Method for Extracting Characteristic Points of Heart Surface Based on Latitude and Longitude Coordinates

Abstract

The present invention discloses a method for extracting characteristic points of a cardiac surface based on longitude and latitude coordinates, including: segmenting the left ventricle, myocardium, and right ventricle of the heart to obtain the boundary contour lines of the three sub-structures of the heart; reconstructing the mesh surfaces of the three cardiac sub-structures according to the boundary contour lines; calibrating longitude and latitude lines on the mesh surface based on distance and angle, and dividing each sub-structure surface into multiple small regions; obtaining the characteristic points of the surface according to the longitude and latitude line information; the method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates combines longitude and latitude lines with the surface, which is flexible and convenient, and can effectively simplify the model; there is no need to pre-give the shape of the cardiac model, as well as a large amount of annotation data and computing resources; the distribution of characteristic points on the surface at different time frames is uniform and the indexing is ordered, which is convenient for dynamic tracking and deformation analysis of the cardiac surface.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical image processing, and particularly relates to a method for extracting characteristic points of a cardiac surface based on longitude and latitude coordinates. Background Art

[0002] The left ventricle of the heart, the inner and outer walls of the myocardium, and the right ventricle surface can be regarded as zero-genus closed surfaces in three-dimensional space. They are topologically homeomorphic to the sphere in space. Longitude and latitude coordinates are a kind of spherical coordinate system used to describe the position and direction of a certain point on the sphere, which can accurately represent the shape and size of the sphere, and the points on the sphere can also be accurately measured.

[0003] Extracting characteristic points of the cardiac surface is an important method for evaluating cardiac function. The traditional method of manually calibrating cardiac characteristic points requires doctors to manually mark points. Since the calibration accuracy directly affects the diagnostic accuracy, doctors need to have rich experience and professional knowledge; the template matching method needs to create a shape template first and then match it with the image to determine the position of the characteristic points, but it requires an accurate template and strong robustness; although the machine learning method can automatically learn the position of the characteristic points, it requires a large amount of labeled data and computing resources. Most of the above methods are based on image calculation methods, and the calculation of characteristic points on 3D mesh surfaces is rare.

[0004] In view of the above technical problems in the prior art, the present invention provides a method for extracting characteristic points of a cardiac surface based on longitude and latitude coordinates. Summary of the Invention

[0005] The present invention proposes a method for extracting characteristic points of a cardiac surface based on longitude and latitude coordinates. The cardiac substructure surface will be divided by longitude and latitude lines, and the surface will be divided into multiple small regions. Calculate the intersection coordinates of the longitude and latitude lines or the average coordinates of the vertices within each small surface region of the surface, and the characteristic points of the cardiac substructure surface can be obtained. This coordinate can provide a more accurate and reasonable description for the motion tracking of the heart in all directions.

[0006] The present invention adopts the following technical solutions:

[0007] A method for extracting characteristic points of a cardiac surface based on longitude and latitude coordinates, comprising:

[0008] Step 1, segment the left ventricle, myocardium, and right ventricle of the heart to obtain the boundary contour lines of the left ventricle, myocardium, and right ventricle of the heart;

[0009] Step 2, reconstruct the mesh surface of the left ventricle, myocardium, and right ventricle structures according to the boundary contour lines;

[0010] Step 3, calibrate the longitude and latitude lines on the mesh surface based on distance and angle, and divide the surfaces of the left ventricle, myocardium, and right ventricle of the heart into multiple regions;

[0011] Step 4: Obtain the feature points of the curved surface according to the longitude and latitude information.

[0012] Further, in Step 4, obtain the feature points of the curved surface according to the intersection coordinates of the longitude and latitude lines.

[0013] Further, in Step 4, obtain the feature points of the curved surface according to the average coordinates of the vertices in each small area.

[0014] Further, in Step 3, draw longitude and latitude lines longitudinally by distance and transversely by angle on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart to divide the curved surface area.

[0015] Further, in Step 1, construct a model through the nnU-Net framework, use cross-entropy as the loss function, input the training data into the nnU-Net model, and use the Adam optimization algorithm to train the model; use the trained model to segment the unsegmented magnetic resonance image to obtain the segmentation results of the left ventricle, myocardium, and right ventricle of the heart.

[0016] Further, in Step 2, for the segmentation results of all slices of each left ventricle, calculate the central coordinates of the segmentation results. Based on the central coordinates of the middle slice, calculate the translation amounts (t x , t y ) of the central coordinates of other slices and the central coordinates of the middle slice in the horizontal and vertical directions. Align other slices through translation operations to correct the misalignment of the left ventricle, myocardium, and right ventricle at the same time, where (x, y) are the original coordinates and (x′, y′) are the translated coordinates; use the Marching Cubes algorithm for three-dimensional geometric shape reconstruction to generate triangular patches and reconstruct the three-dimensional mesh models of the left ventricle, myocardium, and right ventricle of the heart.

[0017] Further, in Step 3, calibrate the latitude lines at equal distance intervals in the direction perpendicular to the z-axis on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart. The coordinates of the vertex P i on the curved surface are (x i , y i , z i ). z max and z min are respectively the maximum and minimum z coordinate values on the curved surface. The number of latitude lines on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart is n lat , and the distance interval of the latitude lines is:

[0018]

[0019] Calibrate a latitude line every s lat on the curved surface;

[0020] Meridians are calibrated at equal angular intervals on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart, and the number of meridians on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart is n lon , and the angular interval of the meridians is:

[0021]

[0022] Starting from the positive x-axis direction in the three-dimensional coordinate and rotating counterclockwise, a meridian is calibrated every time it rotates through an angle of s lon degrees;

[0023] According to the intervals of the meridians and parallels, the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart can be divided into n lat ×n lon regions.

[0024] Furthermore, in step 3, the index of the latitude and longitude region where the vertex P i is located after division is (lat i , lon i ):

[0025]

[0026] where z max is the maximum z coordinate value on the curved surface, z i is the z coordinate value of the vertex P i , s lat is the parallel interval, denotes rounding down;

[0027]

[0028] where x i and y i are the x and y coordinate values of the vertex P i , x c and y c are the central point coordinates of the curved surface in the xy plane:

[0029]

[0030] x max and x min are the maximum and minimum x coordinate values on the curved surface respectively, y max and y min are the maximum and minimum y coordinate values on the curved surface respectively, s lon is the meridian interval, and by traversing each vertex, the vertex is divided into the corresponding latitude and longitude region;

[0031] The coordinates of the feature points within each latitude and longitude region are the average values of the coordinates of all the vertices within the region:

[0032]

[0033] Among them, N mn is the number of vertices in this area, and x i , y i , z i are the x, y, and z coordinates of all vertices in this area respectively. By traversing each latitude-longitude area of the surface, the obtained coordinate set is the characteristic points of the surface.

[0034] Furthermore, in step 3, 20 latitude-longitude lines are drawn longitudinally and transversely on the mesh surface. According to the average value of the maximum distance and the minimum distance longitudinally, 20 equally spaced latitude lines are divided longitudinally, and 20 equally spaced longitude angles are divided transversely for 360 degrees on average.

[0035] Compared with the prior art, the superior effects of the present invention are as follows:

[0036] The method for extracting characteristic points of the heart surface based on latitude-longitude coordinates according to the present invention combines latitude-longitude lines with the surface, which is flexible and convenient, and can effectively simplify the model; using latitude-longitude coordinates to represent the characteristic points of the heart surface can accurately locate and measure the points on the surface, avoiding the subjectivity, susceptibility to interference, non-expandability, and inability to automate of manual point tracing; there is no need to pre-give the shape of the heart model, as well as a large amount of annotation data and computing resources; the distribution of characteristic points on the surface is uniform and the indexing is ordered under different time frames, which is convenient for dynamic tracking and deformation analysis of the heart surface. Description of the Drawings

[0037] Figure 1 is a schematic flowchart of the method for extracting characteristic points of the heart surface based on latitude-longitude coordinates in an embodiment of the present invention. Detailed Embodiments

[0038] In order to more clearly understand the above-mentioned objects, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0039] Embodiment

[0040] The method for extracting characteristic points of the heart surface based on latitude-longitude coordinates, as Figure 1 shown, the original image sequence is a sequence of cardiac magnetic resonance images obtained from a hospital. The left ventricle, myocardium, and right ventricle segmentation results are obtained by segmenting the sequence images, and then the substructure mesh surface reconstruction is performed according to the segmentation results. The surface area is divided using latitude-longitude coordinates for coordinate point calculation, and finally the characteristic points of the substructure surface are obtained;

[0041] Specifically, the feature point extraction method includes:

[0042] Step 1: Segment the left ventricle, myocardium, and right ventricle of the heart to obtain the boundary contour lines of the three sub-structures of the heart.

[0043] Step 2: Reconstruct the mesh surfaces of the three sub-structures of the heart according to the boundary contour lines.

[0044] Step 3: Based on distance and angle, calibrate the latitude and longitude lines on the mesh surface, and divide each sub-structure surface into multiple regions.

[0045] Step 4: Obtain the feature points of the surface according to the latitude and longitude line information.

[0046] In Step 1 of the above embodiment:

[0047] Build a model through the nnU-Net framework, use cross-entropy as the loss function, input the training data into the nnU-Net model, and use the Adam optimization algorithm to train the model; use the trained model to segment the unsegmented magnetic resonance image to obtain the segmentation results of the three sub-structures of the left ventricle, myocardium, and right ventricle of the heart.

[0048] In Step 2 of the above embodiment:

[0049] For the segmentation results of all slices of each left ventricle, calculate its central coordinates. Based on the central coordinates of the middle slice, calculate the translation amounts (tx, ty) of other slices in the horizontal and vertical directions with respect to this central coordinate, and perform alignment of other slices through translation operations Align other slices, and simultaneously perform misalignment correction on the left ventricle, myocardium, and right ventricle, where (x, y) are the original coordinates and (x′, y′) are the translated coordinates; then use the Marching Cubes algorithm for three-dimensional geometric shape reconstruction to generate triangular patches and reconstruct the three-dimensional mesh models of each sub-structure.

[0050] In Step 3 of the above embodiment:

[0051] Calibrate the latitude lines at equal distance intervals in the direction perpendicular to the z-axis of each sub-structure surface. The coordinates of vertex P i on the surface are (x i , y i , z i ), z max and z min are respectively the maximum and minimum z coordinate values on the surface, and the number of latitude lines of the sub-structure surface is n lat , then the distance interval of the latitude lines is:

[0052]

[0053] At every interval of s on the surfacelat Calibrate a latitude line;

[0054] At equal angular intervals perpendicular to the xy plane, calibrate the longitude lines on each sub-structure surface. The number of longitude lines on the sub-structure surface is n lon , then the angular interval of the longitude lines is:

[0055]

[0056] Starting from the positive x-axis direction in the three-dimensional coordinates and rotating counterclockwise, calibrate a longitude line every time it rotates through an angle of s lon angle;

[0057] According to the longitude and latitude intervals, each sub-structure surface of the heart can be divided into n lat ×n lon regions;

[0058] In step 4 of the above embodiment:

[0059] The obtained coordinates of the intersection points of the longitude and latitude lines can be used as the characteristic points of the surface;

[0060] In addition, the average coordinates of the vertices in each small surface area calibrated by the longitude and latitude lines can also be used as the characteristic points of the surface:

[0061] The specific calculation steps are as follows. These calculation steps are applicable to the left ventricle, the inner and outer walls of the myocardium, and the right ventricle surface of the heart;

[0062] The vertex P of the surface i After division, the index of the longitude and latitude region where it is located is (lat i , lon i );

[0063]

[0064] where z max is the maximum z coordinate value on the surface, and z i is the z coordinate value of the vertex P i , s lat is the latitude interval, represents rounding down;

[0065]

[0066] Among them,

[0067] x i and y i are the x and y coordinate values of the vertex P i , and x c and y c are the center point coordinates of the surface in the xy plane:

[0068]

[0069] x max and x min are respectively the maximum and minimum values of the x - coordinate on the surface, y max and y min are respectively the maximum and minimum values of the y - coordinate on the surface, s lon is the meridian interval; traverse each vertex and divide the vertex into the corresponding latitude - longitude region;

[0070] The coordinates of the feature points within each latitude - longitude region are the average values of the coordinates of all vertices within the region:

[0071]

[0072] where N mn is the number of vertices within the region, x i 、y i 、z i are respectively the x, y, and z coordinates of all vertices within the region. Traverse each latitude - longitude region of the surface, and the obtained coordinate set is the feature points of the surface.

[0073] The present invention is not limited by the above - mentioned embodiments. What is described in the above - mentioned embodiments and the specification only illustrates the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims.

Claims

1. A method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates, characterized in that, it includes: Step 1, segment the left ventricle, myocardium, and right ventricle of the heart to obtain the boundary contour lines of the left ventricle, myocardium, and right ventricle of the heart; Step 2, reconstruct the mesh surface of the structures of the left ventricle, myocardium, and right ventricle of the heart according to the boundary contour lines; Step 3: Based on the distance and angle, calibrate the longitude and latitude lines on the mesh surface, and divide the structural surfaces of the left ventricle, myocardium, and right ventricle of the heart into multiple regions; the surface vertex P i The index of the longitude and latitude region where it is located after division is (lat i , lon i ): where z max is the maximum value of the z coordinate on the surface, z i is the z coordinate value of the vertex P i , s lat is the latitude interval, denotes rounding down; Among them, x i and y i are the x and y coordinate values of vertex P i where x c and y c are the coordinates of the center point of the surface on the xy plane: x max and x min are the maximum and minimum x - coordinates on the surface respectively, y max and y min are the maximum and minimum y - coordinates on the surface respectively, s lon is the meridian interval. Traverse each vertex and divide the vertices into the corresponding latitude - longitude regions; Coordinates of feature points within each latitude and longitude region Is the average value of the coordinates of all vertices within the region: where N mn is the number of vertices in this area, and x i , y i , z i are the x, y, and z coordinates of all vertices in this area respectively. By traversing each latitude-longitude area of the surface, the obtained coordinate set is the characteristic points of the surface; Step 4, obtain the characteristic points of the surface according to the longitude and latitude information.

2. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, in Step 4, the characteristic points of the surface are obtained according to the intersection coordinates of the longitude and latitude lines.

3. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, in Step 4, the characteristic points of the surface are obtained according to the average coordinates of the vertices in each small area.

4. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, in Step 3, longitude and latitude lines are drawn longitudinally by distance and transversely by angle on the surfaces of the left ventricle, myocardium, and right ventricle of the heart to divide the surface area.

5. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, in Step 1, a model is constructed through the nnU-Net framework, cross-entropy is used as the loss function, the training data is input into the nnU-Net model, and the Adam optimization algorithm is used to train the model; the trained model is used to segment the unsegmented magnetic resonance image to obtain the segmentation results of the three sub-structures of the left ventricle, myocardium, and right ventricle of the heart.

6. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, In step 2, for the segmentation results of all slices of each left ventricle, calculate the central coordinates of the segmentation results. Based on the central coordinates of the middle slice, calculate the translation amounts (t x , t y ) in the horizontal and vertical directions between the central coordinates of other slices and the central coordinates of the middle slice. Align other slices through translation operations to simultaneously correct the misalignment of the left ventricle, myocardium, and right ventricle. Here, (x, y) are the original coordinates, and (x', y') are the coordinates after translation; use the Marching Cubes algorithm for three-dimensional geometric shape reconstruction to generate triangular patches and reconstruct the three-dimensional mesh models of the left ventricle, myocardium, and right ventricle of the heart.

7. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, In step 3, latitudes are equally spaced and marked in the direction perpendicular to the z-axis on the surfaces of the left ventricle, myocardium, and right ventricle of the heart, and the vertex P on the surface i has coordinates (x i , y i , z i ), where z max and z min are the maximum and minimum z-coordinate values on the surface respectively, and the number of latitudes on the surfaces of the left ventricle, myocardium, and right ventricle of the heart is n lat , and the distance interval between latitudes is: Mark a latitude line every s on the surface lat calibrate a latitude line; Meridians are calibrated at equal angular intervals on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart, with the number of meridians on the curved surfaces of the left ventricle, myocardium, and right ventricle of the heart being n lon , and the angular interval of the meridians is: Starting from the positive x-axis direction in the three-dimensional coordinates and rotating counterclockwise, a meridian is calibrated every s lon angle. The curved surfaces of the left ventricle, myocardium, and right ventricle of the heart can be divided into n lat ×n lon regions according to the intervals of longitude and latitude lines.

8. The method for extracting characteristic points of the cardiac surface based on longitude and latitude coordinates according to claim 1, characterized in that, in Step 3, 20 longitude and latitude lines are drawn longitudinally and transversely on the mesh surface. According to the average value of the maximum and minimum longitudinal distances, 20 equally spaced latitude lines are divided longitudinally, and 20 equally spaced longitude line angles are divided transversely at 360 degrees.

Images