Scoliosis screening method

Abstract

The invention discloses a scoliosis screening method, which utilizes a depth camera to perform three-dimensional depth scanning so as to obtain point cloud data of the back of a human body in a forward bending state, and combines a principle of a spine back trunk rotation angle serving as a spine screening standard and point cloud capable of providing rich three-dimensional coordinate data so as to obtain the scoliosis of the human body. A back ruler intersecting line is obtained, the trunk rotation angle of the point cloud of each level is calculated, and finally, a curve of the back position and the trunk rotation angle of the corresponding level is fitted through spline interpolation so as to replace a traditional spine measurement lateral bending ruler, and the defects that back screening is incomplete and insufficient and the like can be overcome. The device has the advantage that the overall lateral bending condition of the back is visualized.

Description

Technical Field

[0001] The present invention relates to the technical field of scoliosis detection, and in particular to a scoliosis screening method. Background Art

[0002] The trunk rotation angle reflects the rotation of the back structure when the patient's spine is performing the Adams flexion test, usually relative to the horizontal plane. A normal spine should be relatively uniform when flexing, while scoliosis will cause significant twisting and rotation in certain parts. These deformations will cause the back to form an angle with the horizontal plane.

[0003] The commonly used screening method now is to use a scoliosis measuring ruler. The subject performs an Adams flexion test, places the scoliosis measuring ruler on the subject's back, moves the scoliosis measuring ruler along the direction of the spine, and measures the inclination angle between the ulnar-dorsal intersection line and the horizontal plane as the trunk rotation angle for scoliosis screening. However, it often only records the maximum rotation angle of the thoracic and lumbar vertebrae, and then takes the maximum trunk rotation angle of the two parts as the measurement index. Although scoliosis is mainly concentrated in the thoracic and lumbar regions, the use of a scoliosis measuring ruler will have a great impact when scoliosis occurs in other parts, thereby reducing the accuracy of screening. In addition, it cannot provide changes in the overall trunk rotation angle of the back, and it is difficult to use it for comparative analysis with previously measured data.

[0004] Existing technologies also use depth cameras to analyze and process the human back for scoliosis screening, but they do not fully utilize the advantages of using point cloud data. They lack functions such as being able to compare front and back point cloud data to monitor changes in back scoliosis or to visualize back scoliosis.

[0005] Therefore, there is an urgent need to provide a new scoliosis screening method to solve the above problems. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a scoliosis screening method, which adopts a non-contact method and provides a screening method with more comprehensive data results based on the measurement indicators of the trunk rotation angle and the three-dimensional coordinates of the point cloud data.

[0007] In order to solve the above technical problems, a technical solution adopted by the present invention is: to provide a scoliosis screening method, comprising the following steps: Step 1: Use the depth camera and calibration plate to obtain external parameters, and determine the x-axis direction of the measurement coordinate system based on the grid direction of the calibration plate; Step 2: Determine the bending direction of the human body when performing the Adams flexion test according to the x-axis direction of the measurement coordinate system, and use a depth camera to collect and obtain the back three-dimensional point cloud P1 of the human body when performing the Adams flexion test above the human body; Step 3: Obtain the back point cloud P2 to be measured by point cloud clipping; Step 4: Convert the point cloud P2 in the camera coordinate system to the point cloud P3 in the measurement coordinate system; Step 5: According to the x-coordinate value of each point in the point cloud P3, the points are divided into different points and the point cloud levels with abnormal data points are filtered out; Step 6: Analyze and process the point cloud coordinates of each layer, determine the ulnar-dorsal intersection line in each layer of point cloud, and calculate the angle between the ulnar-dorsal intersection line and the XOY plane of the measurement coordinate system as the trunk rotation angle of the back point cloud of this layer, and obtain a set of data on the position of each layer of point cloud and the trunk rotation angle; Step 7: Use spline interpolation to fit the data of each level of point cloud and trunk rotation angle to obtain the relationship curve between back position and trunk rotation angle, and find the maximum trunk rotation angle of the back as the standard for scoliosis screening.

[0008] In a preferred embodiment of the present invention, in step S1, the depth camera is located directly above the middle of the human body test area, and the calibration plate is placed at a position that is on the ground plane or parallel to the ground plane.

[0009] In a preferred embodiment of the present invention, in step S1, the acquired extrinsic parameters include a rotation matrix and a translation vector.

[0010] In a preferred embodiment of the present invention, in step S2, the bending direction of the human body during the Adams flexion test is consistent with the positive direction of the x-axis in the measurement coordinate system.

[0011] In a preferred embodiment of the present invention, the specific steps of step S3 include: cropping the area of ​​interest according to the back three-dimensional point cloud P1 obtained in step 2, selecting the back point cloud from the shoulder position to the waist position, and cropping the point cloud back data P2 for measurement.

[0012] In a preferred embodiment of the present invention, in step S5, a suitable division parameter is selected according to the point cloud density, and the point cloud is layered in the x-axis direction according to the division parameter, and the average coordinate value of the x coordinate of the layer is mapped to the layer index, and this is used as the back length position of the layer of point cloud, and the number of points in each layer of point cloud is calculated. For example, there are m levels of point cloud, each layer of point cloud has n points, and the i-th layer of point cloud has j points.

[0013] In a preferred embodiment of the present invention, in step S5, the point cloud level with abnormal data points is filtered using the quartile method.

[0014] In a preferred embodiment of the present invention, in step S6, the process of determining the ulnar-dorsal intersection line in each layer of the point cloud and calculating the trunk rotation angle of the back point cloud includes the following steps: Step S601: Calculate the back width center value coordinates used to divide the left and right sides in each layer of point cloud; Step S602: Searching for the highest points on both sides of the back width center value in each layer of point cloud; Step S603: Calculate the angle between the line connecting the highest points on both sides of the back width center value in each layer of point cloud and the horizontal plane as the trunk rotation angle of the layer of point cloud.

[0015] Furthermore, in step S601, the back width center value of each layer of point cloud is the average value of the y-value coordinates of the points in the layer.

[0016] Furthermore, in step S602, for the highest point on one side of the back width center value in each layer of point cloud, its y value coordinate is smaller than the average y value coordinate of the layer, and the point set on one side of the layer has the maximum z value, i.e., the maximum back height value; for the highest point on the other side of the back width center value in each layer of point cloud, its y value coordinate is greater than the average y value coordinate of the layer, and the point set on the other side of the layer has the maximum z value, i.e., the maximum back height value.

[0017] The beneficial effects of the present invention are: (1) The present invention uses the calibration of the camera to determine the orientation of the human body for the Adams bend test, clips the back point cloud data to be tested, performs coordinate system conversion on the back point cloud, grids the point cloud and layers the point cloud in the x-axis direction, analyzes each layer of the point cloud, finds the highest points on the left and right sides of each layer of the point cloud, calculates the angle between the line connecting the highest points on the left and right sides of each layer and the ground plane, and then interpolates and fits the data to obtain a curve for screening analysis; (2) The present invention provides a screening method with more comprehensive data results by adopting a non-contact method based on the measurement index of the trunk rotation angle and the three-dimensional coordinates of the point cloud data; (3) The present invention makes full use of the data of the back point cloud to visualize the overall scoliosis of the back, which can fully perform measurement and screening, and can better perform comparative analysis. It can overcome the shortcomings of incomplete and insufficient back screening, and has the advantage of visualizing the overall scoliosis of the back. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of a scoliosis screening method of the present invention; Figure 2 This is a schematic diagram of the layered point cloud on the back; Figure 3 is a schematic diagram of the measurement principle of the trunk rotation angle; Figure 4 It is the back point cloud map and the measurement calculation result map; Figure 5 It is the back point cloud image and the fitting result image after removing abnormal data. DETAILED DESCRIPTION

[0019] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0020] See also Figure 1 , the embodiment of the present invention includes: A scoliosis screening method comprises the following steps: Step 1: Use the depth camera and calibration plate to obtain external parameters, and determine the x-axis direction of the measurement coordinate system based on the grid direction of the calibration plate; In the specific implementation, the sampling position needs to be marked first. This position is the human standing area. The sampling position needs to be able to sample the complete back point cloud. The depth camera is supported by a tripod and connected to a gimbal, placed above the standing area, and collects the back point cloud from top to bottom.

[0021] The rotation matrix and translation vector of the measurement coordinate system and the camera coordinate system are obtained by using the depth camera and the calibration plate, i.e., the external parameters. The calibration chessboard located on the ground plane is selected as the measurement coordinate system, and the ground plane is set as the XOY plane of the measurement coordinate system. The x-axis direction of the measurement coordinate system is determined by the grid direction of the calibration plate. For example, the vertical grid line of the chessboard can be taken as the x-axis forward, the horizontal grid line to the right can be taken as the y-axis, and the vertical chessboard plane can be taken as the z-axis.

[0022] Step 2: Determine the bending direction of the human body when performing the Adams flexion test according to the x-axis direction of the measurement coordinate system, and use a depth camera to collect and obtain the back three-dimensional point cloud P1 of the human body when performing the Adams flexion test above the human body; The Adams test is used in the standing area. When it is implemented, the subject is required to be in the acquisition area, stand naturally with feet together, stretch arms and palms together, naturally hang shoulders and spleen bones, close to the chest wall, slowly bend forward to 90 degrees, put hands together and gradually place them in front of knees, and the direction of bending forward is the x-axis direction of the measurement coordinate system. At this time, the point cloud P1 can be obtained through the depth camera.

[0023] Step 3: Obtain the back point cloud P2 to be measured by point cloud clipping; Point cloud P1 includes point cloud information of the neck, back, hips, and ground. The back point cloud P2 is manually obtained by point cloud cropping. In the specific implementation, the back point cloud from the shoulder position to the waist position is selected, and the back point cloud P2 to be measured is cropped. In point cloud P2, only the back point cloud from the shoulder to the waist is included.

[0024] Step 4: Convert the point cloud P2 in the camera coordinate system to the point cloud P3 in the measurement coordinate system, where the x direction is the back length, the y direction is the back width, and the z direction is the back height; The coordinates of a point in the camera coordinate system and the measurement coordinate system are different. The coordinates of the back point cloud in the measurement coordinate system; the coordinates of the back point cloud in the camera coordinate system: The transformation between the measurement coordinate system and the camera coordinate system can be expressed by formula (1): (1) in, It is called the external parameter matrix from the camera coordinate system to the world coordinate system. is the rotation matrix from the camera coordinate system to the world coordinate system, The translation matrix from the camera coordinate system to the world coordinate system.

[0025] Step 5: According to the x-coordinate value of each point in the point cloud P3, the points are divided into different points and the point cloud levels with abnormal data points are filtered out; The point cloud P3 is layered along the x-axis direction. The point cloud is layered according to the size of the x value. When implementing it, a suitable layering parameter is selected, which is generally around 1mm to 1cm. It can change with the density of the point cloud. The higher the accuracy of the sampled point cloud, the smaller the division parameter can be selected, so as to obtain more accurate and richer data results. The point cloud is layered in the x-axis direction according to the division parameter, and the average coordinate value of the x coordinate of this layer is mapped to the hierarchical index, and this is used as the back length position of the point cloud layer. The number of points in each layer of the point cloud is calculated. The point cloud can be divided into m levels, each layer of the point cloud has a total of n points, and the i-th level of the point cloud has j points. The schematic diagram of the layered point cloud is as follows: Figure 2 Before calculation, the point cloud layers with abnormal point counts can be removed. In this example, the quartile method is used to process abnormal values, as shown in the following formula: (2) (3) in, is the interquartile range, is the lower limit threshold of the number of point clouds in each layer, is the first quartile, The third quartile.

[0026] if , then the point cloud layers with points less than the lower threshold value are eliminated; First, calculate the first quartile, median, and third quartile.

[0027] After sorting the points of each layer of point cloud from small to large, the number in the middle position (50% of the number) is taken as the median. The first quartile and the third quartile are the numbers at 25% and 75% respectively after sorting from small to large.

[0028] Step 6: Analyze and process the point cloud coordinates of each layer, determine the ulnar-dorsal intersection line in each layer of point cloud, and calculate the angle between the ulnar-dorsal intersection line and the XOY plane of the measurement coordinate system as the trunk rotation angle (Angle of Trunk Rotation, ATR) of the back point cloud of this layer, and obtain a set of data on the position of each level of point cloud and the trunk rotation angle; the specific steps include: Step S601: Calculate the back width center value used to divide the left and right positions in each layer of point cloud; In specific implementation, the average value of the y coordinates of the point set of the back point cloud of each level is calculated. The average value is used as the standard for dividing each point in the point cloud of this level into the left or right side. The back width center value of the i-th layer is, according to formula (5), the mapping of the point cloud of this level to the back position is calculated as : (4) (5) Step S602: Search for the highest point on the left side and the highest point on the right side in each layer of point cloud; In specific implementation, for the point cloud of layer i , the standard for dividing the left and right sides of the hierarchical point cloud is , for the highest point on the left side of the layer , its coordinate value , Greater than other points on the left side of the layer; for the highest point on the right side of the layer , its coordinate value , Larger than other points on the right side of the layer.

[0029] Step S603: Calculate the angle between the line between the highest points on the left and right sides of each layer and the horizontal plane as the trunk rotation angle of the back point cloud of this layer; like Figure 3 As shown in the figure, the line between the highest points on the left and right sides is used as the ulnar-dorsal intersection line. This line connects the highest point on the left side of this layer of back point cloud with the highest point on the right side of this layer of back point cloud in space. Then, the angle between this straight line and the XOY plane, i.e., the ground plane, is calculated as the trunk rotation angle α.

[0030] In the specific implementation, after the coordinate system conversion, the horizontal plane of the point cloud P3 is parallel to the xoy of the measurement coordinate system. The highest points on the left and right sides of each layer are calculated according to formula (5): , The angle between the line connecting and the xoy plane of the measurement coordinate system.

[0031] (6) Step 7: Use spline interpolation to fit the data of each level of point cloud and trunk rotation angle to obtain the relationship curve between back position and trunk rotation angle, and find the maximum trunk rotation angle of the back as the standard for scoliosis screening.

[0032] The application of spline interpolation includes image processing, data analysis and other fields. In the case of a large number of calibration data points, spline interpolation can provide a fitting effect that is highly consistent with the actual results. The back trunk rotation angle is continuous with the change of the back length position. The data of the average back length position and the trunk rotation angle in each level of the point cloud are fitted by spline interpolation to obtain the relationship curve between the back position and the trunk rotation angle, and the maximum trunk rotation angle of the back is found as the standard for scoliosis screening. According to the curve, the maximum angle and the minimum negative angle can also be found, which can be compared with the trunk rotation angle standard, and the degree of left or right tilt of the back can be known based on the positive and negative angles.

[0033] The present invention makes full use of the data of the back point cloud to visualize the overall scoliosis of the back, which can fully perform measurement and screening and better perform comparative analysis.

[0034] Figure 4 The above picture shows the back point cloud when standing on the side of the back and looking down from the Z axis of the measurement coordinate system, and the picture below shows the measurement calculation result. The blue points in the picture below are the calculated torso rotation angles of each layer of the point cloud. It can be seen that the changes in the blue points are basically continuous with the changes in the back length, which can represent the validity of the results. However, in general, due to the complexity of the back data and its large number, deviated abnormal points will be calculated from it, but these abnormal points have no application significance for the calculated main data, that is, the overall changes in the back rotation angle.

[0035] Figure 5 The upper picture is a point cloud image of the back when standing on the side of the back and looking downward from the Z-axis of the measurement coordinate system. The lower picture is the measurement calculation result diagram, in which the red curve is the change trend of the back rotation angle and back length after elimination and fitting. This figure has richer information and can facilitate deep learning for more in-depth and accurate scoliosis detection and screening in the future.

[0036] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0037] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for screening scoliosis, characterized in that: The following steps are involved: Step 1: Use the depth camera and calibration plate to obtain external parameters, and determine the x-axis direction of the measurement coordinate system based on the grid direction of the calibration plate; Step 2: Determine the bending direction of the human body when performing the Adams flexion test according to the x-axis direction of the measurement coordinate system, and use a depth camera to collect and obtain the back three-dimensional point cloud P1 of the human body when performing the Adams flexion test above the human body; Step 3: Obtain the back point cloud P2 to be measured by point cloud clipping; Step 4: Convert the point cloud P2 in the camera coordinate system to the point cloud P3 in the measurement coordinate system; Step 5: According to the x-coordinate value of each point in the point cloud P3, the points are divided into different points and the point cloud levels with abnormal data points are filtered out; Step 6: Analyze and process the point cloud coordinates of each layer, determine the ulnar-dorsal intersection line in each layer of point cloud, and calculate the angle between the ulnar-dorsal intersection line and the XOY plane of the measurement coordinate system as the trunk rotation angle of the back point cloud of this layer, and obtain a set of data on the position of each layer of point cloud and the trunk rotation angle; Step 7: Use spline interpolation to fit the data of each level of point cloud and trunk rotation angle to obtain the relationship curve between back position and trunk rotation angle, and find the maximum trunk rotation angle of the back as the standard for scoliosis screening.

2. The scoliosis screening method according to claim 1, characterized in that: In step S1, the depth camera is located directly above the center of the human body test area, and the calibration plate is placed at a position at or parallel to the ground plane.

3. The scoliosis screening method according to claim 1, characterized in that: In step S1, the acquired extrinsic parameters include a rotation matrix and a translation vector.

4. The scoliosis screening method according to claim 1, characterized in that: In step S2, when the human body performs the Adams forward flexion test, the bending direction is consistent with the positive direction of the x-axis in the measurement coordinate system.

5. The scoliosis screening method according to claim 1, characterized in that: The specific steps of step S3 include: cutting out the region of interest according to the back three-dimensional point cloud P1 obtained in step 2, selecting the back point cloud from the shoulder position to the waist position, and cutting out the point cloud back data P2 for measurement.

6. The scoliosis screening method according to claim 1, characterized in that: In step S5, appropriate division parameters are selected according to the point cloud density, and the point cloud is layered in the x-axis direction according to the division parameters. The average coordinate value of the x-coordinate of this layer is mapped to the layer index, and this is used as the back length position of the point cloud layer, and the number of points in each layer of the point cloud is calculated. For example, if there are m levels of point clouds, each layer of point clouds has a total of n points, and the i-th layer of point clouds has j points.

7. The method for screening scoliosis according to claim 1, characterized in that: In step S5, the point cloud level with abnormal data points is filtered using the quartile method.

8. The method for screening scoliosis according to claim 1, characterized in that: In step S6, the process of determining the ulnar-dorsal intersection line in each layer of the point cloud and calculating the trunk rotation angle of the back point cloud includes the following steps: Step S601: Calculate the back width center value coordinates used to divide the left and right sides in each layer of point cloud; Step S602: Searching for the highest points on both sides of the back width center value in each layer of point cloud; Step S603: Calculate the angle between the line connecting the highest points on both sides of the back width center value in each layer of point cloud and the horizontal plane as the trunk rotation angle of the layer of point cloud.

9. The method for screening scoliosis according to claim 8, characterized in that: In step S601 , the back width center value of each layer of point cloud is the average value of the y-value coordinates of the points in that layer.

10. The method for screening scoliosis according to claim 9, characterized in that: In step S602, for the highest point on one side of the back width center value in each layer of point cloud, its y value coordinate is smaller than the average y value coordinate of the layer, and the point set on one side of the layer has the maximum z value, i.e., the maximum back height value; for the highest point on the other side of the back width center value in each layer of point cloud, its y value coordinate is greater than the average y value coordinate of the layer, and the point set on the other side of the layer has the maximum z value, i.e., the maximum back height value.

Images