Method and device for automatically registering and splicing multiple sections of cone beam CT (Computed Tomography) images

Abstract

The invention discloses a cone-beam CT image multi-segment registration and splicing method and device, complete anatomical images of organs can be acquired, and the method comprises the following steps: acquiring volume data of a plurality of cone-beam CTs through movement of a bed body; searching a reference image layer corresponding to registration of the previous data body; determining a to-be-registered image layer of the next data body; calculating a transformation matrix between the to-be-registered image layer and the reference image layer; performing mapping correspondence on the next data body according to the obtained transformation matrix; combining adjacent sequences; and so on, a final splicing sequence is obtained. Cone beam CT acquisition is adopted, a plurality of cone beam CT reconstruction sequences can be set, a bed body can be automatically translated, the adjacent cone beam CT sequences are subjected to registration and splicing processing, the CT scanning width is expanded, and the completeness and continuity of cone beam CT reconstruction organs are guaranteed. Aiming at the problem that the scanning range of traditional cone beam CT equipment is limited, the cone beam CT scanning width is expanded, and the completeness and continuity of CBCT reconstruction organs are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of medical equipment technology for CT imaging, specifically a method for automatic registration and stitching of multi-segment cone-beam CT images. Background Technology

[0002] Medical image registration is a current research hotspot in the medical field, with profound significance in clinical diagnosis and treatment. Medical image registration is a commonly used technique in medical image analysis. It involves transforming the coordinates of one image to that of another reference image, matching corresponding positions in the two images to obtain a registered image. There are many methods for medical image registration, mainly including rigid body transformation, non-rigid body transformation, and affine transformation.

[0003] The scanning range of cone-beam CT is often limited by hardware size and cannot meet the requirements of many applications, such as imaging large objects that exceed the CT scanning range. Therefore, how to extend the scanning range of cone-beam CT has significant theoretical and practical value. Image stitching can extend the scanning range of cone-beam CT, enabling a complete view of the imaged area.

[0004] During cone-beam CT stitching, the automatic translation of the CT bed causes a certain degree of rotation and translation in the reconstructed image. Therefore, a transformation matrix is ​​obtained through registration. By adjusting the transformation of the data volumes during stitching, gaps can be eliminated between the data. Summary of the Invention

[0005] The purpose of this invention is to provide a method and apparatus for multi-segment registration and stitching of cone-beam CT images. By setting the number of stitching sequences, data is acquired through automatic bed movement. Based on image mutual information registration, different sequences are registered using reference sequence image layers, and then merged into a single stitching sequence. This solves the problem of limited scanning range in cone-beam CT, increasing its scanning range. It also addresses the translation and rotation of data volumes during bed movement, enabling efficient and flexible imaging of objects of different sizes.

[0006] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0007] A method for automatic registration and stitching of multi-segment cone-beam CT images includes the following steps:

[0008] 1) Acquire cone-beam CT images and construct a reconstruction dataset;

[0009] 2) Calculate the position of the reference image layer corresponding to the data volume registration at the previous acquisition time in the reconstructed dataset;

[0010] 3) Determine the image layer to be registered for the data volume at the next acquisition time;

[0011] 4) Calculate the transformation matrix between the image layer to be registered and the reference image layer;

[0012] 5) Perform a rigid body transformation on the data volume at the next acquisition time according to the transformation matrix to obtain the mapped sequence;

[0013] 6) Repeat steps 2) to 5) to merge adjacent sequences and complete the splicing of cone-beam CT images.

[0014] Step 1) specifically refers to:

[0015] During the movement of the CT bed, cone-beam CT images are acquired. The CT bed moves forward a fixed distance each time according to the set stitching protocol and the number of sequences. The reconstructed dataset is constructed using images from different positions of the acquired object.

[0016] Step 2) specifically refers to:

[0017] overlapSlice=Dis / betweenSlice

[0018] referSlice=zDim-overlapSlice

[0019] Where Dis represents the bed movement distance, betweenSlice represents the interslice spacing of the scan, overlapSlice represents the number of overlapping slices between two scans, zDim represents the total number of slices in the scan data, and referSlice represents the slice corresponding to the reference slice.

[0020] Step 3) specifically refers to:

[0021] The first 5 layers of the data volume at the next acquisition time are selected and compared with the reference layer (referSlice) of the data volume at the previous acquisition time. The slice with the highest similarity is selected as the slice to be registered.

[0022] The similarity metric is specifically as follows:

[0023]

[0024] SSIM(X,Y)=L(X,Y)*C(X,Y)*S(X,Y)

[0025] Where, μ X μ Y Let σ represent the mean of image X and Y, respectively. X σ Y Let X and Y represent the standard deviations of the image, respectively. Let σ represent the variances of images X and Y, respectively. XYLet X represent the covariance of images X and Y, C1, C2, and C3 be constants, L(X,Y) represent the brightness contrast function value, C(X,Y) represent the contrast function value, S(X,Y) represent the structure contrast function value, and SSIM(X,Y) represent the similarity.

[0026] Step 4) includes the following steps:

[0027] 4.1) Map the reference image space to the image space to be registered;

[0028] 4.2) Define the optimizer type and set the optimization parameters;

[0029] 4.3) The similarity measurement module uses the mean square error criterion to initialize the parameters of the registration type and pass the parameters to the registration process. The similarity function is input into the optimization module for optimization calculation to obtain the final transformation parameters.

[0030] 4.4) Repeat step 4.3) until the mean square error reaches its maximum value, and obtain the optimized parameters;

[0031] 4.5) Resample the image to be registered using the final obtained parameters to obtain the registration result, i.e., the transformation matrix.

[0032] A device for automatic registration and stitching of multi-segment cone-beam CT images, including:

[0033] The cone-beam CT acquisition module is used to acquire cone-beam CT images and build a reconstruction dataset.

[0034] The reference layer calculation module is used to calculate the position of the reference image layer corresponding to the data volume registration at the previous acquisition time in the reconstructed dataset;

[0035] The image layer to be registered calculation module is used to determine the image layer to be registered in the data volume at the next acquisition time.

[0036] The layer mapping module is used to calculate the transformation matrix between the image layer to be registered and the reference image layer, and to perform a rigid body transformation on the data volume at the next acquisition time according to the transformation matrix to obtain the mapped sequence.

[0037] The image stitching module is used to merge adjacent sequences to complete the stitching of cone-beam CT images.

[0038] A system for automatic registration and stitching of multi-segment cone-beam CT images includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the method for automatic registration and stitching of multi-segment cone-beam CT images when the computer program is executed.

[0039] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for automatic registration and stitching of multiple segments of cone-beam CT images.

[0040] The present invention has the following beneficial effects and advantages:

[0041] This invention sets up a stitching protocol, the number of stitched data bodies, and the bed movement distance. Through image similarity and registration algorithms, it completes image stitching. After registration, it ensures seamless stitching, making it convenient for imaging technicians to perform full-body scans on large objects. Attached Figure Description

[0042] Figure 1 This is a flowchart of the present invention;

[0043] Figure 2 This is a block diagram of the device of the present invention;

[0044] Figure 3 To register the framework diagram;

[0045] Figure 4 This is a schematic diagram of the splicing process. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0047] If volumetric data is directly stitched together, the gaps between the data volumes will be very noticeable due to translation and rotation during bed movement. The registration-based stitching method significantly reduces these gaps. Furthermore, by finding similar image layers within a given range, it resolves errors caused during bed movement. This invention includes the following steps:

[0048] Multiple cone-beam CT volume data are acquired by moving the bed.

[0049] Calculate the reference image layer corresponding to the previous data volume registration;

[0050] Determine the image layer to be registered for the next data volume;

[0051] Calculate the transformation matrix between the image layer to be registered and the reference image layer;

[0052] Map the next data volume to the corresponding data according to the obtained transformation matrix;

[0053] Merge adjacent sequences; and so on.

[0054] The specific flowchart of this invention is as follows: Figure 1 As shown:

[0055] Step 101: When capturing data, set the stitching protocol and the number of sequences. During the data acquisition process, the bed will move automatically, moving forward a fixed distance each time to acquire reconstructed data of different positions of the object.

[0056] Step 102: Calculate the position of the reference image layer, specifically:

[0057] overlapSlice=Dis / betweenSlice

[0058] referSlice=zDim-overlapSlice

[0059] Where Dis represents the bed movement distance, betweenSlice represents the interslice spacing, overlapSlice represents the number of overlapping slices between two scans, zDim represents the total number of slices in the scan data, and referSlice represents the slice corresponding to the reference slice.

[0060] Step 103: Determine the image layer to be registered for the next data volume, specifically:

[0061] The first 5 layers of the next data volume are selected and compared with the first data reference layer (referSlice) for similarity measurement. The slice with the highest similarity is selected as the slice to be registered. The similarity calculation method is as follows:

[0062]

[0063] μ X μ Y Let σ represent the mean of image X and Y, respectively. X σ Y Let X and Y represent the standard deviations of the image, respectively. Let σ represent the variances of images X and Y, respectively. XY This represents the covariance of the X and Y graphs. C1, C2, and C3 are constants to maintain stability and avoid the denominator being zero.

[0064] The final SSIM index is

[0065] SSIM(X,Y)=L(X,Y)*C(X,Y)*S(X,Y)

[0066] Step 104: Calculate the transformation parameters between the image layer to be registered and the reference image layer, such as... Figure 3As shown, the details are as follows:

[0067] The transformation that maps the reference image space to the image space to be registered;

[0068] Define the optimizer type and set the optimization parameters. Here, the gradient descent method is used, with the initial amplitude set to 1, the minimum step size to 0.001, the relaxation factor to 0.5, and the maximum number of iterations to 300.

[0069] The similarity measurement module uses the mean squared error criterion to initialize the parameters of the registration type and pass the parameters to the registration process. The similarity function is input into the optimization module for optimization calculation to obtain the final transformation parameters. This process is generally implemented through iteration, that is, repeating the above process until the maximum value is obtained, and the optimized parameters are obtained and output.

[0070] The registration result is obtained by resampling the image to be registered using the final parameters.

[0071] Step 105: Map the next data volume to a rigid body transformation according to the obtained transformation matrix.

[0072] Step 106: Merge adjacent sequences; repeat this process to obtain the final concatenated sequence.

[0073] The specific device block diagram of the present invention is as follows: Figure 2 As shown:

[0074] The cone-beam CT acquisition device 201 uses X-ray equipment and rotation acquisition technology to achieve tomographic imaging of data volumes and acquire stitched data volumes.

[0075] The splicing protocol module 202 can set information such as the number of spliced ​​data bodies and the moving distance.

[0076] The bed moving module 203 can move the bed according to the distance information set in the splicing protocol, and continue to move the bed to scan after the scanning is completed.

[0077] Data splicing and display module 204, such as Figure 4 As shown, the final stitched data volume is obtained and displayed in three dimensions using software.

[0078] The modules and units described in this embodiment of the invention may or may not be physically separate. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Following the above steps of this invention, those skilled in the art can understand and implement it without any creative effort.

Claims

1. A method for automatic registration and stitching of multi-segment cone-beam CT images, characterized in that, Includes the following steps: 1) Acquire cone-beam CT images and construct a reconstruction dataset; 2) Calculate the position of the reference image layer corresponding to the data volume registration at the previous acquisition time in the reconstructed dataset; 3) Determine the image layer to be registered for the data volume at the next acquisition time; 4) Calculate the transformation matrix between the image layer to be registered and the reference image layer; 5) Perform a rigid body transformation on the data volume at the next acquisition time according to the transformation matrix to obtain the mapped sequence; 6) Repeat steps 2) to 5) to merge adjacent sequences and complete the splicing of cone-beam CT images.

2. The method for automatic registration and stitching of multiple segments of cone-beam CT images according to claim 1, characterized in that, Step 1) specifically refers to: During the movement of the CT bed, cone-beam CT images are acquired. The CT bed moves forward a fixed distance each time according to the set stitching protocol and the number of sequences. The reconstructed dataset is constructed using images from different positions of the acquired object.

3. The method for automatic registration and stitching of multiple segments of cone-beam CT images according to claim 1, characterized in that, Step 2) specifically refers to: overlapSlice=Dis / betweenSlice referSlice=zDim-overlapSlice Where Dis represents the bed movement distance, betweenSlice represents the interslice spacing of the scan, overlapSlice represents the number of overlapping slices between two scans, zDim represents the total number of slices in the scan data, and referSlice represents the slice corresponding to the reference slice.

4. The method for automatic registration and stitching of multiple segments of cone-beam CT images according to claim 1, characterized in that, Step 3) specifically refers to: The first 5 layers of the data volume at the next acquisition time are selected and compared with the reference layer (referSlice) of the data volume at the previous acquisition time. The slice with the highest similarity is selected as the slice to be registered.

5. The method for automatic registration and stitching of multiple segments of cone-beam CT images according to claim 4, characterized in that, The similarity metric is specifically as follows: SSIM(X,Y)=L(X,Y)*C(X,Y)*S(X,Y) Where, μ X μ Y Let σ represent the mean of image X and Y, respectively. X σ Y Let X and Y represent the standard deviations of the image, respectively. Let σ represent the variances of images X and Y, respectively. XY Let X represent the covariance of images X and Y, C1, C2, and C3 be constants, L(X,Y) represent the brightness contrast function value, C(X,Y) represent the contrast function value, S(X,Y) represent the structure contrast function value, and SSIM(X,Y) represent the similarity.

6. The method for automatic registration and stitching of multiple segments of cone-beam CT images according to claim 1, characterized in that, Step 4) includes the following steps: 4.1) Map the reference image space to the image space to be registered; 4.2) Define the optimizer type and set the optimization parameters; 4.3) The similarity measurement module uses the mean square error criterion to initialize the parameters of the registration type and pass the parameters to the registration process. The similarity function is input into the optimization module for optimization calculation to obtain the final transformation parameters. 4.4) Repeat step 4.3) until the mean square error reaches its maximum value, and obtain the optimized parameters; 4.5) Resample the image to be registered using the final obtained parameters to obtain the registration result, i.e., the transformation matrix.

7. A device for automatic registration and stitching of multi-segment cone-beam CT images, characterized in that, include: The cone-beam CT acquisition module is used to acquire cone-beam CT images and build a reconstruction dataset. The reference layer calculation module is used to calculate the position of the reference image layer corresponding to the data volume registration at the previous acquisition time in the reconstructed dataset. The image layer to be registered calculation module is used to determine the image layer to be registered in the data volume at the next acquisition time. The layer mapping module is used to calculate the transformation matrix between the image layer to be registered and the reference image layer, and to perform a rigid body transformation on the data volume at the next acquisition time according to the transformation matrix to obtain the mapped sequence. The image stitching module is used to merge adjacent sequences to complete the stitching of cone-beam CT images.

8. A system for automatic registration and stitching of multi-segment cone-beam CT images, characterized in that, Including memory and processor; The memory is used to store a computer program; the processor is used to implement the method for automatic registration and stitching of multi-segment cone-beam CT images as described in any one of claims 1-6 when the computer program is executed.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method for automatic registration and stitching of multiple segments of cone-beam CT images as described in any one of claims 1-6.

Images