CN112233083B

CN112233083B - Spine detection method, spine detection device, electronic equipment and storage medium

Info

Publication number: CN112233083B
Application number: CN202011092026.5A
Authority: CN
Inventors: 李丙生; 何薇
Original assignee: Shenyang Advanced Medical Equipment Technology Incubation Center Co ltd
Current assignee: Neusoft Medical Systems Co Ltd
Priority date: 2020-10-13
Filing date: 2020-10-13
Publication date: 2024-01-02
Anticipated expiration: 2040-10-13
Also published as: CN112233083A

Abstract

The embodiment of the invention provides a spine detection method, a spine detection device, electronic equipment and a storage medium. According to the embodiment of the invention, the three-dimensional spine image is input into the deep learning network model to obtain the segmentation image output by the deep learning network model, the first central point of the spine region is determined on each cross-sectional image corresponding to the three-dimensional spine image according to the segmentation image, the spine central line is obtained according to the first central point on the cross-sectional image, each spine body of the spine is identified according to the segmentation image and the spine central line, the spine body identification is set for each spine body, the spine sagittal image is positioned according to the segmentation image and the identified spine body, the first spine parameter value is determined according to the spine sagittal image, the spine parameter can be fully automatically measured, and the working efficiency is improved.

Description

Spine detection method, spine detection device, electronic equipment and storage medium

Technical Field

The present invention relates to the field of medical image processing technologies, and in particular, to a spine detection method, apparatus, electronic device, and storage medium.

Background

In clinical practice, a clinician with abundant theoretical knowledge and practical experience is usually required to manually measure scoliosis and spinal stenosis from a CT image of a patient, which consumes time and effort of the clinician and is inefficient.

Disclosure of Invention

In order to overcome the problems in the related art, the invention provides a spine detection method, a spine detection device, electronic equipment and a storage medium, and the working efficiency of spine parameter measurement is improved.

According to a first aspect of an embodiment of the present invention, there is provided a spine detection method including:

inputting a three-dimensional spine image into a deep learning network model to obtain a segmented image output by the deep learning network model, wherein at least spine and spine tube are marked in the segmented image;

determining a first center point of a vertebra region on each cross-sectional image corresponding to the three-dimensional vertebra image according to the segmentation image;

obtaining a spinal centerline from a first center point on the cross-sectional image;

identifying each vertebral body of the vertebra according to the segmentation image and the vertebra central line, and setting a vertebral body mark for each vertebral body;

positioning a spinal sagittal image according to the segmented image and the identified vertebral body;

a first vertebral parameter value is determined from the sagittal image.

According to a second aspect of embodiments of the present invention, there is provided a spine detection device comprising:

the segmentation module is used for inputting the three-dimensional spine image into the deep learning network model to obtain a segmentation image output by the deep learning network model, wherein at least the spine and the spinal canal are marked in the segmentation image;

The first center point determining module is used for determining a first center point of a vertebra region on each cross-sectional image corresponding to the three-dimensional vertebra image according to the segmentation image;

the spine center line acquisition module is used for acquiring a spine center line according to the first center points on all the cross-section images;

the vertebral body identification module is used for identifying each vertebral body of the vertebra according to the segmentation image and the vertebra central line and setting a vertebral body mark for each vertebral body;

the positioning module is used for positioning a vertebra sagittal image according to the segmentation image and the identified vertebral body;

and the first parameter value determining module is used for determining a first vertebra parameter value according to the vertebra sagittal image.

According to a third aspect of an embodiment of the present invention, there is provided an electronic apparatus including: an internal bus, and a memory, a processor and an external interface connected by the internal bus, wherein:

the memory is used for storing machine-readable instructions corresponding to the spine detection logic;

the processor is configured to read machine-readable instructions on the memory and execute the instructions to implement the method of any of the first aspects.

According to a fourth aspect of embodiments of the present invention, there is provided a computer readable storage medium having stored thereon a computer program, wherein the program when executed by a processor implements the method of any of the first aspects.

The technical scheme provided by the embodiment of the invention can have the following beneficial effects:

according to the embodiment of the invention, the three-dimensional spine image is input into the deep learning network model to obtain the segmentation image output by the deep learning network model, the first central point of the spine region is determined on each cross-sectional image corresponding to the three-dimensional spine image according to the segmentation image, the spine central line is obtained according to the first central point on the cross-sectional image, each spine body of the spine is identified according to the segmentation image and the spine central line, the spine body identification is set for each spine body, the spine sagittal image is positioned according to the segmentation image and the identified spine body, the first spine parameter value is determined according to the spine sagittal image, the spine parameter can be fully automatically measured, the working efficiency is improved, the workload of doctors is reduced, and the labor and time consumption are saved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the specification and together with the description, serve to explain the principles of the specification.

Fig. 1 is a flowchart illustrating a method for detecting a vertebra according to an embodiment of the present invention.

Fig. 2 is an exemplary view of the spinal centerline and the posterior region of the skull provided by an embodiment of the present invention.

Fig. 3 is a schematic illustration of measurements of the sagittal diameter of the spinal canal and the transverse diameter of the spinal canal.

Fig. 4 is a schematic view of the corner points of the spine, the midpoint of the anterior edge and the midpoint of the posterior edge of the disc.

Fig. 5 is a schematic view of the horizontal tilt of the disc.

Figure 6 is a schematic view of a lumbar anterior lobe.

Fig. 7 is a schematic view of the lumbosacral angle.

Fig. 8 is a schematic view of a sacral midline.

Fig. 9 is a schematic view of Cobb angle.

Fig. 10 is a schematic view of an articular process joint angle.

Fig. 11 is a gray scale image of the medial aspect of the vertebral body.

Fig. 12 is a functional block diagram of a spine detection device according to an embodiment of the present invention.

Fig. 13 is a hardware configuration diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the invention. Rather, they are merely examples of apparatus and methods consistent with aspects of embodiments of the invention as detailed in the accompanying claims.

The terminology used in the embodiments of the invention is for the purpose of describing particular embodiments of the invention only and is not intended to be limiting of embodiments of the invention. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any or all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, third, etc. may be used in embodiments of the present invention to describe various information, these information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, the first information may also be referred to as second information, and similarly, the second information may also be referred to as first information, without departing from the scope of embodiments of the present invention. The word "if" as used herein may be interpreted as "at … …" or "at … …" or "responsive to a determination", depending on the context.

The research of the spine image has important significance for revealing the pathological changes of the spine and the intraspinal canal, and can also provide assistance for actual clinical diagnosis.

Scoliosis is common in recent years, and corrective surgery treatment is still the most effective way among various methods for treating malformed vertebrae, but the risk of such surgery is high. Before the operation, a doctor firstly analyzes the spine morphology of a patient according to various medical images, determines the direction of the spine deformation and the displacement of the deformed spine, analyzes and estimates the cross-sectional area of a spinal canal, and ensures the safe spinal correction posture.

The spinal canal stenosis causes diseases such as cervical and lumbar disc herniation, spinal tumor, spinal cord injury and the like, causes severe injury to spinal cord and neuromyelination tube, causes compression to nerve roots (spinal cord), causes weakness or numbness of legs, shanks and buttocks of a patient, causes great pain to the patient, and even sometimes, the spinal canal stenosis can cause the patient to lose control over some body functions of the patient, so that the patient can feel painful and simultaneously has great influence on life and work of the patient.

Minimally invasive human treatment of lumbar intervertebral disc is becoming the main development direction of treatment of prolapse of intervertebral disc due to the advantages of slight trauma, safety, effectiveness, rapid postoperative recovery, repeatable treatment and the like, and the intervertebral disc puncture path should pass through the middle horizontal plane of the intervertebral disc to avoid serious complications caused by injury of upper and lower end plates of the intervertebral disc during treatment, so the measurement of the horizontal inclination angle of the lumbar intervertebral disc has important clinical significance.

The spine detection method provided by the embodiment of the invention is used for realizing full-automatic spine parameter measurement and improving the working efficiency.

The spine detection method will be described in detail by way of examples.

Fig. 1 is a flowchart illustrating a method for detecting a vertebra according to an embodiment of the present invention. As shown in fig. 1, in this embodiment, the spine detection method may include:

s101, inputting a three-dimensional spine image of a detected object into a trained deep learning network model, and obtaining a segmented image output by the deep learning network model, wherein at least spine and spine tube are marked in the segmented image.

S102, determining a first center point of the spine region on each cross-sectional image corresponding to the three-dimensional spine image according to the segmented image.

S103, obtaining the spine center line according to the first center point on the cross-sectional image.

And S104, identifying each vertebral body of the vertebra according to the segmentation image and the vertebra central line, and setting a vertebral body mark for each vertebral body.

S105, positioning a vertebra sagittal image according to the segmentation image and the identified vertebral body.

And S106, determining a first vertebral parameter value according to the vertebral sagittal image.

In this embodiment, the three-dimensional spine image covers a region of interest to be studied, such as a human spine, a spinal canal, ribs, etc.

In this embodiment, the deep learning network model may be obtained by:

setting a deep learning network model and setting initial parameter values of the deep learning network model;

obtaining a plurality of sets of training data, wherein each set of training data comprises an input image and a label image, the input image is a three-dimensional spine image, and the label image is a segmentation image marked with target tissues such as spine, rib, spine tube and the like;

and training the deep learning network model by utilizing the training data to obtain a trained deep learning network model.

In one example, the structure of the set deep learning network model may be: detection and localization network + splitting network, such as RCNN (Region-CNN) network + U-Net network. The RCNN network may be replaced by other networks, such as CNN (Convolutional Neural Networks, convolutional neural network), RNN (Recurrent Neural Network ), GAN (Generative Adversarial Networks), and generative countermeasure network. Of course, this is merely an example and is not intended to limit the present application, and in practical applications, the structure of the deep learning network model that can be employed is not limited to this example.

Wherein the detection and localization network (e.g., RCNN network) is used to detect and localize the spine, spinal canal, and may also be used to detect and localize ribs; the segmentation network is used for segmenting and marking the spine and the vertebral canal, and can also be used for segmenting and marking the ribs.

The training process of the deep learning network model may be:

in the training process, the parameter value of the deep learning network model corresponding to the 1 st group of training data is the initial parameter value, the parameter value of the deep learning network model corresponding to the j th group of training data is the parameter value regulated after the j-1 st group of training data is trained, j is a natural number, and j is more than or equal to 2; for each set of training data, the following is performed:

Inputting an input image in the set of training data into a deep learning network model corresponding to the set of training data to obtain a segmented image corresponding to the set of training data;

acquiring a difference value between the segmented image and a label image in the set of training data;

if the difference value is larger than a preset threshold value, adjusting the parameter value of the deep learning network model according to the difference value; and if the difference value is smaller than or equal to the preset threshold value, stopping training, and taking the deep learning network model corresponding to the training data as a trained deep learning network model.

In this embodiment, the center point of the region refers to the center of gravity of all points within a three-dimensional cube.

In an exemplary implementation, the method may further include:

determining a second center point of the spinal canal area on each cross-sectional image corresponding to the three-dimensional spine image according to the segmented image;

obtaining a centreline of the spinal canal according to a second central point on the cross-sectional image;

a second vertebral parameter value is determined based on at least one of the segmented image, the vertebral centerline, the spinal canal centerline, and the vertebral body.

It should be noted that, in this embodiment, the first vertebra parameter and the second vertebra parameter are names for convenience of description, and both the first vertebra parameter and the second vertebra parameter are vertebra parameters.

After the first central points and the second central points on all the cross-sectional images are obtained, all the first central points on the cross-sectional images are connected in sequence to obtain a spine central line, and all the second central points on the cross-sectional images are connected in sequence to obtain a spine tube central line.

The human vertebra consists of 33 vertebrae (vertebral bodies), including cervical vertebra (including seven vertebrae), thoracic vertebra (including twelve vertebrae), lumbar vertebra (including five vertebrae), sacral vertebra (including five vertebrae), and coccyx (including four vertebrae), and is formed by connecting ligaments, joints, and intervertebral discs. The upper end of the spine supports the skull, the inferior hip bone, the middle rib, and serves as the back wall of the thorax, abdominal cavity, and pelvic cavity. The spine has functions of supporting the torso, protecting the viscera, protecting the spinal cord, and performing exercise. A longitudinal spinal canal is formed in the spinal column from top to bottom, and spinal fluid and soft tissues are arranged in the spinal canal.

In this embodiment, the vertebral body identifier may be a vertebral body number.

For example, the cervical vertebrae may be denoted by the letter C, followed by cervical vertebrae C1-C7 from top to bottom. Similarly, the thoracic vertebral bodies can be represented by T1 to T12, the lumbar vertebral bodies by L1 to L5, the sacral vertebral bodies by S1 to S5, and the coccyx vertebral bodies by Co1 to Co4, respectively. There are 5 lumbar intervertebral discs, namely L1-L2 (intervertebral discs between L1 and L2), L2-L3, L3-L4, L4-L5 and L5-S1.

In one exemplary implementation, identifying individual vertebral bodies of a vertebra from the segmented image and the vertebra centerline may include:

determining a occipital macropore center point on a vertebra according to the segmented image and the vertebra center line;

and identifying each vertebral body of the vertebra according to the occipital macropore center point.

In one exemplary implementation, determining the occipital macro-aperture center point on the spine from the segmented image and the spine centerline may include:

generating a curved surface unfolding image of the spine according to the front-back direction of the spine according to the spine central line;

identifying a maximum communicated bone region positioned at the rear side of the spine central line in the curved surface expansion image, and marking the maximum communicated bone region as a skull rear side region;

determining a skull bottom contour line according to the skull rear side area;

finding out a target point closest to the spine center line on the skull base contour line; determining a projection point of the target point onto the spine central line as an occipital macropore central point on the spine;

identifying each vertebral body of the vertebra based on the occipital macropore center point, comprising:

determining the central point of each vertebral body of the vertebra according to the segmentation image, and marking the central point as the central point of the vertebral body;

Obtaining the distance value between the central point of each vertebral body and the central point of the occipital macropore;

sequencing each vertebral body according to the distance value between the central point of the vertebral body and the central point of the occipital macropore to obtain a first sequencing result;

and identifying each vertebral body according to the first sequencing result and the vertebral body structure.

Fig. 2 is an exemplary view of the spinal centerline and the posterior region of the skull provided by an embodiment of the present invention. In fig. 2, the left image is an image of the spine centerline CPR ((Curved Planar Reformation, curved reconstruction), and the right image is an image of the area of the back side of the skull.

The occipital macropore center point can be used for distinguishing intracranial from extracranial, the occipital macropore center point is upward intracranial and downward extracranial, and each vertebral body of the vertebra can be obtained through blocking.

In an exemplary implementation, the segmented image is further labeled with ribs; identifying individual vertebral bodies of the spine from the segmented image and the spine centerline may include:

sequencing each rib marked in the segmented image according to the position to obtain a second sequencing result;

and respectively identifying vertebral bodies corresponding to the ribs according to the second sequencing result.

The vertebral bodies of the vertebra are communicated with the ribs, each thoracic vertebral body is usually connected with one rib in the segmented image, and the ribs have obvious up-down sequence relation. Thus, in this embodiment, the order of the individual vertebral bodies is determined according to the ordering of the ribs, thereby identifying the vertebral bodies.

The second ordering result refers to the up-down ordering relation of the ribs.

For example, from top to bottom, the 1 st rib region corresponds to thoracic vertebra T1, and so on, the 12 th rib region corresponds to thoracic vertebra T12, the upward cervical vertebrae of thoracic vertebra T1 are sequentially C7 to C1, the downward cervical vertebrae of thoracic vertebra T12 are sequentially lumbar vertebrae L1 to L5, sacral vertebrae S1 to S5, and coccyx Co1 to Co4.

In one exemplary implementation, locating the sagittal image of the spine based on the segmented image and the identified vertebral body may include:

finding lumbar vertebrae from the identified vertebral bodies on the segmented image;

acquiring an X-axis coordinate value corresponding to the central point of each lumbar region;

determining the average value of X-axis coordinate values corresponding to the central points of the areas of all lumbar vertebrae;

and positioning a vertebra sagittal image in the segmented image according to the average value.

For example, assuming that the X-axis coordinate values of the lumbar vertebrae L1 to L5 are X1 to X5, respectively, an average value of X1 to X5 is calculated, and a spinal sagittal image is located based on the average value.

In one exemplary implementation, the first spinal parameter includes a horizontal tilt of the intervertebral disc;

determining a first spinal parameter value from the spinal sagittal image may include:

On the vertebra sagittal image, according to each cone mark, respectively determining the front edge upper corner point, the front edge lower corner point, the rear edge upper corner point and the rear edge lower corner point of the corresponding cone;

for any intervertebral disc, determining the midpoint of the front edge of the intervertebral disc according to the lower front edge corner of the upper vertebral body of the intervertebral disc and the upper front edge corner of the lower vertebral body of the intervertebral disc; determining a trailing edge midpoint of the intervertebral disc according to a trailing edge lower corner point of an upper vertebral body of the intervertebral disc and a trailing edge upper corner point of a lower vertebral body of the intervertebral disc;

acquiring an included angle value of a connecting line of a midpoint of the front edge of the intervertebral disc and a midpoint of the rear edge of the intervertebral disc and a horizontal line, and taking the included angle value as an angle value of a horizontal inclination angle of the intervertebral disc;

the intervertebral disc comprises a lumbar disc.

According to the embodiment, the horizontal inclination angle of the lumbar intervertebral disc can be automatically measured, and a basis is provided for relevant diagnosis of the lumbar intervertebral disc.

Fig. 4 is a schematic view of the corner points of the spine, the midpoint of the anterior edge and the midpoint of the posterior edge of the disc. In fig. 4, at, ad, pt, pd are respectively a leading edge upper corner point, a leading edge lower corner point, a trailing edge upper corner point, and a trailing edge lower corner point. A. B are the mid-point of the anterior and the mid-point of the posterior edge of the disc, respectively.

According to the present embodiment, four corner points At1, ad1, pt1, pd1 of the lumbar intervertebral discs L1 to L2 may be obtained; four corner points At2, ad2, pt2 and Pd2 of the lumbar intervertebral discs L2-L3; four corner points At3, ad3, pt3 and Pd3 of the lumbar intervertebral discs L3-L4; four corner points At4, ad4, pt4 and Pd4 of the lumbar intervertebral discs L4-L5; four corner points At5, ad5, pt5 and Pd5 of the lumbar intervertebral discs L5-S1. On this basis, the midpoint of the anterior and posterior edges of the disc can be obtained.

For example, lumbar intervertebral discs L1 to L2 are taken as an example. The mid-points Ao 1-2 of the front edges of the lumbar intervertebral discs L1-L2 are as follows: ao1.about.2= (Ad 1+At2)/2, and trailing edge midpoint Po 1.about.2 is: po 1-2= (pd1+pt2)/2. By analogy, a series of leading edge midpoints Ao 1-2, ao 2-3, ao 3-4, ao 4-5, ao 5-S1 and a series of trailing edge midpoints Po 1-2, po 2-3, po 3-4, po 4-5, po 5-S1 can be obtained.

Fig. 5 is a schematic view of the horizontal tilt of the disc. In fig. 5, the solid line is the line connecting the midpoint of the anterior edge and the midpoint of the posterior edge of the disc, the broken line is the horizontal line, and the angle between the solid line and the broken line is the horizontal inclination angle of the disc. As can be seen from fig. 5, the present embodiment can be applied to automatically measure the horizontal tilt of the disc for each disc.

In one exemplary implementation, the first spinal parameter includes TK (Thoracic Kyphosis, thoracic posterior lobe);

on the vertebra sagittal image, determining the upper corner point of the front edge and the upper corner point of the rear edge of the uppermost thoracic vertebra according to the vertebral body mark of the uppermost thoracic vertebra;

determining a first connecting line according to the upper corner point of the front edge and the upper corner point of the rear edge of the uppermost thoracic vertebra;

Determining a front edge lower corner point and a rear edge lower corner point of the lowest thoracic vertebra on the sagittal image of the vertebra and according to the vertebral body identification of the lowest thoracic vertebra;

determining a second connecting line according to the front edge lower corner point and the rear edge lower corner point of the lowest thoracic vertebra;

and acquiring an included angle value of the first connecting line and the second connecting line as an angle value of the thoracic rear lobe.

The embodiment can automatically measure and obtain the angle value of the thoracic kyphosis, provide accurate data for diagnosis and operation preparation related to the vertebra, and assist diagnosis and treatment.

In one exemplary implementation, the first spinal parameter includes LLCobb (Lumar Lordosis, lumbar anterior lobe);

on the vertebra sagittal image, determining the upper corner point of the front edge and the upper corner point of the rear edge of the uppermost lumbar vertebra according to the vertebral body mark of the uppermost lumbar vertebra; determining a third connecting line according to the upper corner point of the front edge and the upper corner point of the rear edge of the uppermost lumbar vertebra; or, on the vertebra sagittal image, determining a plurality of upper edge boundary points of the uppermost lumbar vertebra according to the vertebral body identification of the uppermost lumbar vertebra, and fitting according to the plurality of upper edge boundary points to obtain a third connecting line;

On the vertebra sagittal image, determining the front edge upper corner point and the rear edge upper corner point of the uppermost sacrum according to the vertebral body mark of the uppermost sacrum;

determining a fourth connecting line according to the upper corner point of the front edge and the upper corner point of the rear edge of the uppermost sacral vertebrae;

and acquiring an included angle value of the third connecting line and the fourth connecting line as an angle value of the lumbar anterior lobe.

Figure 6 is a schematic view of a lumbar anterior lobe. As shown in fig. 6, the Angle between the superior border line of the lumbar vertebra L1 and the superior border line of the sacral vertebra S1 is LLCobb (i.e., angle in fig. 6).

In one exemplary implementation, the first spinal parameter includes a lumbosacral angle (Lumbosacral Angle);

and acquiring an included angle value between the fourth connecting line and the horizontal line to serve as an angle value of the lumbosacral angle.

Fig. 7 is a schematic view of the lumbosacral angle. After the human body stands upright, the lumbosacral portion forms a force junction. The lumbosacral angle should generally not exceed 45 °, the greater the lumbosacral angle, the more unstable the spine. According to the embodiment, the angle value of the lumbosacral angle is obtained through automatic measurement, and the stability of the spine can be judged according to the angle value of the lumbosacral angle.

In one exemplary implementation, the first spinal parameter includes a sacral midline (CSVL);

finding out a midpoint of a line segment determined by an upper corner point of the front edge and an upper corner point of the rear edge of the uppermost sacrum;

acquiring a straight line which passes through the midpoint of the line segment and is perpendicular to the horizontal line as a sacrum midwife;

or,

acquiring a plurality of upper edge boundary corner points of the uppermost sacrum on the vertebra sagittal image according to the vertebral body mark of the uppermost sacrum;

determining a midpoint according to coordinate values of a plurality of upper edge boundary corner points of the uppermost sacrum;

a straight line passing through the midpoint and perpendicular to the horizontal line is acquired as a sacral midline.

In this embodiment, the midpoint may be determined by: and respectively averaging the x coordinates, the y coordinates and the z coordinates of a plurality of upper edge boundary corner points of the uppermost sacrum to obtain an x coordinate average value, a y coordinate average value and a z coordinate average value, wherein the x coordinate of a midpoint is equal to the x coordinate average value, the y coordinate of the midpoint is equal to the y coordinate average value, and the z coordinate of the midpoint is equal to the z coordinate average value.

Fig. 8 is a schematic view of a sacral midline. The sacral midline is one of the most important indicators for analysis of scoliosis, and is always perpendicular to the ground from the caudal end to the rostral end.

In one exemplary implementation, the first spinal parameter includes an end vertebra;

determining a first vertebral parameter value from the vertebral sagittal image, comprising:

on the vertebra sagittal image, according to the cone identification of each cone, determining the front edge upper corner point, front edge lower corner point, rear edge upper corner point and rear edge lower corner point of the cone;

for any cone, determining an upper edge line of the cone according to an upper front edge corner and an upper rear edge corner of the cone, and determining a lower edge line of the cone according to a lower front edge corner and a lower rear edge corner of the cone;

respectively acquiring the included angles between the upper edge line of the cone and the lower edge line of other cones, and determining the corresponding included angle range of the cones;

finding out the maximum included angle range from the included angle ranges corresponding to all the vertebral bodies;

finding out vertebral bodies at two ends corresponding to the maximum included angle range, and determining the vertebral body at the upper end of the vertebral bodies at the two ends as an upper-end vertebral body with lateral curvature; and determining the vertebral bodies positioned at the lower ends of the vertebral bodies at the two ends as lateral bent lower-end vertebral bodies.

The embodiment can automatically measure and obtain the upper end vertebrae and the lower end vertebrae of the scoliosis, and provide necessary reference data for diagnosis and treatment preparation of the scoliosis.

In one exemplary implementation, the first spinal parameter includes a top vertebra; the method further comprises the steps of:

acquiring central points of all vertebral bodies between the upper vertebral body and the lower vertebral body as target central points according to the segmentation image;

respectively determining the distance from each target center point to the midsacrum plumb line;

and finding out the maximum distance from the center points of all the targets to the midsacrum plumb line, and determining the vertebral body corresponding to the maximum distance as the apical vertebra.

The embodiment can automatically measure and determine the top vertebra in the scoliosis, and provide necessary reference data for diagnosis and treatment preparation of the scoliosis.

In one exemplary implementation, the spinal parameters include Cobb angle; the method further comprises the steps of:

determining an upper edge line of the upper end vertebra and a lower edge line of the lower end vertebra in the segmented image;

a perpendicular line from the upper edge line to the lower edge line is marked as a first perpendicular line; a perpendicular line from the lower edge line to the upper edge line is marked as a second perpendicular line;

And acquiring an included angle value of the first vertical line and the second vertical line to be used as an angle value of the Cobb angle.

Fig. 9 is a schematic view of Cobb angle. Cobb angle is one of the reference criteria for scoliosis severity and characterizes the magnitude of the scoliosis angle.

In other embodiments, for larger lateral curves, the angle between the superior line of the superior vertebra and the inferior line of the inferior vertebra may be directly calculated as Cobb angle.

In one exemplary implementation, the second spinal parameter includes a spinal length;

determining a second vertebral parameter value based on at least one of the segmented image, the vertebral centerline, the spinal canal centerline, and the vertebral body may include:

acquiring sampling points on the spine central line, wherein the sampling points are intersection points of the spine central line and each cross-sectional image corresponding to the three-dimensional spine image;

acquiring a distance value between the corresponding adjacent sampling points according to coordinates of the two adjacent sampling points in the sampling points;

and multiplying the accumulated sum of the distance values between all adjacent sampling points in the sampling points by the pixel physical unit of the image to obtain the spine length value.

The length of the spine is in millimeters (mm).

In one exemplary implementation, the second spinal parameter includes a spinal sagittal diameter and a spinal transverse diameter;

determining an axial position image of the middle layer of each cone corresponding to each cone according to the segmentation image and the central line of the cone;

acquiring a minimum distance value between a front boundary point and a rear boundary point of the vertebral canal in the range of the vertebral canal on the axial position image, and taking the minimum distance value as a length value of a sagittal diameter of the vertebral canal;

obtaining a maximum distance value between a left boundary point and a right boundary point of the vertebral canal in the vertebral canal range on the axial position image, and taking the maximum distance value as a length value of the transverse diameter of the vertebral canal;

and determining the length ratio of the transverse diameter to the sagittal diameter according to the length value of the sagittal diameter of the vertebral canal and the length value of the transverse diameter of the vertebral canal.

The ratio of the length of the transverse to sagittal diameters is equal to the length of the transverse diameter divided by the length of the sagittal diameter.

The bone spinal canal stenosis refers to a disease which causes the reduction of the spinal canal diameter line and the reduction of the effective area of the section to generate corresponding spinal nerve dysfunction due to various reasons, and CT measurement of the bone spinal canal stenosis is mainly carried out on the sagittal diameter and the transverse diameter of the bone spinal canal. The sagittal diameter of the spinal canal is the shortest distance from the posterior edge of the vertebral body to the basal line of the spinous process.

FIG. 3 is a schematic view of the measurement of sagittal diameter of the spinal canal and transverse diameter of the spinal canal, and in FIG. 3, the left line a is the sagittal diameter of the level spinal canal and the line b is the transverse diameter of the level spinal canal; the dashed line on the right represents the axial image position of the middle level of a certain vertebral body.

According to the measured length values of the sagittal diameter and the transverse diameter of the vertebral canal, whether the vertebral canal stenosis exists or not can be judged. There are two ways to determine spinal stenosis: one is to compare the sagittal diameter with a preset threshold, for example, if the sagittal diameter of the cervical spine is less than 10mm, and if the sagittal diameter of the lumbar spine is less than 15mm, it is considered to be narrow; another is based on the Pavlov ratio, pavlovratio=b/a, and when the ratio is less than 0.75, it can be determined that the cervical canal is narrow.

In one exemplary implementation, the second spinal parameter includes a lumbar facet joint angle;

acquiring the upper edge layer cross section of each lumbar vertebra according to the segmentation image and the vertebral body identification of each lumbar vertebra;

determining a centrum central line on the cross section of the upper edge layer, wherein the centrum central line is a connecting line of a first central point of a lumbar vertebra and a second central point of a vertebral canal;

Determining left front and back vertex connecting lines and right front and back vertex connecting lines of articular surfaces on the cross section of the upper edge layer;

acquiring an included angle of a connecting line between the centrum midline and the front and rear vertexes of the left side of the articular process joint surface as a first included angle, and acquiring an included angle of a connecting line between the centrum midline and the front and rear vertexes of the right side of the articular process joint surface as a second included angle;

and determining the average value of the first included angle and the second included angle as the angle value of the articular process joint angle of the corresponding lumbar vertebra.

Fig. 10 is a schematic view of an articular process joint angle. The left view of fig. 10 shows a superior limbic cross-section of the lumbar spine, where L1 is the centrum midline, L2 is the left anterior-posterior apex line of the articular surface, and L3 is the right anterior-posterior apex line of the articular surface. In the right view of fig. 10, the position indicated by the solid line is the position of the upper lamina cross-section of the lumbar spine shown in the left view.

The change of the structure of the lumbar articular process is a main cause of the instability of lumbar vertebra and a plurality of lumbar lesions, and when the angle of the articular process is reduced, the stability of the vertebral body is correspondingly reduced, so that the measurement of the angle of the lumbar articular process is of great significance. The embodiment can automatically measure the angle value of the lumbar articular process joint angle and assist in searching the cause of lumbar pathological changes.

In one exemplary implementation, the spinal parameter includes bone density;

determining a second vertebral parameter value based on at least one of the segmented image, the vertebral centerline, the spinal canal centerline, and the vertebral body, comprising:

determining a middle level of each vertebral body according to the vertebral centerline;

acquiring gray values of the vertebral bodies on the middle layer surface of each vertebral body;

selecting a measuring area, wherein the measuring area is a circular area taking a first central point of a vertebral body on the middle side surface as a center, and the diameter of the circular area is half of the sagittal diameter of the vertebral canal;

acquiring an average gray value in the measurement area as a gray value corresponding to the middle layer of the cone;

and generating a cone gray density curve according to gray values corresponding to the middle layers of the cones, wherein the bone density of the cones is positively correlated with the corresponding gray density on the cone gray density curve.

Fig. 11 is a gray scale image of the medial aspect of the vertebral body. In fig. 11, for a specified circular region, the area of the region, the maximum and minimum values of the gradation, and the mean square error are automatically measured.

Osteoporosis is a common disease of middle-aged and elderly people, and is liable to cause multiple fracture and multiple serious complications. Bone density measurement is a gold standard for diagnosing fracture loosening and complications. The embodiment can automatically measure and obtain the bone density, and provides accurate basis for diagnosing the osteoporosis degree.

In an exemplary implementation, the method may further include:

at least one of the segmented image, the sagittal image, and the first vertebral parameter value is output and displayed.

In an exemplary implementation, the method may further include:

outputting and displaying the second vertebra parameter value.

The display content of the divided image may include:

(1) The segmented image may be displayed as an MPR (Multi-planar Reformation, multi-planar reconstruction) image or as a 3D (three-dimensional) image. In the segmented image, the spine, ribs, spinal canal tissues can be displayed in the same or different colors, wherein the spine can also display different tissues such as cervical vertebra, thoracic vertebra, lumbar vertebra, sacral vertebra, coccyx and the like by parts in the same or different colors.

(2) In the divided images, different vertebral bodies can be displayed in the same or different colors, and the names of the vertebral bodies are marked, including cervical vertebral bodies C1-C7, thoracic vertebral bodies T1-T12, lumbar vertebral bodies L1-L5, sacral vertebral bodies S1-S5 and coccyx vertebral bodies Co 1-Co 4.

In an exemplary implementation, the method further includes:

and outputting and displaying the parameter value of the preset vertebra parameter.

The displaying of the parameter values of the spinal parameters may include:

(3) Automatically displaying the central lines of the spine and the vertebral canal and the corresponding CPR images;

(4) Automatically displaying the positioned sagittal image of the spine;

(5) Displaying the calculated vertebra length value in mm;

(6) Automatically positioning an axial image of the middle layer of each vertebral body of the cervical vertebra and the lumbar vertebra, displaying sagittal lines and transverse lines of the vertebral canal with the same or different colors, and displaying sagittal and transverse diameter values (unit mm) as diagnosis basis of the vertebral canal stenosis;

(7) In the sagittal image of the vertebra, the connecting line from the midpoint of the front edge to the midpoint of the rear edge of the lumbar intervertebral discs L1-2, L2-3, L3-4, L4-5 and L5-S1 and the intersecting horizontal line are displayed, and the horizontal inclination angles of 5 lumbar intervertebral discs are displayed;

(8) In a sagittal image of the spine, displaying a scoliosis measurement value, including a thoracic rear lobe, a lumbar front lobe, a lumbosacral angle, a sacral midline, a Cobb angle, displaying an upper edge line of an upper vertebra, a lower edge line of a lower vertebra, and displaying serial numbers of the upper vertebra, the lower vertebra and an apex vertebra, as diagnostic basis of scoliosis;

(9) Automatically displaying the optimal measuring plane of each lumbar vertebra L1-L5, displaying the centrum central line, displaying the vertex connecting line of the articular surfaces of the left side and the right side, and displaying the articular angle value of the lumbar vertebra segment;

(10) And automatically displaying the middle layer of the cone, displaying the gray value of the interest area in the center of the cone, and displaying the gray density curve of the cone in a graph.

The organization and classification support user color editing, point, line and angle support user manual editing, and parameter values of various vertebra parameters can be marked on images and can be displayed in an intuitive mode such as a chart and four-corner information, so that a basis is provided for assisting clinical omnibearing and multi-view diagnosis of vertebra diseases.

According to the spine detection method provided by the embodiment of the invention, the three-dimensional spine image is input into the deep learning network model, the segmentation image output by the deep learning network model is obtained, the first central point of the spine region is determined on each cross-sectional image corresponding to the three-dimensional spine image according to the segmentation image, the spine central line is obtained according to the first central point on the cross-sectional image, each spine body of the spine is identified according to the segmentation image and the spine central line, the spine body identification is set for each spine body, the spine sagittal image is positioned according to the segmentation image and the identified spine body, the first spine parameter value is determined according to the spine sagittal image, the spine parameter can be fully automatically measured, the working efficiency is improved, the workload of doctors is reduced, and the labor and time consumption are saved.

In addition, the spine detection method provided by the embodiment of the invention can avoid the error of the diagnosis result caused by the influence of different knowledge accumulation, practice experience and subjective factors of different doctors, and improve the accuracy of the diagnosis result.

Based on the method embodiment, the embodiment of the invention also provides a corresponding device, equipment and storage medium embodiment.

Fig. 12 is a functional block diagram of a spine detection device according to an embodiment of the present invention. As shown in fig. 12, in the present embodiment, the spine detection device may include:

the segmentation module 310 is configured to input a three-dimensional spine image into a deep learning network model, and obtain a segmented image output by the deep learning network model, where at least a spine and a spinal canal are identified in the segmented image;

a first center point determining module 320, configured to determine a first center point of a vertebra region on each cross-sectional image corresponding to the three-dimensional vertebra image according to the segmented image;

a spine centerline acquisition module 330 for acquiring a spine centerline from a first center point on the cross-sectional image;

the vertebral body identification module 340 is configured to identify each vertebral body of the vertebra according to the segmented image and the vertebra center line, and set a vertebral body identifier for each vertebral body;

A positioning module 350, configured to position a sagittal image of the spine according to the segmented image and the identified vertebral body;

a first parameter value determination module 360 is configured to determine a first vertebral parameter value from the sagittal image.

In one exemplary implementation, the vertebral body identification module 340 is specifically configured to:

In one exemplary implementation, determining an occipital macro-aperture center point on a vertebra from the segmented image and the vertebra centerline includes:

determining a skull bottom contour line according to the skull rear side area;

In an exemplary implementation, the segmented image is further labeled with ribs; the vertebral body recognition module 340 is specifically configured to:

In one exemplary implementation, the positioning module 350 may be specifically configured to:

the first parameter value determining module 360 may be specifically configured to:

the intervertebral disc comprises a lumbar disc.

In one exemplary implementation, the first spinal parameter includes a posterior thoracic lobe;

In one exemplary implementation, the first spinal parameter includes a lumbar anterior lobe;

In one exemplary implementation, the first spinal parameter includes a lumbosacral angle;

In one exemplary implementation, the first spinal parameter includes a sacral midline;

Or,

In one exemplary implementation, the first spinal parameter includes a top vertebra; the first parameter value determination module 360 is further configured to:

acquiring the central points of all vertebral bodies between the upper vertebral body and the lower vertebral body as target central points;

In one exemplary implementation, the first spinal parameter includes a Cobb angle; the first parameter value determination module 360 is further configured to:

determining an upper edge line of the upper end vertebra and a lower edge line of the lower end vertebra;

In an exemplary implementation, the method further includes:

the second center point determining module is used for determining a second center point of the vertebral canal area on each cross-sectional image corresponding to the three-dimensional vertebra image according to the segmentation image;

The centrum line acquisition module is used for acquiring a centrum line of the centrum according to the second central point on the cross-section image;

and a second parameter value determining module configured to determine a second spinal parameter value for at least one of the segmented image, the spinal centerline, the spinal canal centerline, and the vertebral body. In one exemplary implementation, the second spinal parameter includes a spinal length;

the second parameter value determining module may be specifically configured to:

In one exemplary implementation, the second spinal parameter includes bone density;

In an exemplary implementation, the apparatus further includes:

and the first output module is used for outputting and displaying at least one of the segmentation image, the vertebra sagittal image and the first vertebra parameter value.

In an exemplary implementation, the apparatus further includes:

And the second output module is used for outputting and displaying the second vertebra parameter value.

The embodiment of the invention also provides electronic equipment. Fig. 13 is a hardware configuration diagram of an electronic device according to an embodiment of the present invention. As shown in fig. 13, the electronic device includes: an internal bus 401, and a memory 402, a processor 403 and an external interface 404 connected by the internal bus, wherein,

the memory 402 is configured to store machine-readable instructions corresponding to the spine detection logic;

the processor 403 is configured to read the machine readable instructions on the memory 402 and execute the instructions to implement any of the aforementioned methods of spine detection.

The embodiment of the invention also provides a computer readable storage medium, on which a computer program is stored, wherein the program, when executed by a processor, implements any one of the aforementioned spine detection methods.

For the device and apparatus embodiments, reference is made to the description of the method embodiments for the relevant points, since they essentially correspond to the method embodiments. The apparatus embodiments described above are merely illustrative, wherein the modules illustrated as separate components may or may not be physically separate, and the components shown as modules may or may not be physical, i.e., may be located in one place, or may be distributed over a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purposes of the present description. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Other embodiments of the present description will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This specification is intended to cover any variations, uses, or adaptations of the specification following, in general, the principles of the specification and including such departures from the present disclosure as come within known or customary practice within the art to which the specification pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the specification being indicated by the following claims.

It is to be understood that the present description is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present description is limited only by the appended claims.

The foregoing description of the preferred embodiments is provided for the purpose of illustration only, and is not intended to limit the scope of the disclosure, since any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the disclosure are intended to be included within the scope of the disclosure.

Claims

1. A method of spinal detection, comprising:

determining a first vertebral parameter value from the vertebral sagittal image;

the identifying each vertebral body of the vertebra from the segmented image and the vertebra centerline comprises:

identifying each vertebral body of the vertebra according to the occipital macropore center point;

the determining the occipital macropore center point on the vertebra according to the segmented image and the vertebra center line comprises the following steps:

determining a skull bottom contour line according to the skull rear side area;

the identifying each vertebral body of the vertebra according to the occipital macropore center point comprises the following steps:

2. The method of claim 1, wherein the segmented image is further labeled with ribs;

identifying respective vertebral bodies of a vertebra from the segmented image and the vertebra centerline, comprising:

3. The method of claim 1, wherein locating a sagittal image from the segmented image and the identified vertebral body comprises:

4. The method of claim 1, wherein the first spinal parameter comprises a horizontal tilt of an intervertebral disc;

the intervertebral disc comprises a lumbar disc.

5. The method of claim 1, wherein the first spinal parameter comprises a posterior thoracic lobe;

6. The method of claim 1, wherein the first spinal parameter comprises a lumbar anterior lobe;

7. The method of claim 1, wherein the first spinal parameter comprises a lumbosacral angle;

8. The method of claim 1, wherein the first spinal parameter comprises a sacral midline;

or,

9. The method of claim 1, wherein the first spinal parameter comprises an end vertebra;

10. The method of claim 9, wherein the first spinal parameter comprises a top vertebra; the method further comprises the steps of:

11. The method of claim 9, wherein the first spinal parameter comprises Cobb angle; the method further comprises the steps of:

12. The method as recited in claim 1, further comprising:

13. The method of claim 12, wherein the second spinal parameter comprises a spinal length;

14. The method of claim 12, wherein the second spinal parameter comprises a spinal sagittal diameter and a spinal transverse diameter;

15. The method of claim 12, wherein the second spinal parameter comprises a lumbar facet joint angle;

16. The method of claim 12, wherein the second spinal parameter comprises bone density;

17. The method according to claim 1, wherein the method further comprises:

18. The method according to claim 12, wherein the method further comprises:

outputting and displaying the second vertebra parameter value.

19. A spinal detection device, comprising:

the first parameter value determining module is used for determining a first vertebra parameter value according to the vertebra sagittal image;

the vertebral body identification module is specifically used for:

determining a skull bottom contour line according to the skull rear side area;

20. An electronic device, comprising: an internal bus, and a memory, a processor and an external interface connected by the internal bus, wherein:

the processor is configured to read machine-readable instructions on the memory and execute the instructions to implement the method of any one of claims 1-18.

21. A computer readable storage medium, characterized in that a computer program is stored thereon, wherein the program, when executed by a processor, implements the method of any of claims 1-18.