A semi-manual ultrasound and CT fusion registration method

Abstract

The application relates to the field of detection registration technology, and discloses a semi-manual fusion registration method of ultrasound and CT, which comprises the following steps: S1: CT image point cloudization, enhanced CT scanning and ultrasonic scanning are carried out on a patient, after scanning, a staff member manually selects easily recognizable blood vessels or bronchial tubes based on the CT scanning result of the patient, carries out tissue segmentation on the CT image, marks soft tissues such as blood vessels and liver which need to be recognized, and reconstructs a point cloud image; S2: ultrasonic image point cloudization; S3: identification of a central axis. The point cloud image of the whole blood vessel is replaced by the central axis, so that the efficiency and precision of registration can be improved; the intersection with high recognition degree is manually selected by a doctor, so that the accuracy and speed of registration are improved; based on the semi-automatic registration mode, the doctor selects the optimal one, and the automatic registration mode improves the registration efficiency and guarantees the accuracy of registration in the easily deformed soft tissue.

Description

Technical Field

[0001] The present invention relates to the technical field of detection and registration, and in particular to a semi-manual ultrasound and CT fusion registration method. Background Art

[0002] Medical image fusion technology is a hot topic of research both domestically and internationally. Medical imaging modalities include X-rays, ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), single-photon emission tomography (SPECT), positron emission tomography (PET), infrared, digital subtraction angiography (DSA), and fluoroscopy. Medical images of various modalities reflect information about the human body from different perspectives, and comprehensive diagnostic information cannot be obtained from a single image alone. If doctors rely solely on spatial assumptions and speculation to comprehensively assess the information they need from multiple images, their accuracy will be affected by subjective judgment and may even overlook some information.

[0003] Tissue registration is of great significance for medical surgical positioning and is a necessary step for automated guidance. For example, in the liver, lungs, and kidneys, soft tissue deformation at different stages is inevitable. How to complete local registration under deformation is very meaningful. Due to spatial relationships or inconsistent tissue sizes obtained at different times, automatic registration is difficult and has low accuracy. Therefore, a semi-manual ultrasound and CT fusion registration method is proposed. Summary of the Invention

[0004] (1) Technical problems solved

[0005] In response to the shortcomings of the existing technology, the present invention provides a semi-manual ultrasound and CT fusion registration method, which solves the problems of difficulty in automatic registration and low accuracy due to spatial relationships or inconsistent tissue sizes acquired at different times.

[0006] (2) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A semi-manual ultrasound and CT fusion registration method comprises the following steps:

[0009] S1: Convert CT images into point clouds and perform enhanced CT scans and ultrasound scans on the patient. After the scans, staff manually select easily identifiable blood vessels or bronchi based on the patient's CT scan results, perform tissue segmentation on the CT images, mark the blood vessels, liver and other soft tissues that need to be identified, and reconstruct the point cloud image.

[0010] S2: Ultrasound image point clouding. The ultrasound probe installed on the robotic arm continuously scans to obtain real-time ultrasound video information, and then obtains the posture information corresponding to the ultrasound video information and the ultrasound probe posture information (x, y, z, γ, β, α). Then, the ultrasound image is matched with the corresponding posture information. The organs in the ultrasound image are identified and the recognition results are displayed and sampled. Based on the sampled organ contour sampling points, the three-dimensional coordinates are changed to obtain the actual position of each contour point in the frame image in the robotic arm coordinate system, and the sparse points are interpolated based on the contour features. Each converted point is depicted in 3D space for visualization. The coordinate system of this space is set according to the robot coordinate system.

[0011] S3: Identify the central axis, which is based on the spatial circle model to identify the central axis of the blood vessel or bronchial point cloud model, including horizontal central axis extraction, cross-section extraction based on the horizontal central axis, and setting the cross-section equation based on the spatial circle model fitting the central axis;

[0012] S4: Select intersection points. The doctor manually selects 2-5 intersection points on the central axis with high identifiability.

[0013] S5: Align the point cloud positions. Based on multiple intersection points, align the point cloud positions of CT and ultrasound. First, correct the deviation between the ultrasound plane and the CT section, and convert multiple points of the CT slice and ultrasound image to the same CT physical coordinate system. The marked pixel points [A1, A2, A3] are calculated to the 3D physical coordinate system of CT using the current slice parameters. On the ultrasound image, the marked pixel points [B1, B2, B3] are calculated to the 3D physical coordinate system of CT using the current coarse registration matrix and the current posture matrix of the robotic arm.

[0014] As a further solution of the present invention, in S2, for the ultrasound probe posture information, a predetermined number of continuous posture information are sampled every predetermined time difference as a posture information group, each posture information group corresponds to a video frame with the same video frame number as its own group number, and the first posture information in each group of ultrasound probe posture information is selected to match the corresponding video frame.

[0015] Furthermore, in S2, image recognition is performed on each frame of the collected image, the corresponding organ contour for which a model is to be established is extracted, and the organ contour is sampled to obtain contour sampling points.

[0016] Based on the above scheme, the robot coordinate system, probe coordinate system and image coordinate system are defined as {B}, {P} and {I} respectively in S2, and the transformation matrix from {P} to {B} is The transformation matrix to {P} is,

[0017] Furthermore, in S3, the projection point set of the blood vessel or bronchus on the X0Y horizontal plane is obtained based on the projection calculation, a Delaunay triangulation is constructed for the projection point set to extract the boundary point set, and the left and right boundary lines are extracted based on the distance between the turning points in the boundary point set and the relationship between the tube diameter and the forward direction; finally, the KD tree algorithm is used to extract the initial horizontal central axis.

[0018] Based on the above scheme, in S3, a straight line equation is constructed based on each two adjacent points on the horizontal central axis of the acquired ground, and the corresponding midpoint and the normal vector of the cross section at the midpoint are obtained, thereby obtaining the cross section equation at the midpoint, and then the three-dimensional data of the pipeline between the corresponding two points is intercepted according to the equation to form a cross section;

[0019] Determine the seven points of the horizontal axis according to the direction of blood circulation, and then u (x mu ,y mu , z mu ) and M v (x mv ,y mv , z mv ) Construct the equation of the line, that is, the slope of the corresponding line Then the midpoint of the corresponding line segment can be obtained as M uv (x uv ,y uv , z uv ) and its normal vector is e(1, k u , 0);

[0020] That is, the cross-section equation at the midpoint can be expressed as:

[0021] xx uv +k u (yy uv )=0(1), X, Y are any set of points on the X axis or Y axis.

[0022] In a further embodiment of the present invention, the cross-sectional equation of the central axis in S5 is:

[0023] Ax+By+Cz+D=0

[0024] Combined with the cross-sectional equation obtained from formula (1), A=1,B=k r , C=0,D=-x r -k r ·y r , then the normal vector of the corresponding cross section is e r (1, k r , 0);

[0025] Next, let the corresponding spherical equation be:

[0026] (x-x1) 2 +(y-y1) 2 +(z-z1) 2 =R1 2

[0027] Among them, the coordinates of the sphere center O1 are (x0, y0, z0), and the radius of the sphere is R1;

[0028] According to the error equation and its constraints, the approximate sphere center coordinates O1(x1, y1, z1) are obtained through continuous iteration;

[0029] Based on the approximate spherical equation, the vertical distance D from the center of the sphere to the cross section is obtained r ;Finally, based on the corresponding cross-section normal vector e r (1, k r , 0), Kr is the coordinate of the Y axis, project the coordinates of the sphere center to the cross-section equation to obtain the center of the precise space circle O(x0, y0, z0), and calculate the corresponding space circle radius R;

[0030]

[0031]

[0032] Where: D r The vertical distance from the center of the sphere to the cross section, R is the radius of the exact circle in space, and R1 is the radius of the approximate circle in space.

[0033] (3) Beneficial effects

[0034] Compared with the existing technology, the present invention provides a semi-manual ultrasound and CT fusion registration method, which has the following beneficial effects:

[0035] 1. In the present invention, the point cloud of ultrasound and CT images can express the spatial outline and specific position of the object, and different point clouds can be directly fused as long as they are in the same coordinate system.

[0036] 2. In the present invention, the three-dimensional point cloud image is represented by a simple and clear space curve, which can filter out useless point cloud information, making the image representation simpler, reducing the difficulty of matching, and improving the accuracy and efficiency of registration.

[0037] 3. In the present invention, by replacing the point cloud image of the entire blood vessel with the central axis, the efficiency and accuracy of the registration can be improved. The doctor manually selects highly recognizable intersections to improve the accuracy and speed of the registration. Based on this semi-automatic registration method, the doctor selects the best and automatically registers, which not only improves the efficiency of the registration, but also ensures the accuracy of the registration in deformable soft tissues. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic diagram of the process structure of a semi-manual ultrasound and CT fusion registration method proposed in the present invention;

[0039] Figure 2 This is a schematic diagram of the spatial circle model of a semi-manual ultrasound and CT fusion registration method proposed in the present invention. DETAILED DESCRIPTION

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0041] Reference Figure 1-2 A semi-manual ultrasound and CT fusion registration method comprises the following steps:

[0042] Step 1: Convert CT images into point clouds;

[0043] Step 1-1: Perform enhanced CT scan and ultrasound scan on the patient;

[0044] Step 1-2: The doctor manually selects easily identifiable blood vessels or bronchi based on the patient's CT scan results;

[0045] Steps 1-3: Segment the CT image, mark the blood vessels and soft tissues of the liver that need to be identified, and reconstruct the point cloud image. The principle is shown in the following formula:

[0046]

[0047] Among them, x, y, z represent the three-dimensional coordinate points in the world coordinate system, that is, the coordinate points after point clouding, x', y' are any coordinate points in the image coordinate system, and D is the depth value; f x , f y is the focal length, the two values ​​are the same;

[0048] Step 2: Ultrasound image point cloud conversion:

[0049] Step 2-1: The ultrasound probe mounted on the robotic arm continuously scans to obtain real-time ultrasound video information;

[0050] Step 2-2: Obtain the posture information corresponding to the ultrasound video information, and obtain the ultrasound probe posture information (x, y, z, γ, β, α);

[0051] Step 2-3: Matching the ultrasound image with the corresponding posture information;

[0052] Specifically, for the ultrasound probe posture information, a predetermined number of consecutive posture information are sampled at predetermined time intervals as a posture information group. Each posture information group corresponds to a video frame with the same video frame number as its own group number. The first posture information in each group of ultrasound probe posture information is selected to match the corresponding video frame.

[0053] Step 2-4: Identification of organs in the ultrasound image and display of identification results;

[0054] Specifically, image recognition is performed on each frame of the collected image to extract the corresponding organ contour for which a model is to be built, and the organ contour is sampled to obtain contour sampling points;

[0055] The sampling points for obtaining organ contours include:

[0056] (a) Arrange all points on the organ contour in order along the contour to generate a point sequence

[0057] S P

[0058] S P ={P1(x1, y1), P2(x1, y1)…Pm(xm, ym)…Pn(xn, yn)} (1)

[0059] Where P represents the contour point, (x, y) is the coordinate of the contour point, and n is the number of contour points;

[0060] (b) Based on a certain point and its two preceding and following points, solve the changing trend of the organ contour at that point, and express it as the vector angle between each point and the two adjacent points to generate an angle sequence. The vector angle of point Pm is θ m :

[0061]

[0062] (c) Find the average value θ of each angle a , and process the vector angle series as follows to generate the acquisition sequence S s

[0063]

[0064] (d) Using θ m Start sampling, assuming that the probability of the point to be collected is

[0065] c, c∈(0,1], then point P m The following conditions are met and the first point sequence S is generated. P1 , the first sampling point sequence S P1 The sampling points in satisfy formula (4)

[0066] I is a positive integer, and

[0067] (e) is to extract the feature points on the contour, and the angle change exceeds the average value θ a All points with a certain multiple of e are sampled to generate the second sampling point sequence S P2 , the second sampling point sequence S P2 The sampling points in satisfy formula (5)

[0068] θ m >e*θ a , e>1 (5)

[0069] (f) The two sampling sequences may have some overlapping points. Taking their union generates the sampling contour point sequence S as follows:

[0070] S=S P1 ∪S P2 (6);

[0071] Step 2-5: Based on the sampled organ contour sampling points, perform three-dimensional coordinate transformation to obtain the actual position of each contour point in the frame image in the robotic arm coordinate system, and perform contour feature-based interpolation processing on sparse points;

[0072] Specifically, the robot coordinate system, probe coordinate system and image coordinate system are defined as {B}, {P} and {I} respectively, and the transformation matrix from {P} to {B} is The transformation matrix from {I} to {P} is,

[0073] (a) Conversion of ultrasound image pixel coordinates to probe coordinates;

[0074] Assume that the pixel coordinates of a certain point A in the ultrasound image after sampling are A(a, b), then its physical coordinates after conversion are A(a', b', c'), where m and n are scale coefficients, a is the x-axis coordinate, b is the y-axis coordinate, a' is the x-axis coordinate, b' is the y-axis coordinate, and c' is the z-axis coordinate. The conversion formula is as follows:

[0075] a′=m*a (7)

[0076] b′=n*b (8)

[0077] c′=0 (9)

[0078] According to the actual positional relationship between the ultrasound image and the probe, the image coordinates of point A are converted to the coordinates (a″, b″, c″) in the probe coordinate system {P} as follows:

[0079] a″=0 (10)

[0080] b″=a′-0.5*w (11)

[0081] c″=-b′ (12)

[0082] Where w is the width of the ultrasound probe bottom;

[0083] (b) Transformation from probe coordinates to robot coordinates. According to the robot coordinate changes, the coordinate transformation from the probe coordinate system to the robot coordinate system is:

[0084]

[0085] If the posture corresponding to a certain ultrasound image is obtained as (x, y, z, γ, β, α), (x, y, z) refers to the position of the probe in the robot coordinate system, and (γ, β, α) refers to the rotation of the probe robot coordinate system. According to the Euler coordinate system transformation, Formula 7 is expanded into Formula 8, where c is the abbreviation of cosine and s is the abbreviation of sin.

[0086]

[0087] (c) Interpolation of sparse models;

[0088] When the minimum distance d between a sampling point A(x1, y1, z1) in the current frame and B(x2, y2, z2) among all sampling points in the current frame is greater than the predetermined accuracy threshold q, based on the previous frame where point A is located, the contour point with the minimum distance from all contour points on it to point A is C(x3, y3, z3). The arc f(x) is fitted based on points A, B, and C using the following interpolation method:

[0089] Based on f(x), interpolation points are inserted from point A to point B in sequence, so that the arc length between the first interpolation point and point A and the arc length between adjacent interpolation points are both l, until the arc length between the current interpolation point and point B is less than or equal to l.

[0090] The calculation formula for l is:

[0091] l=q / (n a n b s a s b ) (15)

[0092] In the above formula, n a Indicates the total number of sampling points in the frame where point A is located, s a Indicates the value of point A in the acquisition sequence, n b Indicates the total number of sampling points of point B in its frame,

[0093] s b Indicates the value of point B in the acquisition sequence;

[0094] Step 2-6: Draw each transformed point into a 3D space for visualization, where the coordinate system is set according to the robot coordinate system;

[0095] Converting ultrasound and CT images into point clouds can express the spatial outline and specific location of objects, and different point clouds can be directly fused as long as they are in the same coordinate system;

[0096] Step 3: Identify the central axis of the blood vessel or bronchial point cloud model based on the spatial circle model;

[0097] Step 3-1: Extract the horizontal central axis;

[0098] Specifically, the projection point set of the blood vessels or bronchus on the XOY horizontal plane is obtained through projection forward calculation. A Delaunay triangulation is constructed from the projection point set to extract the boundary point set. The left and right boundary lines are extracted based on the relationship between the distance between the turning points in the boundary point set and the tube diameter and the forward direction. Finally, a KD tree (a data structure for organizing points in K-dimensional Euclidean space, or K-Dimensional) algorithm is used to extract the initial horizontal central axis.

[0099] Step 3-2: Extract the cross section based on the horizontal central axis;

[0100] Specifically, a straight line equation is constructed based on every two adjacent points on the horizontal central axis of the acquired site, and the normal vector of the corresponding midpoint and the cross section at the midpoint is obtained, thereby obtaining the cross section equation at the midpoint. Then, the three-dimensional pipeline data between the corresponding two points is intercepted according to the equation to form a cross section.

[0101] Determine the seven points of the horizontal axis according to the direction of blood circulation, and then u (x mu ,y mu , z mu ) and M v (x mv ,y mv , z mv ) Construct the equation of the line, that is, the slope of the corresponding line Then the midpoint of the corresponding line segment can be obtained as M uv (x uv ,y uv , z uv ) and its normal vector is e(1, k u , 0);

[0102] That is, the cross-section equation at the midpoint can be expressed as:

[0103] xx uv +k u (yy uv)=0(1), X, Y are any set of points on the X-axis or Y-axis;

[0104] Step 3-3: Based on the spatial circle model, fit the central axis and set the cross-section equation as:

[0105] Ax+By+Cz+D=0

[0106] Combined with the cross-sectional equation obtained from formula (1), A=1,B=k r , C=0,D=-x r -k r ·y r , then the normal vector of the corresponding cross section is e r (1, k r , 0);

[0107] Next, let the corresponding spherical equation be:

[0108] (x-x1) 2 +(y-y1) 2 +(z-z1) 2 =R1 2

[0109] Among them, the coordinates of the sphere center O1 are (x0, y0, z0), and the radius of the sphere is R1;

[0110] According to the error equation and its constraints, the approximate sphere center coordinates O1(x1, y1, z1) are obtained through continuous iteration;

[0111] Based on the approximate spherical equation, the vertical distance D from the center of the sphere to the cross section is obtained r ;Finally, based on the corresponding cross-section normal vector e r (1, k r , 0), Kr is the coordinate of the Y axis, project the coordinates of the sphere center to the cross-section equation to obtain the center of the precise space circle O(x0, y0, z0), and calculate the corresponding space circle radius R;

[0112]

[0113]

[0114] Where: D r The vertical distance from the center of the sphere to the cross section, R is the radius of the exact circle in space, and R1 is the radius of the approximate circle in space;

[0115] Representing the 3D point cloud with simple and clear space curves can filter out useless point cloud information, making the image representation simpler, reducing the difficulty of matching, and improving the accuracy and efficiency of registration;

[0116] Step 4: The doctor manually selects the intersection points on three highly identifiable central axes;

[0117] Step 5: Align the point cloud positions of CT and ultrasound based on the three intersection points;

[0118] Step 5-1: Correct the deviation between the ultrasound plane and the CT section;

[0119] Specifically, the current ultrasound plane and CT section are approximately coplanar, but there are deviations in the plane positions and angles. This means that based on the current coarse registration matrix as the coordinate system reference, only the plane direction (x, y, z = 0) and the plane normal vector rotation angle (rx = 0, ry = 0, rz) need to be revised. The rotation matrix is ​​expressed as:

[0120]

[0121] Translation vector representation

[0122]

[0123] Step 5-2: Convert the CT slice and the three points of the ultrasound image to the same CT physical coordinate system. The three marked pixel points [A1, A2, A3] are calculated to the CT 3D physical coordinate system using the current slice parameters:

[0124]

[0125] On the ultrasound image, the three marked pixel points [B1, B2, B3] are calculated to the CT 3D physical coordinate system using the current coarse registration matrix and the current robot arm posture matrix:

[0126]

[0127] By replacing the point cloud image of the entire blood vessel with the central axis, the efficiency and accuracy of the registration can be improved. After the doctor manually selects highly recognizable intersections, the accuracy and speed of the registration are improved. Based on this semi-automated registration method, the doctor selects the best and automatically registers, which not only improves the efficiency of the registration, but also ensures the accuracy of the registration in easily deformed soft tissues.

[0128] In the description herein, it should be noted that relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include," "comprise," or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A semi-manual ultrasound and CT fusion registration method, characterized in that: The following steps are involved: S1: Convert CT images into point clouds and perform enhanced CT scans and ultrasound scans on the patient. After the scans, staff manually select easily identifiable blood vessels or bronchi based on the patient's CT scan results, perform tissue segmentation on the CT images, mark the blood vessels, liver and other soft tissues that need to be identified, and reconstruct the point cloud image. S2: Ultrasound image point clouding. The ultrasound probe installed on the robotic arm continuously scans to obtain real-time ultrasound video information, and then obtains the posture information corresponding to the ultrasound video information and the ultrasound probe posture information (x, y, z, γ, β, α). Then, the ultrasound image is matched with the corresponding posture information. The organs in the ultrasound image are identified and the recognition results are displayed and sampled. Based on the sampled organ contour sampling points, the three-dimensional coordinates are changed to obtain the actual position of each contour point in the frame image in the robotic arm coordinate system. The sparse points are interpolated based on the contour features, and each converted point is depicted in the 3D space for visualization. The coordinate system of the space is set according to the robot coordinate system. S3: Identify the central axis, which is based on the spatial circle model to identify the central axis of the blood vessel or bronchial point cloud model, including horizontal central axis extraction, cross-section extraction based on the horizontal central axis, and setting the cross-section equation based on the spatial circle model fitting the central axis; S4: Select intersection points. The doctor manually selects 2-5 intersection points on the central axis with high identifiability. S5: Align the point cloud positions. Based on multiple intersection points, align the point cloud positions of CT and ultrasound. First, correct the deviation between the ultrasound plane and the CT section, and convert multiple points of the CT slice and ultrasound image to the same CT physical coordinate system. The marked pixel points [A1, A2, A3] are calculated to the 3D physical coordinate system of CT using the current slice parameters. On the ultrasound image, the marked pixel points [B1, B2, B3] are calculated to the 3D physical coordinate system of CT using the current coarse registration matrix and the current robot arm posture matrix.

2. A semi-manual ultrasound and CT fusion registration method according to claim 1, characterized in that: In S2, for the ultrasound probe posture information, a predetermined number of continuous posture information are sampled at predetermined time difference intervals as a posture information group, each posture information group corresponds to a video frame with the same video frame number as its own group number, and the first posture information in each group of ultrasound probe posture information is selected to match the corresponding video frame.

3. The semi-manual ultrasound and CT fusion registration method according to claim 2, characterized in that: In the step S2, image recognition is performed on each frame of the collected image to extract the corresponding contour of the organ for which a model is to be established, and the contour of the organ is sampled to obtain contour sampling points.

4. The semi-manual ultrasound and CT fusion registration method according to claim 1, characterized in that: Specifically in S2, the robot coordinate system, the probe coordinate system and the image coordinate system are defined as {B}, {P} and {I} respectively, and the transformation matrix from {P} to {B} is The transformation matrix from {I} to {P} is, 5. The semi-manual ultrasound and CT fusion registration method according to claim 1, characterized in that: In S3, the projection point set of the blood vessel or bronchus on the XOY horizontal plane is obtained based on the projection calculation, a Delaunay triangulation is constructed for the projection point set to extract the boundary point set, and the left and right boundary lines are extracted based on the distance between the turning points in the boundary point set and the relationship between the tube diameter and the forward direction; finally, the KD tree algorithm is used to extract the initial horizontal central axis.

6. The semi-manual ultrasound and CT fusion registration method according to claim 5, characterized in that: In said S3, a straight line equation is constructed based on each two adjacent points on the horizontal central axis of the obtained ground, and the corresponding midpoint and the normal vector of the cross section at the midpoint are obtained, thereby obtaining the cross section equation at the midpoint, and then the three-dimensional data of the pipeline between the corresponding two points is intercepted according to the equation to form a cross section; Determine the seven points of the horizontal axis according to the direction of blood circulation, and then u (x mu ,y mu ,z mu ) and M v (x mv ,y mv ,z mv ) Construct the equation of the line, that is, the slope of the corresponding line Then the midpoint of the corresponding line segment can be obtained as M uv (x uv ,y uv ,z uv ) and its normal vector is e(1,k u ,0); That is, the cross-section equation at the midpoint can be expressed as: xx uv +k u (yy uv )=0 (1), X, Y are any set of points on the X axis or Y axis.

7. The semi-manual ultrasound and CT fusion registration method according to claim 6, characterized in that: The cross-section equation of the central axis of S5 is: Ax+By+Cz+D=0 Combined with the cross-sectional equation obtained from formula (1), A=1, B=k in this equation r , C=0, D=-x r -k r ·y r , then the normal vector of the corresponding cross section is e r (1,k r ,0); Next, let the corresponding spherical equation be: (x-x1) 2 +(y-y1) 2 +(z-z1) 2 =R1 2 Among them, the coordinates of the sphere center O1 are (x0, y0, z0), and the radius of the sphere is R1; According to the error equation and its constraints, the approximate coordinates of the sphere center O1(x1, y1, z1) are obtained through continuous iteration; Based on the approximate spherical equation, the vertical distance D from the center of the sphere to the cross section is obtained r ;Finally, based on the corresponding cross-section normal vector e r (1,k r ,0), Kr is the coordinate of the Y axis, project the coordinate of the sphere center to the cross-section equation to obtain the center of the precise space circle O(x0,y0,z0), and calculate the corresponding space circle radius R; Where: D r The vertical distance from the center of the sphere to the cross section, R is the radius of the exact circle in space, and R1 is the radius of the approximate circle in space.

Images