the segmentation results, in order to avoid incomplete or inaccurate spine boundaries, morphological dilation is performed on the spine in the segmented image to obtain a morphologically dilated spine image.

[0043] The morphologically dilated spine image in volume coordinates is transformed into the first spine image in image coordinates.

[0044] The morphologically expanded spinal boundary information, i.e., the first spinal boundary information, is obtained from the first spinal image.

[0045] According to some embodiments, the first spinal boundary information includes the left and right boundaries of the morphologically expanded spinal image in the direction from one transverse process of the vertebra to the other (x-axis), the anterior and posterior boundaries in the direction from the spinous process of the spine to the vertebral body (y-axis), and the upper and lower boundaries in the direction from the sacrum to the cervical spine (z-axis).

[0046] In step S120, the position information of the first vertebra is obtained.

[0047] For example, in step S120, the location information of the first vertebra is obtained through the second network model, wherein the location information of the first vertebra is the location information of a single vertebra in the spine in the original CT image.

[0048] In a specific embodiment of this application, the specific steps for obtaining the position information of the first vertebra are as follows:

[0049] A second network model is constructed, and the original CT image is input into the second network model to obtain the position information of a single vertebra in the spine in the original CT image, i.e., the position information of the first vertebra.

[0050] In step S130, an initial image of a single vertebra is obtained based on the first vertebral boundary information and the first vertebral position information.

[0051] For example, in step S130, the boundary information of a single vertebra is determined based on the first vertebral boundary information and the first vertebral position information, and the initial image of the single vertebra is obtained from the first vertebral image corresponding to the morphologically expanded vertebral image based on the boundary information of the single vertebra.

[0052] In a specific embodiment of this application, the specific steps for obtaining an initial image of a single vertebra based on the first spinal boundary information and the first vertebra position information are as follows:

[0053] Based on the boundary information and position information of the first vertebral column, the boundary information of each individual vertebral column is determined.

[0054] An initial image of a single vertebra is obtained from the first vertebral image based on the boundary information of the single vertebra.

[0055] In step S140, the position of the single vertebra in the initial image of the single vertebra is marked to generate an updated image of the single vertebra.

[0056] For example, in step S140, the second spinal boundary information is first obtained, and then the position information of a single vertebra is updated based on the second spinal boundary information. The position of the single vertebra in the initial image of the single vertebra is marked according to the updated position information of the single vertebra to generate an updated image of the single vertebra.

[0057] In a specific embodiment of this application, the specific steps for marking the position of a single vertebra in an initial image of a single vertebra to generate an updated image of that single vertebra are as follows:

[0058] Determine the spinal boundary information in the initial image of a single vertebra, i.e., the second spinal boundary information.

[0059] The positional information of each individual vertebra in the initial image of the individual vertebra is updated using the second spinal boundary information.

[0060] Based on the updated position information of each individual vertebra, the positions of the individual vertebrae in the initial image of the individual vertebrae are marked, and the corresponding position-marked images of the individual vertebrae in volume coordinates are obtained. According to some embodiments, each position mark of each individual vertebra in the initial image of the individual vertebrae is assigned a corresponding pixel label value.

[0061] An updated image of a single vertebra is generated based on the initial image and the location-marked image of the single vertebra.

[0062] In a specific embodiment of this application, the specific steps for generating an updated image of a single vertebra based on an initial image and a position marker image of the single vertebra are as follows:

[0063] Transform the initial image of a single vertebra into an initial image of a single vertebra in volume coordinates.

[0064] A third network model is constructed by overlaying the initial image of a single vertebra in volume coordinates with the position marker image of the single vertebra, and inputting them into the third network model in a preset order, so as to obtain the updated image of the single vertebra through the third network model in a preset order.

[0065] The updated image of a single vertebra in volume coordinates obtained through the third network model is transformed into an updated image of a single vertebra in image coordinates.

[0066] According to some embodiments, the updated image of a single vertebra includes a location marker for each vertebra, and each location marker for a single vertebra contains a corresponding pixel label value.

[0067] In step S150, the single vertebrae of the reconstructed image corresponding to the original CT image are determined based on the updated image of the single vertebra.

[0068] For example, in step S150, based on the single vertebrae corresponding to the updated image of the single vertebrae in the image coordinates, the single vertebrae in the original image are marked in a preset order, and the boundary information of the single vertebrae in the original image is updated, so as to determine the single vertebrae in the restored image corresponding to the original image.

[0069] In a specific embodiment of this application, the single vertebral segment is determined from the updated image of the original CT image to the reconstructed image of the single vertebral segment. The specific steps are as follows:

[0070] First, the acquired raw CT images are transformed into their original coordinate system to convert them into the original images in the image coordinate system.

[0071] Based on the updated images of individual vertebrae in the image coordinate system, the individual vertebrae in the original image are marked in a preset order. According to some embodiments, a zero-pixel image of the same size as the original image in the image coordinate system is generated. Following a preset order of the updated images of individual vertebrae output by the third network model, the pixel value of the position marker of each individual vertebra in the updated image of the individual vertebrae in the image coordinate system replaces the pixel label value of the corresponding position in the zero-pixel image of the original image, so that the position marker of each individual vertebra in the updated image of the individual vertebrae becomes the position marker of each individual vertebra in the zero-pixel image of the original image. Furthermore, the pixel label value corresponding to the position marker of each individual vertebra in the zero-pixel image of the original image is modified to the same value as the preset order of the updated images of individual vertebrae output by the third network model. For example, the pixel label value 1 of the position marker of each individual vertebra in the zero-pixel image of the original image is modified to 1, 2, 3… in a preset order.

[0072] After marking the position of each vertebra in the zero-pixel image corresponding to the original image, the position marking of each vertebra in the original image is obtained accordingly.

[0073] Based on the updated image of each individual vertebra in the image coordinate system, the boundary information of the original image with marked individual vertebrae is updated. The boundary information of each individual vertebra in the original image with marked individual vertebrae is replaced with the boundary information of the individual vertebrae in the updated image of the individual vertebrae in the image coordinate system, thereby determining the individual vertebrae in the original image. According to some embodiments, after obtaining the position markers of each individual vertebra in the zero-pixel image corresponding to the original image, since the original image in the image coordinate system and its corresponding zero-pixel image have the same size, the position markers of each individual vertebra in the original image can be obtained accordingly. The boundary information of each individual vertebra is obtained through the updated image of the individual vertebrae in the image coordinate system, and the boundary information of each individual vertebra in the original image is replaced with the boundary information of each individual vertebra in the updated image of the individual vertebrae in the image coordinate system. Furthermore, based on the position markers of each individual vertebra in the original image and the replaced boundary information of each individual vertebra in the original image, the individual vertebrae in the original image can be determined.

[0074] After each individual vertebra in the original image is identified, the original image with the identified individual vertebrae is used as the reconstructed image corresponding to the original CT image. According to some embodiments, each individual vertebra in the reconstructed image is identified and displayed in the reconstructed image with pixel label values ​​in the same preset order as the updated images of the individual vertebrae output by the third network model. The display of each individual vertebra in the reconstructed image can be seen in... Figure 2 The image shown is a rendering of a single vertebra.

[0075] According to the embodiments of this application, the technical solution of this application can perform single-segment segmentation of the spine in CT images through a neural network model, which can quickly obtain each single vertebra in the spine and improve the segmentation accuracy.

[0076] Figure 3 A flowchart illustrating the acquisition of an initial image of a single vertebra according to an example embodiment of this application is shown.

[0077] like Figure 3 As shown, in step S131, the first spine image in the image coordinate system is obtained.

[0078] For example, in step S131, a morphologically dilated spine image is obtained, and the morphologically dilated spine image in volume coordinates is transformed into a first spine image in image coordinates.

[0079] In step S132, the boundary information of a single vertebra is determined based on the first vertebral boundary information and the first vertebral position information.

[0080] For example, in step S132, the boundary information of the morphologically expanded spinal image (i.e., the first spinal boundary information) and the position information of a single vertebra in the spinal column (i.e., the first vertebra position information) are obtained, and the boundary information of each single vertebra is determined accordingly.

[0081] In a specific embodiment of this application, after obtaining the morphologically dilated spinal image through the first network model, the first spinal boundary information corresponding to the morphologically dilated spinal image is obtained, including the upper boundary (top) and lower boundary (bottom) of the spine in the z-axis direction.

[0082] After obtaining the location information of the first vertebra through the second network model, the boundary information of a single vertebra in the x-axis, y-axis and z-axis directions is determined based on the location information of the first vertebra and the obtained boundary information of the first vertebra. The boundaries of the single vertebra in the x-axis and y-axis directions are consistent with the boundaries of the morphologically dilated spine image in the x-axis and y-axis directions.

[0083] According to some embodiments, the first vertebral position information is obtained through a second network model, and each individual vertebra in the spine is sorted along the z-axis direction, with the sorting result denoted as order.

[0084] According to some embodiments, the upper boundary (top) and lower boundary (bottom) in the z-axis direction of the first spinal boundary information are obtained. If a single vertebra is the first in the sorting result, its corresponding z-axis coordinate range is [bottom, order[i+1].z]; if a single vertebra is the last in the sorting result, its corresponding z-axis coordinate range is [order[i-1].z, top]; if a single vertebra is any one in the sorting result other than the first and last, its corresponding z-axis coordinate range is [order[i-1], order[i+1].z]. Where order[i].z represents the z-axis coordinate of the i-th single vertebra corresponding to the sorting result.

[0085] Determine the z-axis coordinate range of a single vertebra based on the sorting results [curmin] z curmax z [ ], which refers to the boundary information of a single vertebra in the z-axis direction.

[0086] In step S133, an initial image of a single vertebra is obtained from the first spinal image based on the boundary information of the single vertebra.

[0087] For example, in step S133, an initial image of each individual vertebra is extracted from the first spinal image based on the boundary information of the individual vertebrae, wherein the boundaries of the initial image of the individual vertebra in the x-axis and y-axis directions are consistent with the boundaries of the first spinal image in the x-axis and y-axis directions.

[0088] Figure 4 A flowchart illustrating the acquisition of an updated image of a single vertebra according to an example embodiment of this application is shown.

[0089] like Figure 4 As shown, in step S141, the position information of a single vertebra in the initial image of a single vertebra is updated according to the second spinal boundary information.

[0090] For example, in step S141, the spinal boundary information in the initial image of a single vertebra is obtained, namely the second spinal boundary information, and the position information of each single vertebra in the initial image of the single vertebra is updated according to the second spinal boundary information.

[0091] In one specific embodiment of this application, the boundary information of the second vertebra is determined based on an initial image of a single vertebra.

[0092] According to some embodiments, spinal boundary information in an initial image of a single vertebra can be obtained using the numpy.where() method of the numpy scientific computing library.

[0093] The positional information of each individual vertebra in the initial image of a single vertebra is updated using the second spinal boundary information.

[0094] In step S142, based on the updated position information of the individual vertebrae, the position of the individual vertebrae in the initial image of the individual vertebrae is marked to obtain a position-marked image of the individual vertebrae in the body coordinate system.

[0095] For example, in step S142, based on the updated position information of individual vertebrae, the positions of individual vertebrae in the initial image of the individual vertebrae are marked, and the position marks of the individual vertebrae are obtained. The initial image of the individual vertebrae containing the position marks of the individual vertebrae is transformed into a coordinate system image of the position marks of the individual vertebrae in a volume coordinate system.

[0096] In a specific embodiment of this application, the position marker of a single vertebra in the initial image of the single vertebra is obtained based on the updated position information of the single vertebra.

[0097] According to some embodiments, a zero-pixel map with the same initial image size as the individual vertebrae in the image coordinate system is first generated. Based on the updated position information of the individual vertebrae, each individual vertebra is marked in the zero-pixel map to obtain a position-marked image of the individual vertebrae in the image coordinate system.

[0098] According to some embodiments, the point corresponding to the updated position coordinates of a single vertebra can be used as the center of a sphere. A small sphere with a radius within a preset range (which can be adjusted according to actual needs, for example, the preset range is [1,7]) is set to replace the point corresponding to the position coordinates of the single vertebra in the zero-pixel image, serving as the position marker for each single vertebra in the zero-pixel image, and the pixel label value of the small sphere is set to 1. The initial image of the single vertebra is then overlaid with the zero-pixel image that has already marked the position of the single vertebra to obtain the position marker of the single vertebra in the initial image of the single vertebra.

[0099] The initial image of a single vertebra containing location markers is transformed into a location marker image of the single vertebra in volume coordinates.

[0100] In step S143, an updated image of the single vertebra is generated based on the initial image of the single vertebra and the position marker image of the single vertebra in the body coordinate system.

[0101] For example, in step S143, the initial image of a single vertebra is converted into an initial image of a single vertebra in volume coordinates. The initial image of the single vertebra in volume coordinates and the position marker image of the single vertebra in volume coordinates are superimposed and input into the pre-set third network model, and the updated image of the single vertebra is obtained through the third network model.

[0102] In a specific embodiment of this application, a third network model is constructed, and the obtained initial image of a single vertebra is transformed into an initial image of a single vertebra in volume coordinates.

[0103] The initial image of a single vertebra in volume coordinates and the position marker image of a single vertebra in volume coordinates are superimposed and input into the third network model in a preset order. The third network model then obtains the updated image of the single vertebra in the preset order of the input data.

[0104] According to some embodiments, the updated image of a single vertebra includes a location marker for each vertebra, and each location marker for a single vertebra contains a corresponding pixel label value.

[0105] Figure 5 A flowchart illustrating the training process of a first network model according to an example embodiment of this application is shown.

[0106] like Figure 5 As shown, in step S210, the first dataset is obtained.

[0107] For example, in step S210, a first dataset is obtained from a public dataset according to a preset specification for training a first network model.

[0108] In a specific embodiment of this application, a first dataset is obtained from a public dataset. The first dataset includes various types of spinal CT images and standard position information of each individual vertebra in the spinal CT images, wherein the standard position information of each individual vertebra is position information in the image coordinate system.

[0109] According to some embodiments, the preset specification for acquiring spinal CT images can be 512*512*N, where N represents the number of layers of CT image data and 512*512 represents the size of each layer of CT image data.

[0110] According to some embodiments, the various categories of spinal CT images included in the first dataset correspond to spinal CT images from the cervical to the sacral vertebrae that include complete ribs; spinal CT images that include complete ribs, part of the thoracic vertebrae and part of the lumbar vertebrae, but not the sacral vertebrae; and spinal CT images that include complete sacral vertebrae, part of the thoracic vertebrae and part of the lumbar vertebrae, but not the ribs.

[0111] According to some embodiments, in order to avoid classification errors of the spine and sacrum caused by sacralization, the first dataset also includes CT images of the spine from the cervical to the sacral vertebrae containing complete ribs and sacral vertebrae that have transformed into lumbar morphology.

[0112] In step S220, the first network model is trained using the first dataset.

[0113] For example, in step S220, a first network model is set up and trained using the obtained first dataset.

[0114] In a specific embodiment of this application, after constructing the first network model, the spinal CT images from the first dataset are input into the first network model, and the segmentation results of the spine, sacrum, and ribs in the spinal CT images are obtained through the first network model. The segmentation results of the spine, sacrum, and ribs in the spinal CT images can be seen as follows: Figure 6 The diagram shows the segmentation of the spine, sacrum, and ribs.

[0115] According to some embodiments, the first network model may be an nnUNet neural network model.

[0116] Figure 7 A flowchart illustrating the training process of a second network model according to an example embodiment of this application is shown.

[0117] like Figure 7 As shown, in step S310, the second dataset is obtained.

[0118] For example, in step S310, a second dataset is obtained from a public dataset according to a preset specification for training a second network model.

[0119] In a specific embodiment of this application, a second dataset is obtained from a public dataset. The second dataset includes various types of spinal CT images and standard position information of each individual vertebra in the spinal CT images, wherein the standard position information of each individual vertebra is position information in the image coordinate system.

[0120] According to some embodiments, the preset specification for acquiring spinal CT images can be 512*512*N, where N represents the number of layers of CT image data and 512*512 represents the size of each layer of CT image data.

[0121] According to some embodiments, the multi-category spinal CT images included in the second dataset correspond to complete spinal CT images from the cervical spine to the sacrum; spinal CT images containing parts of the cervical spine and parts of the thoracic spine; spinal CT images containing parts of the thoracic spine and parts of the lumbar spine; and spinal CT images containing parts of the thoracic spine and complete lumbar spine.

[0122] In step S320, the labels of the second network model are obtained based on the second dataset.

[0123] For example, in step S320, the spinal CT images in the second dataset are converted into spinal images in an image coordinate system. Each individual vertebra in the spinal images in the image coordinate system is labeled according to its standard location information. Labels for the second network model are generated based on the labeled spinal images of each individual vertebra.

[0124] In one specific embodiment of this application, the spinal CT images in the second dataset in the volume coordinate system are converted into spinal images in the image coordinate system.

[0125] According to some embodiments, the mutual conversion between the volume coordinate system and the image coordinate system can be achieved by calling the application programming interface of the corresponding software, such as the Python runtime library SimpleITK.

[0126] Furthermore, a zero-pixel image with the same dimensions as the spine image in the image coordinate system is generated.

[0127] Based on the standard position information of each individual vertebra, the position of each individual vertebra in the spine is marked in the zero-pixel image.

[0128] According to some embodiments, the point corresponding to the standard position coordinates of each individual vertebra can be used as the center of a sphere. A small sphere with a radius within a preset value range (which can be adjusted according to actual needs, for example, the preset value range is [1,7]) can be rendered to replace the point corresponding to the position coordinates of the individual vertebra in the zero-pixel image, so as to serve as the position mark of each individual vertebra in the zero-pixel image, and the pixel label value of the small sphere can be set to 1.

[0129] The spine image in the image coordinate system is overlaid with a zero-pixel image that has marked the position of each individual vertebra in the spine to obtain a position-marked image of each individual vertebra in the spine in the image coordinate system.

[0130] The positional label map of each individual vertebra of the spine in the image coordinate system is transformed into the positional label map of each individual vertebra of the spine in the volume coordinate system, and this is used as the label of the second network model.

[0131] In step S330, the second network model is trained based on the labels of the second dataset and the second network model.

[0132] For example, in step S330, a second network model is set up and trained using the second dataset and the labels of the second network model.

[0133] In a specific embodiment of this application, after setting up a second network model, the spinal CT images from the second dataset are input into the second network model, and the labeled image of each individual vertebra of the spine in the volume coordinate system is used as the label. The second network model outputs the position information of each individual vertebra in the spinal CT image within the spine. The position of each individual vertebra in the spinal CT image within the spine can be seen as follows: Figure 8 The image shows a rendering of the location information of a single vertebra within the spine.

[0134] Figure 9 A flowchart illustrating the training process of a third network model according to an example embodiment of this application is shown.

[0135] like Figure 9 As shown, in step S410, the third dataset is obtained.

[0136] For example, in step S410, a third dataset is obtained from the public dataset according to a preset specification for training a third network model.

[0137] In a specific embodiment of this application, a third dataset is obtained from a public dataset. The third dataset includes multi-category spinal CT images, standard position information of each individual vertebra in the spinal CT images, and segmentation label information of each segment of the spine in the spinal CT images. The standard position information of each individual vertebra and the segmentation label information of each segment of the spine in the spinal CT images are position information in the image coordinate system.

[0138] According to some embodiments, the multi-category spinal CT images included in the third dataset correspond to spinal CT images containing each segment of the complete cervical vertebrae; spinal CT images containing each segment of the complete thoracic vertebrae; and spinal CT images containing each segment of the complete cervical vertebrae.

[0139] In step S420, the labels of the third network model are obtained based on the third dataset.

[0140] For example, in step S420, the spinal CT images in the third dataset are converted into spinal images in an image coordinate system. Based on the segmentation label information of each segment of the spine, the spinal images in the image coordinate system are segmented to obtain images of each individual vertebra. Pixel processing is performed on the images of each individual vertebra, and these pixel-processed images of the individual vertebrae are used as labels for the third network model.

[0141] In a specific embodiment of this application, after converting the spinal CT images in the third dataset into spinal images in the image coordinate system, the position information of each individual vertebra in the spinal image in the image coordinate system is obtained based on the segmentation label information of each segment of the spine.

[0142] According to some embodiments, the position information of each individual vertebra in the spine image in the image coordinate system can be obtained using the `numpy.where()` method of the NumPy scientific computing library. The position coordinates of an individual vertebra can be obtained in the following way.

[0143] position=[[min_z,max_z],[min_y,max_y],[min_x,max_x]]

[0144] Where min represents the minimum position coordinate of a single vertebra, and max represents the maximum position coordinate of a single vertebra.

[0145] Based on the positional information of each individual vertebra, an image of each individual vertebra in the image coordinate system is obtained.

[0146] In the image of each individual vertebra in the image coordinate system, the pixel values ​​of the regions other than the vertebral region are set to zero. The image with zero pixel values ​​is then superimposed on the image of each individual vertebra in the image coordinate system to generate a label image of the individual vertebra in the image coordinate system.

[0147] The label map of a single vertebra in the image coordinate system is transformed into a label map of a single vertebra in the volume coordinate system, and this is used as the label of the third network model.

[0148] In step S430, the third network model is trained based on the labels of the third dataset and the third network model.

[0149] For example, in step S430, a third network model is set up and trained using a third dataset and the labels of the third network model.

[0150] In a specific embodiment of this application, the image of each vertebra in the image coordinate system obtained in step S420 is transformed into the image of each vertebra in the volume coordinate system, and used as input data for the third network model.

[0151] According to some embodiments, the voxel spacing and direction of an image of a single vertebra in body coordinates are consistent with the voxel spacing and direction of the spinal CT images in a third dataset.

[0152] First, the image size of a single vertebra in volume coordinates is calculated. The image size of a single vertebra in volume coordinates is obtained by multiplying the size of the single vertebra image in each dimension by the voxel spacing of the spinal CT images in the third dataset. This result is denoted as new. size .

[0153] Based on the image dimensions of a single vertebra in volume coordinates, the position of the image center point of that vertebra in volume coordinates is calculated using the following formula. The image center point of a single vertebra in volume coordinates is labeled as `origin`.

[0154] origin[i]=direction[i][0]*new size[0] +direction[i][1]*new size[1]

[0155] +direction[i][2]*new size[2]

[0156] Where i = 1, 2, 3, and direction is the orientation of the spinal CT image in the third dataset.

[0157] Based on the obtained voxel spacing and orientation, and the calculated image center point position, the image of each individual vertebra in the image coordinate system is transformed into an image of each individual vertebra in the volume coordinate system. The image of each individual vertebra in the volume coordinate system can be found in [reference needed]. Figure 10 The diagram shows a single vertebral segment in volume coordinates.

[0158] Based on the standard position information of each individual vertebra, the position information of each individual vertebra in the image of each individual vertebra in the image coordinate system is updated.

[0159] According to some embodiments, the standard position information of a single vertebra in the third dataset is denoted as spine_pos(x,y,z), and the position information of a single vertebra in the updated image of the single vertebra is denoted as spine_pos(x1,y1,z1). The position information of a single vertebra in the updated image of the single vertebra is calculated using the following formula.

[0160]

[0161] Where min represents the minimum position coordinate of a single vertebra, and max represents the maximum position coordinate of a single vertebra.

[0162] Generate a zero-pixel image with the same image size as each individual vertebra in the image coordinate system.

[0163] Based on the updated position information of each individual vertebra, the position of each individual vertebra is marked in the zero-pixel image to obtain a position-marked image of the individual vertebra in the image coordinate system.

[0164] According to some embodiments, the point corresponding to the updated position coordinates of a single vertebra can be used as the center of a sphere. A small sphere with a radius within a preset range (which can be adjusted according to actual needs, for example, the preset range is [1,7]) can be rendered, replacing the point corresponding to the position coordinates of the single vertebra in the zero-pixel image. This sphere serves as the position marker for each single vertebra in the zero-pixel image, and the pixel label value of the small sphere is set to 1. The image of each single vertebra in the image coordinate system is then overlaid with the zero-pixel image that has been marked with the position of the single vertebra, to obtain the position marker image of the single vertebra in the image coordinate system.

[0165] The positional label images of a single vertebra in the image coordinate system are transformed into positional label images of a single vertebra in the volume coordinate system, which are then used as another input data for the third network model.

[0166] The image of each individual vertebra in the volume coordinate system and the image with the position label of each individual vertebra in the volume coordinate system are superimposed and input into the third network model, and the third network model is trained according to the label of the third network model.

[0167] Figure 11 A block diagram of a segmentation device for a single vertebra according to an example embodiment of this application is shown.

[0168] like Figure 11 As shown, the segmentation device 500 includes a data acquisition module 510, a data processing module 520, and a data output module 530.

[0169] The data acquisition module 510 is used to acquire raw CT images.

[0170] According to some embodiments, the data acquisition module 510 is further configured to acquire a first dataset for training a first network model from a public dataset, acquire second data for training a second network model, and acquire a third dataset for training a third network model.

[0171] The data processing module 520 constructs the first network model and trains the first network model.

[0172] The data processing module 520 obtains the segmentation results of the spine, sacrum and ribs in the original CT image through the first network model, and performs morphological dilation on the spine in the segmented image of the original CT image according to the segmentation results to obtain the morphologically dilated spine image and the first spine boundary information corresponding to the morphologically dilated spine image.

[0173] The data processing module 520 constructs a second network model and trains the second network model.

[0174] The data processing module 520 obtains the position information of a single vertebra in the spine from the original CT image through the second network model, namely the position information of the first vertebra.

[0175] Based on the first vertebral boundary information and the first vertebral position information, the data processing module 520 determines the boundary information of a single vertebral segment, and based on the boundary information of the single vertebral segment, obtains the initial image of the single vertebral segment from the first vertebral image in the image coordinate system obtained by transforming the morphologically dilated vertebral image.

[0176] After determining the second spinal boundary information corresponding to the initial image of a single vertebra, the data processing module 520 updates the position information of the single vertebra in the initial image of the single vertebra, and thereby obtains the position marker of the single vertebra in the initial image of the single vertebra.

[0177] The data processing module 520 converts the initial image of a single vertebra containing the location markers of the single vertebra into a location marker image of the single vertebra in a volume coordinate system, and converts the initial image of the single vertebra into an initial image of the single vertebra in a volume coordinate system.

[0178] The data processing module 520 constructs a third network model and trains the third network model.

[0179] The data processing module 520 inputs the initial image of a single vertebra in volume coordinates and the position marker image of a single vertebra in volume coordinates into the third network model in a preset order, and obtains the updated image of the single vertebra through the third network model. Furthermore, the data processing module 520 converts the updated image of the single vertebra in volume coordinates obtained through the third network model into an updated image of the single vertebra in image coordinates.

[0180] The data processing module 520 converts the original CT image into an original image in the image coordinate system, and determines each individual vertebra in the original image and marks it in a preset order based on the updated image of the individual vertebra in the image coordinate system.

[0181] The data output module 530 outputs the original image of each individual vertebra as the corresponding reconstructed image of the original CT image.

[0182] Figure 12 A block diagram of an electronic device according to an example embodiment of this application is shown.

[0183] like Figure 12 As shown, the electronic device 600 is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0184] like Figure 12 As shown, the electronic device 600 is manifested in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different system components (including the storage unit 620 and the processing unit 610), a display unit 640, etc. The storage unit stores program code, which can be executed by the processing unit 610, causing the processing unit 610 to perform the methods described in this specification according to the various exemplary embodiments of this application. For example, the processing unit 610 can perform, for example... Figure 1 The method shown.

[0185] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 6201 and / or cache memory 6202, and may further include a read-only memory (ROM) 6203.

[0186] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0187] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0188] Electronic device 600 can also communicate with one or more external devices 700 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0189] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. The technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this application.

[0190] Software products may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections with one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0191] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0192] Program code for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0193] The aforementioned computer-readable medium carries one or more programs, which, when executed by a device, cause the computer-readable medium to perform the aforementioned functions.

[0194] Those skilled in the art will understand that the above modules can be distributed in the device as described in the embodiments, or they can be modified accordingly and placed in one or more devices that are unique to this embodiment. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0195] According to some embodiments of this application, the technical solution of this application can perform single-segment segmentation of the spine through a convolutional network model while ensuring speed and accuracy, so as to obtain images of single vertebrae corresponding to the spine, thereby improving the accuracy of spine segmentation.

[0196] The embodiments of this application have been described in detail above. These descriptions are solely for the purpose of helping to understand the method and core ideas of this application. Furthermore, any changes or modifications made by those skilled in the art based on the ideas of this application, its specific implementation methods, and its application scope, are all within the scope of protection of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for segmenting a single vertebra, characterized in that, include: Obtain first spinal boundary information, which is the boundary information of the spine in the segmentation result obtained after image segmentation of the original CT image; Obtain first vertebral position information, which is the position information of a single vertebra in the spine in the original CT image; Based on the first spinal boundary information and the first vertebral position information, an initial image of a single vertebra is obtained; The position of the single vertebra in the initial image of the single vertebra is marked to generate an updated image of the single vertebra; Based on the updated image of the single vertebra, determine the single vertebra in the restored image corresponding to the original CT image; Specifically, determining the single vertebra in the restored image corresponding to the original CT image based on the updated image of the single vertebra includes: The original CT image is converted into an original CT image in an image coordinate system; Obtain the zero-pixel image corresponding to the original CT image in the image coordinate system; Based on the updated image of the single vertebra in the image coordinate system, the single vertebra is marked in a preset order in the zero-pixel image corresponding to the original CT image in the image coordinate system. Based on the zero-pixel image corresponding to the original CT image in the image coordinate system with the marked individual vertebrae, obtain the position marker of the individual vertebrae in the original CT image in the image coordinate system. Based on the updated image of the single vertebra in the image coordinate system, update the boundary information of the single vertebra in the original CT image in the image coordinate system of the marked single vertebra, so as to determine the single vertebra in the original CT image in the image coordinate system. The original CT image in the image coordinate system of the identified single vertebra is determined as the restored image corresponding to the original CT image.

2. The method according to claim 1, characterized in that, Obtain the first spinal boundary information, including: Construct the first network model; The segmentation results of the spine, sacrum and ribs in the original CT image are obtained through the first network model to obtain the segmented image of the original CT image; The spine in the segmented image is morphologically dilated to obtain a morphologically dilated spine image; The morphologically dilated spine image is transformed into a first spine image in the image coordinate system; Obtain the boundary information of the first spine corresponding to the first spine image.

3. The method according to claim 1, characterized in that, Obtain the location information of the first vertebra, including: Construct a second network model; The second network model is used to obtain the first vertebral position information from the original CT image.

4. The method according to claim 2, characterized in that, Based on the first spinal boundary information and the first vertebral position information, an initial image of a single vertebra is obtained, including: Based on the first spinal boundary information and the first vertebral position information, the boundary information of a single vertebral segment is determined; Based on the boundary information of the single vertebral segment, an initial image of the single vertebral segment is obtained from the first spinal image.

5. The method according to claim 4, characterized in that, Marking the position of the single vertebra in the initial image of the single vertebra to generate an updated image of the single vertebra, including: Determine the second spinal boundary information, which is the spinal boundary information in the initial image of the single vertebra; Based on the second spinal boundary information, update the position information of the single vertebra in the initial image of the single vertebra; Based on the updated position information of the individual vertebrae, obtain the position marker of the individual vertebrae in the initial image of the individual vertebrae; The initial image of the single vertebra, which includes the location markers of the single vertebra, is converted into a location marker image of the single vertebra in volume coordinates. An updated image of the single vertebra is generated based on the initial image of the single vertebra and the position marker image of the single vertebra.

6. The method according to claim 5, characterized in that, Based on the initial image of the single vertebra and the location marker image of the single vertebra, an updated image of the single vertebra is generated, including: Construct a third network model; The initial image of the single vertebra is transformed into an initial image of the single vertebra in volume coordinates. The initial image of a single vertebra in the body coordinate system and the position marker image of the single vertebra are input into the third network model in the preset order. The updated images of the single vertebral segment are obtained through the third network model in the preset order; The updated image of the single vertebra is transformed into an updated image of the single vertebra in an image coordinate system.

7. A segmentation device for a single vertebra, characterized in that, include: The data acquisition module acquires the raw CT images; The data processing module sets up a first network model, a second network model, and a third network model; it obtains segmentation results of the spine, sacrum, and ribs from the original CT image using the first network model; it obtains first spinal boundary information based on the segmentation results; it obtains first vertebral position information from the original CT image using the second network model; and it obtains an initial image of a single vertebral segment based on the first spinal boundary information and the first vertebral position information. The position of the single vertebra in the initial image of the single vertebra is marked; based on the initial image of the single vertebra and the position marked image of the single vertebra, an updated image of the single vertebra is generated through the third network model; based on the updated image of the single vertebra, the single vertebra in the restored image corresponding to the original CT image is determined; The data output module outputs the reconstructed image corresponding to the original CT image after determining a single vertebral segment; The data processing module is also used for The original CT image is converted into an original CT image in an image coordinate system; Obtain the zero-pixel image corresponding to the original CT image in the image coordinate system; Based on the updated image of the single vertebra in the image coordinate system, the single vertebra is marked in a preset order in the zero-pixel image corresponding to the original CT image in the image coordinate system. Based on the zero-pixel image corresponding to the original CT image in the image coordinate system with the marked individual vertebrae, obtain the position marker of the individual vertebrae in the original CT image in the image coordinate system. Based on the updated image of the single vertebra in the image coordinate system, update the boundary information of the single vertebra in the original CT image in the image coordinate system of the marked single vertebra, so as to determine the single vertebra in the original CT image in the image coordinate system. The original CT image in the image coordinate system of the identified single vertebra is determined as the restored image corresponding to the original CT image.

8. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the method as described in any one of claims 1-6.

Images