Method and system for quickly acquiring initial value of spine registration

Abstract

The present invention discloses a method and system for quickly acquiring initial values ​​of spinal registration. Through a small amount of interactive information input, six-dimensional initial parameters are quickly calculated according to spinal posture and surgical characteristics, which match well with the final registration result, thereby significantly improving the calculation efficiency and result accuracy of the registration iterative algorithm while reducing the dependence on operating experience. There are few interactive operations, which reduces the dependence on operating experience. The calculation formula is highly interpretable and easy to understand. The calculation amount is low, so the calculation speed is fast, and the result is accurate, which improves the efficiency of the registration algorithm and shortens the operation time.

Description

Technical Field

[0001] The present invention relates to the field of electromechanical equipment, and in particular to a method and system for quickly acquiring an initial value of spinal registration. Background Art

[0002] The spine is an extremely important bone in the human body. When the mechanical balance of the spine is destroyed, it may further affect the peripheral nerves, blood vessels and other tissues, leading to diseases of other systems in the human body. Therefore, it is particularly important to properly treat spinal lesions. Spinal screw placement surgery is a common type of orthopedic surgery. By placing screws in the pedicles to fuse multiple vertebrae, the spinal vertebrae sequence reconstruction and local stabilization are completed. At the same time, due to the presence of a large number of nerves and blood vessels in the spinal canal, improper screw placement can cause serious consequences, and spinal surgery technology assisted by robot navigation has begun to develop. Multimodal medical image registration is one of the key links in robot-assisted navigation technology. In the field of spinal surgery, a common type of multimodal medical image registration is the registration of three-dimensional CT images and two-dimensional X-ray images. After completing the surgical screw placement planning based on the preoperative CT images, X-ray images containing navigation calibration objects are taken at multiple angles during the operation. After the two medical images are registered through multimodal image registration technology, the necessary navigation parameters are provided to the robot. Due to the large differences in image acquisition methods and the different patient postures before and during surgery, in the multimodal registration process of CT images and X-ray images, an accurate and efficient method for obtaining the initial registration parameters can significantly improve the accuracy and computational efficiency of the registration algorithm, thereby shortening the operation time.

[0003] At present, the main methods for obtaining the initial parameters for the registration of CT images and X-ray images include empirical methods, random generation methods, and multiple registration methods. In the empirical method, the robot operator manually inputs the initial parameters based on experience. The disadvantage is that it is extremely dependent on the operator's experience and is time-consuming. The random generation method randomly generates a set of initial parameters through a computer. When the initial parameters are far from the final registration result, the number of iterations of the registration algorithm will be greatly increased, which will increase the time consumption and the risk of registration failure. The multiple registration method first obtains rough initial registration parameters through some rough registration algorithms, and then inputs them into the fine registration algorithm for subsequent calculations. It also has the disadvantages of complex process, increased time consumption, and the risk of errors in the initial parameters.

[0004] In summary, a method and system for quickly acquiring the initial value of spinal registration is needed to solve the deficiencies in the prior art. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for quickly acquiring initial values ​​of spine registration, aiming to solve the above problems.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for quickly acquiring the initial value of spine registration, comprising the following steps:

[0007] Step S1: Obtain a 3D CT image of the spine, read the spine posture recorded in the label, and obtain the spine region and each vertebral segment number and vertebral segment center through a segmentation algorithm;

[0008] Step S2: obtaining at least two X-ray images of different angles containing the marker ball array, and recording the position of the point light source during the shooting;

[0009] Step S3: Calculate the coordinate system conversion matrix according to the X-ray image;

[0010] Step S4: manually mark or select the X-ray image to calculate the center coordinates of two different vertebrae;

[0011] Step S5: Taking the space of an X-ray image as a reference, setting the point light source position of the X-ray image as the spatial origin, and calculating the initial parameters of the three-axis rotation according to the information comparison between the CT image and the X-ray image;

[0012] Step S6: calculating the initial parameters of the three-axis translation, including calculating the y-axis translation Ty according to the X-ray lateral image, and calculating the x-axis translation Tx and the z-axis translation Tz according to the X-ray anteroposterior image, to obtain the initial parameters of the three-axis translation;

[0013] Step S7: Input the initial parameters of the three axes of rotation and the three axes of translation into the registration, obtain the registration result, and verify it.

[0014] Optionally, in step S2, the best combination of the two X-ray images at different angles is a standard frontal view and a standard lateral view.

[0015] Optionally, the coordinate system conversion matrix calculated in step S3 is:

[0016] The coordinate system transformation matrix T of image B to the orthogonal image A is calculated based on the coordinate matrix corresponding to the marker ball array in the two X-ray images. ba .

[0017] Optionally, the initial parameters of the three-axis rotation are calculated in step S5 in the following manner:

[0018] Step S51: gradually adjusting the rotation parameters according to the CT image posture label and the actual spine posture, so that the CT image posture is consistent with the standard surgical posture;

[0019] Step S52: According to the difference between the spine direction in the X-ray image and the spine direction in the CT image, the rotation parameters Ry and Rz are adjusted to make the spine directions in the CT image consistent with those in the X-ray image, and the initial rotation parameters R (Rx, Ry, Rz) of the three axes are obtained.

[0020] Optionally, the rotation parameters are adjusted in step S51 in the following manner:

[0021] Step A1: Determine the CT posture label: if the label is head-first prone, increase Rz by 180°; if the label is feet-first supine, increase Ry by 180°; if the label is feet-first prone, increase Rx by 180°;

[0022] Step A2: Determine the actual spine posture: if the actual position is head first, increase Rz by 180°; if the actual position is feet first, increase Ry by 180°; if the actual position is feet first, increase Rx by 180°;

[0023] Step A3, the common posture in spinal surgery is head-first prone, increasing Rz by 180°.

[0024] Optionally, the three-axis initial parameters R (Rx, Ry, Rz) of the rotation in step S52 are obtained in the following manner:

[0025] Rotate the CT image to obtain the coordinates of the two vertebral center in the X-ray image A (Va1X, Va1Y) and (Va2X, Va2Y), and obtain the two vertebral centers in the corresponding CT image (Vct1X, Vct1Y, Vct1Z) and (Vct2X, Vct2Y, Vct2Z);

[0026] If (Va1Y - Va2Y)*(Vct1Z - Vct2Z) < 0, increase Ry by 180;

[0027] Take the point light source above the patient as the benchmark. If the point light source is below the patient, increase Rz by 180.

[0028] The matrix R (Rx, Ry, Rz) is the initial parameters of the three-axis rotation.

[0029] Optionally, the initial parameters of the three-axis translation in step S6 are calculated in the following manner:

[0030] Step S61: Calculate Ty according to the X-ray image A and the X-ray lateral image B in the reference space;

[0031] Step S62: Calculate Tx and Tz according to the X-ray anterior image A.

[0032] Optionally, Ty is calculated in step S61 in the following manner:

[0033] In the reference space S, the coordinates of the point light source of X-ray image A are La(0, 0, 0), and the unit ray vector of X-ray image A is La(0, 0, 1). In the coordinate system of X-ray image B, the coordinates of the point light source are Lb(0, 0, 0), and the unit ray vector of X-ray image B is Lb(0, 0, 1).

[0034] Calculate the vector Lb'=Tba*Lb of Lb in space S. Calculate the projection vector Lb' of Lb' on the xz plane of space, and calculate the coordinates P(0, 0, Pz) of the intersection of vector Lb' and La;

[0035] Calculate the corrected pixel spacing rb of the X-ray image B = |Vct1Z - Vct2Z| / |Vb1Y - Vb2Y|;

[0036] Calculate the y-axis translation of the CT image Ty=Pz+(rb*Vb1X-Vct1Y).

[0037] Optionally, in step S62, Tx and Tz are calculated in the following manner:

[0038] Calculate the corrected pixel spacing ra of the X-ray image A = |Vct1Z - Vct2Z| / |Va1Y - Va2Y|;

[0039] Calculate the x-axis translation of the CT image Tx=ra*Va1X-Vct1X;

[0040] Calculate the z-axis translation of the CT image Tz=ra*Va1Y-Vct1Z;

[0041] The matrix T (Tx, Ty, Tz) obtained is the initial parameters of the three-axis translation.

[0042] A system for rapidly acquiring initial values ​​of spine registration, using the method for rapidly acquiring initial values ​​of spine registration, comprises:

[0043] A data acquisition module is used to acquire CT images and X-ray images and read relevant data in the images;

[0044] The module for calculating the initial parameters of the three-axis rotation includes a posture judgment submodule and a direction correction submodule, which are used to calculate the initial parameters of the three-axis rotation;

[0045] A translation three-axis initial parameter calculation module, including a lateral image processing submodule and an orthogonal image processing submodule, is used to calculate the translation three-axis initial parameters;

[0046] Initial parameter output module: used to output the calculated initial parameters of the three-axis rotation and the three-axis translation.

[0047] Beneficial effects of the present invention:

[0048] 1. In the present invention, a small amount of interactive information is input, and according to the spinal posture and surgical characteristics, the six-dimensional initial parameters are quickly calculated, which match well with the final registration result, thereby reducing the dependence on operating experience and significantly improving the calculation efficiency and result accuracy of the registration iterative algorithm;

[0049] 2. In the present invention, there are few interactive operations, which reduces the dependence on operating experience. The calculation formula is highly interpretable and easy to understand. The amount of calculation is low and the calculation speed is fast. The results are accurate, which improves the efficiency of the registration algorithm and shortens the operation time. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The present invention is a schematic flow chart of a method.

[0051] Figure 2 This is a schematic diagram of a framed vertebral segment of the present invention.

[0052] Figure 3 This is a schematic diagram of an initial parameter application of the present invention.

[0053] Figure 4 It is a schematic diagram of a registration result of the present invention. DETAILED DESCRIPTION

[0054] In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] like Figures 1 to 4 As shown, a method for quickly obtaining the initial value of spine registration includes the following contents:

[0056] 1. Obtaining CT-related information

[0057] Read the three-dimensional CT image of the spine, read the spinal posture (HFP / HFS / FFP / FFS) recorded in the CT image label, read the spinal area obtained by the segmentation algorithm (manual segmentation, traditional method, deep model), read the actual spinal posture (HFP / HFS / FFP / FFS) calculated from the segmentation result, and read each vertebral segment number and vertebral segment center.

[0058] 2. Obtaining X-ray related information

[0059] Read two X-ray images A and B with different shooting angles that contain the marker sphere array. The best combination is the standard frontal image A and the standard lateral image B; or try to make image A a frontal image and image B close to a lateral image. Record whether the point light source is above or below the patient when shooting.

[0060] The coordinate system conversion matrix Tba from image B to the orthogonal image A is calculated based on the coordinate matrix corresponding to the marker ball array in the two X-ray images.

[0061] Obtained in two X-ray images (manual marking or box selection) Figure 1 ) and then calculate the center point of the frame and the depth segmentation model) of two different vertebral center coordinates A1 (Va1X, Va1Y), A2 (Va2X, Va2Y) and B1 (Vb1X, Vb1Y), B2 (Vb2X, Vb2Y).

[0062] 3. Calculation of initial parameters of three-axis rotation

[0063] Taking the space S of X-ray image A as the reference, set the position of the point light source of X-ray image A as the origin of the space, so that the projection position of the point light source on image A is the center point of image A, the x-axis and y-axis of the space are respectively consistent with the x-axis and y-axis of image A, and the z-axis of the space is consistent with the emission direction of the point light source ray. The center of the CT image is overlapped with the origin, the x-axis of the CT image is consistent with the x-axis direction of the space, the y-axis of the CT image is consistent with the z-axis direction of the space, and the z-axis of the CT image is consistent with the y-axis direction of the space. At this time, the rotation three-axis matrix (Rx, Ry, Rz) of the CT image is set to (0°, 0°, 0°).

[0064] 3.1. CT image rotation based on CT reconstruction principles and actual spine posture

[0065] Since CT data will be reconstructed based on the posture labels input during acquisition, the target posture is head-first supine (HFS). In order to deal with the problem that the input posture labels may be incorrect, the head-first supine HFS is used as the reference target, and the CT image is rotated step by step according to the CT reconstruction principle and the actual posture of the spine.

[0066] First, determine the CT posture label: if the label is head-first prone (HFP), increase Rz by 180°; if the label is feet-first supine (FFS), increase Ry by 180°; if the label is feet-first prone (FFP), increase Rx by 180°.

[0067] Then determine the actual spinal posture: if it is actually head-first prone (HFP), increase Rz by 180°; if it is actually feet-first supine (FFS), increase Ry by 180°; if it is actually feet-first prone (FFP), increase Rx by 180°

[0068] Finally, the head-first prone (HFP) position is commonly used in spinal surgery, thus increasing Rz by 180°.

[0069] 3.2. CT image rotation obtained by comparison with X-ray image information

[0070] Due to different processing or settings of different X-ray machines, the spine direction in the obtained X-ray image may be different from the patient's spine direction in space, so the CT image is rotated again. The coordinates of the centers of the two vertebrae in the interactively obtained X-ray image A are (Va1X, Va1Y) and (Va2X, Va2Y), and the corresponding coordinates of the centers of the two vertebrae in the CT image are (Vct1X, Vct1Y, Vct1Z) and (Vct2X, Vct2Y, Vct2Z). If (Va1Y - Va2Y)*(Vct1Z - Vct2Z)<0, increase Ry by 180°.

[0071] Taking the point light source above the patient as the benchmark, if the point light source is below the patient, increase Rz by 180°.

[0072] The matrix R (Rx, Ry, Rz) obtained after a series of rotation processing is the initial parameters of the three-axis rotation.

[0073] 4. Calculation of initial parameters of three-axis translation

[0074] It is assumed that the center of the CT image coincides with the origin of space S, that is, the position of the point light source A in the X-ray image, and that the y-axis of the CT image coincides with the z-axis of space, that is, the direction of emission of the point light source rays. The CT image needs to be moved to the actual position of the spine so that the projection of the CT image under the point light source can be similar to the X-ray image.

[0075] 4.1. Calculation of Ty based on the X-ray lateral image B

[0076] In the reference space S, the coordinates of the point light source of X-ray image A are La(0, 0, 0), and the unit ray vector of X-ray image A is La(0, 0, 1). In the coordinate system of X-ray image B, the coordinates of its point light source are Lb(0, 0, 0), and the unit ray vector of X-ray image B is Lb(0, 0, 1). Calculate the vector Lb'=Tba*Lb of Lb in space S. Calculate the projection vector Lb'' of Lb' on the xz plane of space, and calculate the coordinates of the intersection of vector Lb'' and La, P(0, 0, Pz).

[0077] Since X-ray images are projected images, their image pixel spacing cannot represent the real space spacing and needs to be corrected. The pixel spacing of CT images can be used as a correct benchmark to calculate the corrected pixel spacing rb of X-ray image B = |Vct1Z- Vct2Z| / |Vb1Y - Vb2Y|.

[0078] Calculate the y-axis translation of the CT image Ty=Pz+(rb*Vb1X-Vct1Y).

[0079] 4.2 Calculation of Tx and Tz based on X-ray image A

[0080] Calculate the corrected pixel pitch ra=|Vct1Z - Vct2Z| / |Va1Y - Va2Y| of the X-ray image A.

[0081] Calculate the x-axis translation of the CT image Tx=ra*Va1X-Vct1X.

[0082] Calculate the z-axis translation of the CT image Tz=ra*Va1Y-Vct1Z.

[0083] The matrix T (Tx, Ty, Tz) obtained is the initial parameters of the three-axis translation.

[0084] like Figure 3 As shown in the figure, the initial parameter application is schematically shown: (a) and (b) are X-ray images A and B, respectively, which are standard frontal and lateral images; (a1) and (b1) are the frontal and lateral images of the CT image after the initial translation parameters are applied; (a2) and (b2) are the frontal and lateral images of the CT image after the initial translation and rotation parameters are applied.

[0085] CT image and X-ray image registration

[0086] The initial parameters of rotation and translation are input into the registration process, and the registration result is obtained through an iterative algorithm.

[0087] like Figure 4 As shown in the figure, the registration results are shown as follows: (a) is the matching of the CT projection image and the X-ray image using the initial parameters of translation and rotation; (b) is the registration result of the CT image and the X-ray image obtained by the registration algorithm.

[0088] A system for rapidly acquiring initial values ​​of spine registration, using the method for rapidly acquiring initial values ​​of spine registration, comprises:

[0089] A data acquisition module is used to acquire CT images and X-ray images and read relevant data in the images;

[0090] The module for calculating the initial parameters of the three-axis rotation includes a posture judgment submodule and a direction correction submodule, which are used to calculate the initial parameters of the three-axis rotation;

[0091] A translation three-axis initial parameter calculation module, including a lateral image processing submodule and an orthogonal image processing submodule, is used to calculate the translation three-axis initial parameters;

[0092] Initial parameter output module: used to output the calculated initial parameters of the three-axis rotation and the three-axis translation.

[0093] The present invention uses a small amount of interactive information input to quickly calculate the six-dimensional initial parameters according to the spinal posture and surgical characteristics, which match well with the final registration result, thereby significantly improving the calculation efficiency and result accuracy of the registration iterative algorithm while reducing the dependence on operating experience;

[0094] The present invention has few interactive operations, reduces dependence on operating experience, has a highly interpretable calculation formula that is easy to understand, has a low amount of calculation and thus a fast calculation speed, and accurately results in improved efficiency of the registration algorithm and shortened surgery time.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent substitution or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for quickly acquiring the initial value of spine registration, characterized in that: The following steps are involved: Step S1: Obtain a 3D CT image of the spine, read the spine posture recorded in the label, and obtain the spine region and each vertebral segment number and vertebral segment center through a segmentation algorithm; Step S2: obtaining at least two X-ray images of different angles containing the marker ball array, and recording the position of the point light source during the shooting; Step S3: Calculate the coordinate system conversion matrix according to the X-ray image, and calculate the coordinate system conversion matrix Tba from image B to the orthogonal image A according to the coordinate matrix corresponding to the marker ball array in the two X-ray images; Step S4: manually mark or select the X-ray image to calculate the center coordinates of two different vertebrae; Step S5: Taking the space of an X-ray image as a reference, setting the point light source position of the X-ray image as the spatial origin, and calculating the initial parameters of the three-axis rotation according to the information comparison between the CT image and the X-ray image; Step S6: calculating the initial parameters of the three-axis translation, including calculating the y-axis translation Ty according to the X-ray lateral image, and calculating the x-axis translation Tx and the z-axis translation Tz according to the X-ray anteroposterior image, to obtain the initial parameters of the three-axis translation; Rotate the CT image to obtain the coordinates of the two vertebral center in the X-ray image A (Va1X, Va1Y) and (Va2X, Va2Y), and obtain the two vertebral centers in the corresponding CT image (Vct1X, Vct1Y, Vct1Z) and (Vct2X, Vct2Y, Vct2Z); Calculate Ty in the following way: In the reference space S, the coordinates of the point light source of X-ray image A are Pa(0, 0, 0), and the unit ray vector of X-ray image A is La(0, 0, 1). In the coordinate system of X-ray image B, the coordinates of the point light source are Pb(0, 0, 0), and the unit ray vector of X-ray image B is Lb(0, 0, 1). In the X-ray shadow box, the corresponding coordinates of two different vertebral center points A1 (Va1X, Va1Y), A2 (Va2X, Va2Y) and B1 (Vb1X, Vb1Y), B2 (Vb2X, Vb2Y) are obtained. Calculate the vector Lb'=Tba*Lb of Lb in space S. Calculate the projection vector Lb' of Lb' on the xz plane of space, and calculate the coordinates P(0, 0, Pz) of the intersection of vector Lb' and La; Calculate the corrected pixel spacing rb of the X-ray image B = |Vct1Z - Vct2Z| / |Vb1Y - Vb2Y|; Calculate the y-axis translation of the CT image Ty=Pz+(rb*Vb1X-Vct1Y); Calculate Tx and Tz in the following way: Calculate the corrected pixel spacing ra of the X-ray image A = |Vct1Z - Vct2Z| / |Va1Y - Va2Y|; Calculate the x-axis translation of the CT image Tx=ra*Va1X-Vct1X; Calculate the z-axis translation of the CT image Tz=ra*Va1Y-Vct1Z; The matrix T (Tx, Ty, Tz) is the initial parameters of the three-axis translation; Step S7: Input the initial parameters of the three axes of rotation and the three axes of translation into the registration, obtain the registration result, and verify it.

2. The method for rapidly acquiring the initial value of spine registration according to claim 1, characterized in that: In step S2, the best combination of the two X-ray images at different angles is a standard frontal view and a standard lateral view.

3. The method for rapidly acquiring the initial value of spine registration according to claim 1, characterized in that: In step S5, the initial parameters of the three axes of rotation are calculated in the following manner: Step S51: gradually adjusting the rotation parameters according to the CT image posture label and the actual spine posture, so that the CT image posture is consistent with the standard surgical posture; Step S52: According to the difference between the spine direction in the X-ray image and the spine direction in the CT image, the rotation parameters Ry and Rz are adjusted to make the spine directions in the CT image consistent with those in the X-ray image, and the initial rotation parameters R (Rx, Ry, Rz) of the three axes are obtained.

4. The method for rapidly acquiring the initial value of spine registration according to claim 3, characterized in that: The rotation parameters are adjusted in step S51 in the following manner: Step A1: Determine the CT posture label: if the label is head-first prone, increase Rz by 180°; if the label is feet-first supine, increase Ry by 180°; if the label is feet-first prone, increase Rx by 180°; Step A2: Determine the actual spine posture: if the actual position is head first, increase Rz by 180°; if the actual position is feet first, increase Ry by 180°; if the actual position is feet first, increase Rx by 180°; Step A3, the common posture in spinal surgery is head-first prone, increasing Rz by 180°.

5. The method for rapidly acquiring the initial value of spine registration according to claim 3, characterized in that: In step S52, the three-axis initial parameters R (Rx, Ry, Rz) are rotated in the following manner: Rotate the CT image to obtain the coordinates of the two vertebral center in the X-ray image A (Va1X, Va1Y) and (Va2X, Va2Y), and obtain the two vertebral centers in the corresponding CT image (Vct1X, Vct1Y, Vct1Z) and (Vct2X, Vct2Y, Vct2Z); If (Va1Y - Va2Y)*(Vct1Z - Vct2Z) < 0, increase Ry by 180; Take the point light source above the patient as the benchmark. If the point light source is below the patient, increase Rz by 180. The matrix R (Rx, Ry, Rz) is the initial parameters of the three-axis rotation.

6. The method for rapidly acquiring the initial value of spine registration according to claim 1, characterized in that: The initial parameters of the three-axis translation in step S6 are calculated in the following manner: Step S61: Calculate Ty according to the X-ray image A and the X-ray lateral image B in the reference space; Step S62: Calculate Tx and Tz according to the X-ray anterior image A.

7. A system for rapidly acquiring initial values ​​of spine registration, using the method for rapidly acquiring initial values ​​of spine registration as claimed in claim 1, characterized in that: include: A data acquisition module is used to acquire CT images and X-ray images and read relevant data in the images; The module for calculating the initial parameters of the three-axis rotation includes a posture judgment submodule and a direction correction submodule, which are used to calculate the initial parameters of the three-axis rotation; A translation three-axis initial parameter calculation module, including a lateral image processing submodule and an orthogonal image processing submodule, is used to calculate the translation three-axis initial parameters; Initial parameter output module: used to output the calculated initial parameters of the three-axis rotation and the three-axis translation.

Images