A method and system for generating digitally reconstructed radiographic images

Abstract

The present invention discloses a method and system for generating digitally reconstructed radiographic images, wherein the method includes: obtaining a three-dimensional image of a patient; designing corresponding weights in a forward projection ray transfer function according to different surgical objectives, wherein the weight of the surgical target area is greater than that of other tissues; and performing forward projection using a forward projection model based on the obtained forward projection ray transfer function to obtain a digitally reconstructed radiographic image. The present invention designs different forward projection ray transfer functions according to different surgical objectives, so that the forward projection process focuses on the surgical target structure, and can dynamically adjust the response in the ray transfer function, and obtain a DRR image that meets the requirements of the surgical process through multiple iterations.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and system for generating digitally reconstructed radiographic images. Background Art

[0002] Currently, surgical robots need to obtain information about the patient's body tissue structure and the location and morphology of the lesion before performing surgery. Since CT images have high resolution, fast scanning speed and moderate price (ultrasound images have poor resolution, magnetic resonance imaging has slow scanning speed and high price), CT images are usually used to obtain information about the patient's body tissue structure, and then the doctor formulates a surgical plan.

[0003] During surgery, it is necessary to obtain the spatial position information of the lesion and surgical instruments in real time and accurately to prevent the surgical process from deviating from the surgical plan and causing accidents. Usually, a C-arm X-ray machine is used during surgery to observe the surgical process in real time. The C-arm obtains the attenuation of X-rays passing through human tissue through the detector and then obtains a 2D perspective image. It is difficult to accurately locate the position of the lesion by directly observing the perspective image. It is necessary to align the perspective image with the preoperative CT scan image to obtain more accurate intraoperative information. However, since the preoperative CT image is a reconstructed 3D image of the human tissue structure, it cannot be directly aligned with the 2D perspective image. The usually adopted solution is to generate a digitally reconstructed radiograph (DRR) from the 3D CT image through a simulated ray projection algorithm, and then use the DRR image to align with the C-arm perspective image to achieve real-time positioning during the operation.

[0004] Currently, the calculation process of DRR images usually requires manual adjustment of the ray projection weights of different tissues to obtain clear and accurate DRR images. In addition, compared with conventional clinical CT, animal CT images often lack standards for CT value calibration, and the image CT values ​​fluctuate greatly, resulting in a more unstable DRR image acquisition process, which increases the time and complexity of the DRR acquisition process. Summary of the Invention

[0005] Purpose of the invention: In response to the above-mentioned shortcomings, the present invention proposes a method and system for generating digitally reconstructed radiographic images. The forward projection process focuses on the surgical target structure and can dynamically adjust the response in the ray transfer function. Multiple iterations are performed to obtain DRR images that meet the requirements of the surgical process.

[0006] Technical solution:

[0007] A method for generating a digitally reconstructed radiographic image, comprising:

[0008] Obtain three-dimensional images of patients;

[0009] The weights in the corresponding forward projection ray transfer function are designed according to different surgical objectives, where the weight of the surgical target area is greater than that of other tissues;

[0010] The digitally reconstructed radiographic image is obtained by performing orthographic projection on the orthographic projection model according to the obtained orthographic projection ray transfer function.

[0011] After acquiring the patient's three-dimensional image, a recalibration step is also included:

[0012] The voxel value distribution in the three-dimensional image is counted, the voxel values ​​of soft tissue and air in the three-dimensional image are identified, the voxel values ​​of soft tissue and air are recalibrated, and the voxel values ​​of other tissues are recalibrated based on the voxel values ​​of soft tissue and air and their recalibrated values.

[0013] Coarsely dividing the voxel values ​​in the three-dimensional image by a first interval;

[0014] The groups represented by air and soft tissue are distinguished according to the number of voxel occurrences in each group, and the voxel value range R corresponding to air and soft tissue is obtained accordingly. A 、R S ;

[0015] Voxel value range R corresponding to air and soft tissue A 、R S The second interval is used to divide the voxel values ​​in the group with the largest number of voxel occurrences into the average value E A and E S as the current voxel value of air and soft tissue;

[0016] The voxel values ​​of air and soft tissue are recalibrated, and the voxel values ​​of other tissues are recalibrated according to the voxel values ​​of air and soft tissue and their recalibrated values.

[0017] Before recalibration, the process also includes: identifying a region of interest based on a round window mask in the three-dimensional image, and uniformly setting the voxel values ​​outside the mask to 0.

[0018] The weights in the orthographic projection ray transfer function designed according to different surgical objectives are as follows:

[0019] 1) For tissues with clearly distinguishable voxel value distribution, the ray transfer function f n Set as a piecewise function spaced by voxel values:

[0020]

[0021] Among them, v1, v2, ..., v j Respectively represent the voxel value separation value; λ1, λ2, ..., λ j Represent the corresponding weights, u nis the voxel value corresponding to the voxel point n that the ray passes through, where the weight of the voxel point corresponding to the surgical target area is greater than the weight of the voxel point in other areas;

[0022] 2) For tissues with no obvious difference in voxel value distribution, the spatial coordinate range of each tissue region is obtained by segmenting the 3D image, and the ray transfer function f is set n , as follows:

[0023]

[0024] Among them, Ω1, Ω2, ..., Ω j Respectively represent the spatial coordinate range of the voxel points corresponding to each region, w n Represents the spatial coordinates of the region corresponding to voxel point n, where the weight of the voxel point corresponding to the surgical target area is greater than the weight of the voxel points in other areas.

[0025] The segmentation method for segmenting the three-dimensional image to obtain the spatial coordinates of the corresponding area of ​​each tissue adopts region growing, graph cut, UNet or VNet neural network segmentation algorithm.

[0026] Also includes the calibration steps:

[0027] The calculated voxel value distribution on the digitally reconstructed radiographic image is compared with the expected voxel value distribution of the pre-set surgical target area to determine whether it meets the expectations; if so, the digitally reconstructed radiographic image is output; if not, the weights in the ray transfer function are adjusted to re-project the image, and this step is repeated until the obtained digitally reconstructed radiographic image meets the expected voxel value distribution of the pre-set surgical target area.

[0028] Adjust the ray transfer function f n The weights λ1, λ2, …, λ j Specifically:

[0029] The expected voxel value distribution of the pre-set surgical target area includes the expected mean value range [E1, E2] and the corresponding variance range [V1, V2] of the image voxel values ​​of the surgical target area;

[0030] Calculate the average voxel value E of the surgical target area in the current digitally reconstructed radiographic image k ;

[0031] If E k <E1, then increase the weight of the transfer function corresponding to the surgical target area according to the corresponding difference; if E k >E2, then the weight of the transfer function corresponding to the surgical target area is reduced according to the corresponding difference;

[0032] Through the above adjustment, the average voxel value E of the surgical target area in the digitally reconstructed radiographic image is obtained. k ∈[E1,E2], calculate the corresponding variance V k ;

[0033] If V k <V1, then increase the weight of the transfer function corresponding to the surgical target area according to the corresponding difference; if V k > V2, then the weight of the transfer function corresponding to the surgical target area is reduced according to the corresponding difference;

[0034] Repeat the adjustment until the mean and variance of the voxel values ​​of the surgical target area in the current digitally reconstructed radiographic image meet the expected voxel value distribution, and output the projection result.

[0035] A system for generating a digitally reconstructed radiographic image, comprising:

[0036] 3D imaging equipment, used to scan the patient's preoperative 3D image and send it to the recalibration module;

[0037] The processing module designs different orthographic projection ray transfer functions according to different surgical objectives, and orthographically projects the recalibrated preoperative three-dimensional image through the orthographic projection model to obtain a digitally reconstructed radiographic image; the design of the orthographic projection ray transfer function is as described above.

[0038] The system also includes a recalibration module for recalibrating the corresponding areas of each tissue in the patient's preoperative three-dimensional image.

[0039] The beneficial effects of the present invention are as follows: the present invention first uniformly calibrates the three-dimensional CT images, and then designs different orthographic projection ray transfer functions according to different surgical targets. The response within the CT value range of structures such as soft tissue and air is set to a lower level, so that the orthographic projection process focuses on the surgical target structure, and the response in the ray transfer function can be dynamically adjusted, and multiple iterations are performed to obtain DRR images that meet the requirements of the surgical process. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A flow chart of a method for generating a digitally reconstructed radiographic image according to the present invention;

[0041] Figure 2 Schematic diagram of ray attenuation for the orthographic projection model. DETAILED DESCRIPTION

[0042] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments.

[0043] The present invention is a method for adaptively generating digitally reconstructed radiographic images that meet surgical needs based on three-dimensional image data. Figure 1 As shown, the following steps are included:

[0044] (1) Scan and obtain the patient's preoperative three-dimensional image, and perform unified recalibration on the preoperative three-dimensional image;

[0045] In the present invention, a preoperative three-dimensional image of a patient is obtained by scanning. The voxel values ​​of the image data have generally been calibrated to a standard. The voxel values ​​of the air and soft tissue areas are generally uniformly calibrated. For example, the image area where the soft tissue is located is calibrated to 0, and the image area where the air is located is calibrated to -1000. Some data, such as preoperative three-dimensional image data of animals, have not been calibrated to a standard. In the present invention, both standard-calibrated and non-standard-calibrated three-dimensional image data need to be uniformly recalibrated.

[0046] The present invention first identifies the region of interest based on the round window mask in the 3D image, uniformly sets the voxel values ​​outside the mask to 0 to eliminate interference, and obtains the preoperative 3D image. The preoperative 3D image data is then uniformly recalibrated. Specifically, the image area containing air is calibrated as t, and the image area containing soft tissue is calibrated as c. In this embodiment, t = 0 and c = 1000. Then, the corresponding area in the 3D image needs to be identified. Specifically, the voxel value distribution in the 3D image can be statistically analyzed using the histogram method:

[0047] The voxel values ​​in the three-dimensional image are roughly divided by a first interval and a histogram is constructed;

[0048] Because the distribution differences of the voxel values ​​of air, soft tissue and bone tissue in the histogram are very obvious, the groups represented by air, soft tissue and bone tissue can be distinguished according to the number of voxel occurrences in the corresponding groups in the histogram, thereby obtaining the corresponding areas in the three-dimensional image and the corresponding voxel value range R. A 、R S and R B ; In the three-dimensional image, the voxel values ​​that appear more frequently are air, soft tissue, and bone tissue. According to the distribution of human tissues, the voxel value occurrence frequency decreases in the order of soft tissue, bone tissue, and air. The voxel value ranges of soft tissue, bone tissue, and air can be obtained based on the histogram statistics of the voxel distribution.

[0049] Voxel value range R corresponding to air and soft tissue A 、R S The second interval is divided into two groups respectively, and a histogram is constructed, and the average value E of the voxel value in the group with the largest number of voxels is taken as A and E S As the current voxel value of air and soft tissue, the voxel value of each voxel in the three-dimensional image can be recalibrated according to the above recalibration of air and soft tissue using the following formula:

[0050] I1=(I0-EA ) / (E S -E A )*(ct)+t

[0051] Where I1 represents the recalibrated voxel value of a voxel in the 3D image, and I0 represents the original voxel value of the voxel. According to the general design practice, t = 0, c = 1000, then

[0052] I1=(I0-E A ) / (E S -E A )*1000

[0053] In the above process, when counting the voxel value ranges of air, soft tissue, and bone tissue, the statistics can be accelerated by extracting the original data three-dimensional image or downsampling some slices thereof, so as to complete the calibration more quickly.

[0054] In the present invention, CT scanning is used to obtain the patient's preoperative CT image.

[0055] (2) For the recalibrated preoperative three-dimensional image obtained in step (1), different orthographic projection ray transfer functions are designed according to different surgical objectives, and a digitally reconstructed radiographic image is obtained by performing orthographic projection on the orthographic projection model;

[0056] The orthographic projection model refers to the modeling of the process in which rays are emitted from the tube, passed through the scanned object, are absorbed and attenuated, and finally irradiate the detector sensor unit. Therefore, the various parameters of the orthographic projection are known, including the tube coordinates, detector coordinates, detector size, resolution, and projection fan angle. The orthographic projection model can be generally divided into point model, line model, and area model according to the size of the tube, the size of the detector unit, and the differences in ray shape modeling. In addition, it can also be divided into pixel-driven model, ray-driven model, distance-driven model, etc. according to the calculation process of the projection value.

[0057] The ray attenuation process of the orthographic projection model obeys the Lambert-Beer law, that is:

[0058]

[0059] Among them, I i is the intensity of the ray before it penetrates the target, I0 is the intensity of the ray after it penetrates the target, and p is the line integral of the attenuation coefficient on the ray path, that is, the Radon transform of the ray path:

[0060] p=∫u(x)dx

[0061] Where u(x) is the attenuation coefficient at position x on the ray. Taking the line model as an example, the orthographic projection model of the present invention is represented on the 3D image as the voxel value of the voxel point x on the ray. Therefore, the above formula can be discretized as:

[0062]

[0063] Where N is the number of voxel points passed through by the ray path; u n is the voxel value corresponding to the voxel point n that the ray passes through; l n is the length of the ray within the voxel point n, which can be calculated by connecting the intersection points of the ray and the edges of each voxel, such as Figure 2 As shown;

[0064] The present invention can set different weights f according to the different responses of each voxel point on the ray path. n , as follows:

[0065]

[0066] Among them, the weight is the ray transfer function f n Designed according to different surgical objectives, a higher weight is set for the area where the surgical target is located, and a lower weight is set for other areas, so that the orthographic projection result can obtain a clear surgical target and reduce the influence of other tissues; the details are as follows:

[0067] 1) For parts where the voxel value distribution can be clearly distinguished, such as human bones, the ray transfer function f n It can be set as a piecewise function with intervals of voxel values:

[0068]

[0069] Among them, v1, v2, ..., v j They represent the voxel value separation values, which have a natural distinction for the parts of the voxel value distribution that can be clearly distinguished, so it can be clearly obtained; λ1, λ2, ..., λ j They represent the corresponding weight parameters, and their initial values ​​are set based on experience. The weight of the voxel point corresponding to the surgical target is greater than the weight of the voxel points in other areas.

[0070] 2) For parts where the voxel value distribution is not obviously different, such as various soft tissue organs (liver, prostate, etc.), the ray transfer function f n The surgical target cannot be distinguished directly by voxel values, so the first step is to segment the surgical target into different regions, so that the surgical target can be obtained. The specific segmentation method can use traditional segmentation algorithms such as region growing and graph cut, or neural network segmentation algorithms such as UNet and VNet. After the surgical target is segmented, the ray transfer function f is set according to the coordinate range of the surgical target. n , as follows:

[0071]

[0072] Among them, Ω1, Ω2, ..., Ω j They represent the voxel point sets corresponding to the regions segmented by the aforementioned segmentation algorithms, w n Represents the spatial coordinates corresponding to the voxel point n; the weight of the voxel point corresponding to the surgical target is greater than the weight of the voxel points in other areas;

[0073] The orthographic projection model of the present invention can also adopt other models, but the ray transfer function f n The design method according to different surgical goals is consistent with the wire model, as above;

[0074] (3) According to step (2), the digitally reconstructed radiographic image is obtained, and the voxel value distribution on the digitally reconstructed radiographic image is calculated. The digitally reconstructed radiographic image is compared with the expected voxel value distribution of the pre-set surgical target area to determine whether it meets the expectation. If it meets the expectation, the digitally reconstructed radiographic image is output. If it does not meet the expectation, the ray transfer function f is adjusted. n The weight parameters λ1, λ2, ..., λ j Re-projection is performed and this step is repeated until the digitally reconstructed radiographic image meets the expected voxel value distribution of the pre-set surgical target area;

[0075] Adjust the ray transfer function f n The weight parameters λ1, λ2, ..., λ j Specifically:

[0076] The expected voxel value distribution of the preset surgical target area includes the expected mean value range [E1, E2] and the corresponding variance range [V1, V2] of the image voxel values ​​of the surgical target area;

[0077] Calculate the average voxel value E of the surgical target area in the current digitally reconstructed radiographic image k ;

[0078] If E k <E1, then the weight parameter in the transfer function corresponding to the surgical target area is increased according to the corresponding difference, as follows:

[0079]

[0080] in, is the ray transfer function f n The updated value of the mth weight parameter, is the ray transfer function f n The current value of the mth weight parameter, α is the weight adjustment parameter corresponding to the average value of the voxel value;

[0081] If E k> E2, then the weight parameter in the transfer function corresponding to the surgical target area is reduced according to the corresponding difference, as follows:

[0082]

[0083] That is, through the above adjustment, the average voxel value E of the surgical target area in the digitally reconstructed radiographic image is obtained. k ∈[E1,E2], and then calculate the corresponding variance V k , adjust it as follows:

[0084] If V k < V1, then the weight parameter in the transfer function corresponding to the surgical target area is increased according to the corresponding difference, as follows:

[0085]

[0086] Among them, β is the weight adjustment parameter corresponding to the variance;

[0087] If V k > V2, then the weight parameter in the transfer function corresponding to the surgical target area is reduced according to the corresponding difference, as follows:

[0088]

[0089] In the present invention, when adjusting the weight parameters of the ray transfer function, the weight parameters in the ray transfer function are first adjusted according to the average value of the voxel values ​​until the average value of the voxel values ​​is within the average value threshold range, and then the weight parameters in the ray transfer function are adjusted according to the variance of the voxel values ​​until the variance of the voxel values ​​is within the variance threshold range. When the average value of the voxel values ​​and the variance of the voxel values ​​both meet the threshold range, the projection result is output.

[0090] The present invention also provides a system for generating a digitally reconstructed radiographic image, comprising:

[0091] A three-dimensional imaging device is used to scan the patient's preoperative three-dimensional image and send it to the recalibration module; in the present invention, the three-dimensional imaging device is a CT device;

[0092] The recalibration module is set in the host computer and is used to recalibrate the patient's preoperative three-dimensional images;

[0093] The processing module is set in the host computer, and different orthographic projection ray transfer functions are designed according to different surgical objectives. The orthographic projection model is used to orthographically project the recalibrated preoperative three-dimensional image to obtain a digitally reconstructed radiographic image. The design of the orthographic projection ray transfer function is as described above.

[0094] The correction module corrects the digitally reconstructed radiographic image obtained by the processing module. The specific correction method is as described in the above step (3).

[0095] The present invention infers the CT value ranges of different structures, such as air, soft tissue, and bone, based on the statistical distribution of pixels in the input three-dimensional CT image. After obtaining the CT values ​​of each tissue, the entire three-dimensional image is further calibrated (for example, air is 0 and water is 1000) to obtain uniformly calibrated data. Different forward projection ray transfer functions are then designed based on different surgical objectives. For example, if orthopedic surgery focuses primarily on bone tissue, the forward projection ray transfer function is set to a higher response within the bone tissue CT value range and a lower response within the CT value range of structures such as soft tissue and air, so that the forward projection process focuses on the surgical target structure. The forward projection process can adopt pixel-driven, ray-driven, or distance-driven models. The forward projection model combines the response of the ray transfer function after the recalibration of the CT value of each pixel to obtain a DRR image that focuses on the surgical target structure. Based on the DRR image, the pixel distribution range of the corresponding target structure is analyzed. If it does not meet the pre-set distribution, the response in the ray transfer function can be dynamically adjusted. Multiple iterations are performed to obtain a DRR image that meets the requirements of the surgical process. The embodiments disclosed in the present invention are applicable to the field of medical surgical instruments.

[0096] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations (such as quantity, shape, position, etc.) can be made to the technical solution of the present invention, and these equivalent transformations all fall within the scope of protection of the present invention.

Claims

1. A method for generating a digitally reconstructed radiographic image, characterized in that: include: Obtain three-dimensional images of patients; The weights in the corresponding forward projection ray transfer function are designed according to different surgical objectives, where the weight of the surgical target area is greater than that of other tissues; According to the obtained orthographic projection ray transfer function, the orthographic projection model is used to perform orthographic projection to obtain a digitally reconstructed radiographic image; The calculated voxel value distribution on the digitally reconstructed radiographic image is compared with the expected voxel value distribution of the pre-set surgical target area to determine whether it meets the expectations; if so, the digitally reconstructed radiographic image is output; if not, the weights in the ray transfer function are adjusted to re-perform the forward projection, and this step is repeated until the obtained digitally reconstructed radiographic image meets the expected voxel value distribution of the pre-set surgical target area; Among them, the adjustment ray transfer function Weight 、 、…、 Specifically: The expected voxel value distribution of the preset surgical target area includes the expected mean value range [E1, E2] and the corresponding variance range [V1, V2] of the image voxel values ​​of the surgical target area; Calculate the average voxel value E of the surgical target area in the current digitally reconstructed radiographic image k ; If E k <E1, then increase the weight of the transfer function corresponding to the surgical target area according to the corresponding difference; if E k >E2, then the weight of the transfer function corresponding to the surgical target area is reduced according to the corresponding difference; Through the above adjustment, the average voxel value E of the surgical target area in the digitally reconstructed radiographic image is obtained. k ∈[E1,E2], calculate the corresponding variance V k ; If V k <V1, then increase the weight of the transfer function corresponding to the surgical target area according to the corresponding difference; if V k > V2, then the weight of the transfer function corresponding to the surgical target area is reduced according to the corresponding difference; Repeat the adjustment until the mean and variance of the voxel values ​​of the surgical target area in the current digitally reconstructed radiographic image meet the expected voxel value distribution, and output the projection result.

2. The method for generating a digitally reconstructed radiographic image according to claim 1, wherein: After acquiring the patient's three-dimensional image, the method further includes a recalibration step: The voxel value distribution in the three-dimensional image is counted, the voxel values ​​of soft tissue and air in the three-dimensional image are identified, the voxel values ​​of soft tissue and air are recalibrated, and the voxel values ​​of other tissues are recalibrated based on the voxel values ​​of soft tissue and air and their recalibrated values.

3. The method for generating a digitally reconstructed radiographic image according to claim 2, wherein: Coarsely dividing the voxel values ​​in the three-dimensional image by a first interval; The groups represented by air and soft tissue are distinguished according to the number of voxel occurrences in each group, and the voxel value range R corresponding to air and soft tissue is obtained accordingly. A 、R S ; Voxel value range R corresponding to air and soft tissue A 、R S The second interval is used to divide the voxel values ​​in the group with the largest number of voxel occurrences into the average value E A and E S as the current voxel value of air and soft tissue; The voxel values ​​of air and soft tissue are recalibrated, and the voxel values ​​of other tissues are recalibrated according to the voxel values ​​of air and soft tissue and their recalibrated values.

4. The method for generating a digitally reconstructed radiographic image according to claim 2, wherein: Before recalibration, the process also includes: identifying a region of interest based on a round window mask in the three-dimensional image, and uniformly setting the voxel values ​​outside the mask to 0.

5. The method for generating a digitally reconstructed radiographic image according to claim 1, wherein: The weights in the orthographic projection ray transfer function designed according to different surgical objectives are as follows: 1) For tissues with clearly distinguishable voxel value distribution, ray transfer function Set as a piecewise function spaced by voxel values: ； in, 、 、…、 Respectively represent the voxel value separation value; 、 、…、 Represent the corresponding weights, is the voxel value corresponding to the voxel point n that the ray passes through, where the weight of the voxel point corresponding to the surgical target area is greater than the weight of the voxel point in other areas; 2) For tissues with no obvious difference in voxel value distribution, the spatial coordinate range of each tissue region is obtained by segmenting the 3D image, and the ray transfer function is set , as follows: ； in, 、 、…、 Respectively represent the spatial coordinate range of the voxel points corresponding to each area, Represents the spatial coordinates of the region corresponding to voxel point n, where the weight of the voxel point corresponding to the surgical target area is greater than the weight of the voxel points in other areas.

6. The method for generating a digitally reconstructed radiographic image according to claim 5, wherein: The segmentation method for segmenting the three-dimensional image to obtain the spatial coordinates of the corresponding area of ​​each tissue adopts region growing, graph cut, UNet or VNet neural network segmentation algorithm.

7. A system for generating a digitally reconstructed radiographic image, configured to execute the method for generating a digitally reconstructed radiographic image according to any one of claims 1 to 6, comprising: 3D imaging equipment, used to scan the patient's preoperative 3D image and send it to the recalibration module; The processing module designs different orthographic projection ray transfer functions according to different surgical objectives, and orthographically projects the recalibrated preoperative 3D image through the orthographic projection model to obtain a digitally reconstructed radiographic image.

8. The system for generating digitally reconstructed radiographic images according to claim 7, wherein: The system also includes a recalibration module for recalibrating the corresponding areas of each tissue in the patient's preoperative three-dimensional image.

Images