System for processing unstructured multi-medical image registration algorithm and method of using same

Abstract

According to a system for processing an unstructured multi-medical image registration algorithm and a method of using same, proposed in the present invention, the system comprises: a first image registration processing unit for performing image registration through a normalized cross-correlation (NCCS) process based on a Sobel operator; a second image registration processing unit for performing image registration through a normalized cross-correlation (NCCL) process based on a LOG operator; and a multi-resolution processing unit for performing image processing, with multiple resolutions, on each of registration images of the first image registration processing unit and the second image registration processing unit, thereby improving the accuracy of unstructured image registration by using a registration algorithm between unstructured multi-medical images using normalized cross-correlation based on the Sobel operator or normalized cross-correlation based on the LOG operator.

Description

Atypical multi-medical image registration algorithm processing system and method of using the same

[0001] The present invention was devised as a result of the research on the industry-academia R&BD collaboration commercialization project "Development of an AI-based Body Balance Diagnosis / Correction Platform through Motion Big Data Analysis," which was conducted with the support of the Daegu Digital Innovation Promotion Agency. In particular, it relates to a processing system for an unstructured multi-medical image registration algorithm and a method for using the same. More specifically, it relates to a processing system for an unstructured multi-medical image registration algorithm and a method for using the same that can improve the accuracy of unstructured image registration by utilizing a registration algorithm between unstructured multi-medical images that uses normalized cross-correlation based on a Sobel operator or normalized cross-correlation based on a logarithmic operator.

[0002] The content described in this section merely provides background information regarding an embodiment of the present invention and does not constitute prior art.

[0003]

[0004] In general, medical imaging provides the internal structure of an object and displays to the user structural details, internal tissues, and fluid flow within the body as captured and processed. Examples of such medical imaging include magnetic resonance imaging (MRI) for providing magnetic resonance images, computed tomography (CT) images, X-ray images, and ultrasound images. Medical imaging represents the object in various ways depending on the type of imaging device and the imaging method.

[0005]

[0006] Users, such as doctors, can utilize these medical images to check a patient's health status and diagnose diseases. For example, lung CT images can be used to diagnose lung diseases. Lung CT images are used to diagnose various lung diseases, and for an accurate diagnosis, it is necessary to provide images that include the information required for that diagnostic purpose.

[0007]

[0008] As such, accurate diagnosis of medical images requires the registration of multiple medical images of different, unstructured types. This registration of medical images is performed through a process of extracting features from each of the multiple images and matching the extracted features. This feature matching method can be executed smoothly when the difference in semantic information between images is not significant. However, when the difference in semantic information between images increases due to factors such as differences in sensor domains, there was a problem involving limitations where the accuracy of feature extraction and matching for each image decreases, inevitably leading to a decline in registration performance. Korean Registered Patent Publication No. 10-2285530 and Published Patent Publication No. 10-2017-0096088 are disclosed as prior art documents.

[0009]

[0010] The aforementioned background technology is technical information that the inventor possessed for the derivation of the present invention or acquired during the process of deriving the present invention, and it cannot be considered as publicly known technology disclosed to the general public prior to the filing of the present invention.

[0011] The present invention is proposed to solve the aforementioned problems of previously proposed methods, and aims to provide a processing system for an unstructured multi-medical image registration algorithm and a method for using the same, which can improve the accuracy of unstructured image registration by utilizing an unstructured multi-medical image registration algorithm that uses normalized cross-correlation based on a Sobel operator or normalized cross-correlation based on a logarithmic operator, by comprising a first image registration processing unit that performs image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator, a second image registration processing unit that performs image registration using a normalized cross-correlation (NCCL) process based on a logarithmic operator, and a multi-resolution processing unit that processes the registered images of the first image registration processing unit and the second image registration processing unit at multiple resolutions.

[0012]

[0013] In addition, another objective of the present invention is to provide a processing system for an unstructured multi-medical image registration algorithm and a method for using the same, which can improve the accuracy of unstructured image registration by using a registration algorithm for unstructured multi-medical images that utilizes a normalized cross-correlation based on a Sobel operator or a normalized cross-correlation based on a logarithmic operator, thereby improving the performance of the registration algorithm for unstructured multi-medical images and, along with the improvement in registration accuracy, increasing the convergence range of the algorithm.

[0014]

[0015] However, the technical problem that the present invention aims to solve is not limited to the technical problem described above, and other technical problems may exist.

[0016] An unstructured multi-medical image registration algorithm processing system according to the features of the present invention for achieving the above-mentioned purpose,

[0017] As an atypical multi-medical image registration algorithm processing system,

[0018] A first image matching processing unit that performs image matching using a normalized cross-correlation (NCCS) process based on the Sobel operator;

[0019] A second image matching processing unit that performs image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator; and

[0020] The configuration is characterized by including a multi-resolution processing unit that processes the respective matched images of the first image matching processing unit and the second image matching processing unit at multiple resolutions.

[0021]

[0022] Preferably, the first image matching processing unit is,

[0023] Image registration is performed using a normalized cross-correlation (NCCS) process based on the Sobel operator, and registration can be achieved by comparing the angles between the reference image and the DRR image.

[0024]

[0025] More preferably, the first image matching processing unit is,

[0026] The angle between the reference image and the DRR image is compared to match them, and the smaller the angle between the gradient vectors, the more similar the 2D reference image and the DRR image can become.

[0027]

[0028] Even more preferably, the first image matching processing unit is,

[0029] By comparing the angle between the reference image and the DRR image to perform matching, and making the 2D reference image and the DRR image more similar as the angle between the gradient vectors becomes smaller, the matching result between the 2D reference image and the 3D plotted image can be made more accurate.

[0030]

[0031] Preferably, the second image matching processing unit is,

[0032] Image registration is performed using a normalized cross-correlation (NCCL) process based on the logarithmic (LOG) operator, and registration can be achieved by combining the normalized cross-correlation coefficients and image boundary information.

[0033]

[0034] More preferably, the second image matching processing unit is,

[0035] Image registration is performed using a normalized cross-correlation (NCCL) process based on the logarithmic (LOG) operator, where the LOG (Laplacian Of Gaussian) operator can extract image edge information.

[0036]

[0037] Even more preferably, the second image matching processing unit is,

[0038] Normalized cross-correlation (NCCL) based on the LOG operator is obtained, and the Laplacian image can be obtained by convolving the reference image and the DRR image with the LOG operator.

[0039]

[0040] Preferably, the multi-resolution processing unit is,

[0041] Each of the first image matching processing unit and the second image matching processing unit may be processed at multiple resolutions, and a Gaussian parameter may be used for downsampling to obtain a low-resolution image from an original high-resolution image.

[0042]

[0043] A method for using an unstructured multi-medical image registration algorithm processing system according to the features of the present invention for achieving the above-mentioned purpose is,

[0044] As a method of utilizing an atypical multi-medical image registration algorithm processing system,

[0045] (1) A first image matching processing unit performs image matching using a normalized cross-correlation (NCCS) process based on the Sobel operator;

[0046] (2) A step in which a second image matching processing unit performs image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator; and

[0047] (3) The multi-resolution processing unit is characterized by the step of processing the matching images of the first image matching processing unit and the second image matching processing unit in multiple resolutions.

[0048]

[0049] Preferably, the first image matching processing unit is,

[0050] Image registration is performed using a normalized cross-correlation (NCCS) process based on the Sobel operator, and registration can be achieved by comparing the angles between the reference image and the DRR image.

[0051]

[0052] More preferably, the first image matching processing unit is,

[0053] The angle between the reference image and the DRR image is compared to match them, and the smaller the angle between the gradient vectors, the more similar the 2D reference image and the DRR image can become.

[0054]

[0055] Even more preferably, the first image matching processing unit is,

[0056] By comparing the angle between the reference image and the DRR image to perform matching, and making the 2D reference image and the DRR image more similar as the angle between the gradient vectors becomes smaller, the matching result between the 2D reference image and the 3D plotted image can be made more accurate.

[0057]

[0058] Preferably, the second image matching processing unit is,

[0059] Image registration is performed using a normalized cross-correlation (NCCL) process based on the logarithmic (LOG) operator, and registration can be achieved by combining the normalized cross-correlation coefficients and image boundary information.

[0060]

[0061] More preferably, the second image matching processing unit is,

[0062] Image registration is performed using a normalized cross-correlation (NCCL) process based on the logarithmic (LOG) operator, where the LOG (Laplacian Of Gaussian) operator can extract image edge information.

[0063]

[0064] Even more preferably, the second image matching processing unit is,

[0065] Normalized cross-correlation (NCCL) based on the LOG operator is obtained, and the Laplacian image can be obtained by convolving the reference image and the DRR image with the LOG operator.

[0066]

[0067] Preferably, the multi-resolution processing unit is,

[0068] Each of the first image matching processing unit and the second image matching processing unit may be processed at multiple resolutions, and a Gaussian parameter may be used for downsampling to obtain a low-resolution image from an original high-resolution image.

[0069] According to the unstructured multi-medical image registration algorithm processing system and method of use proposed in the present invention, by comprising a first image registration processing unit that performs image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator, a second image registration processing unit that performs image registration using a normalized cross-correlation (NCCL) process based on a logarithmic operator, and a multi-resolution processing unit that processes the registered images of each of the first image registration processing unit and the second image registration processing unit at multiple resolutions, the accuracy of unstructured image registration can be improved using an unstructured multi-medical image registration algorithm that utilizes normalized cross-correlation based on a Sobel operator or normalized cross-correlation based on a logarithmic operator.

[0070]

[0071] In addition, according to the system for processing an unstructured multi-medical image registration algorithm and the method for using the same of the present invention, the accuracy of unstructured image registration can be improved by using a registration algorithm between unstructured multi-medical images that utilizes normalized cross-correlation based on a Sobel operator or normalized cross-correlation based on a logarithmic operator, thereby improving the performance of the registration algorithm for unstructured multi-medical images and, along with the improvement in registration accuracy, increasing the convergence range of the algorithm.

[0072]

[0073] Furthermore, the various and beneficial advantages and effects of the present invention are not limited to those described above and may be more easily understood in the process of explaining specific embodiments of the present invention.

[0074] FIG. 1 is a diagram illustrating the configuration of an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention in functional blocks.

[0075] FIG. 2 is a diagram illustrating the processing function configuration of a first image registration processing unit of an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention.

[0076] FIG. 3 is a diagram illustrating the processing function configuration of the second image registration processing unit of an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention.

[0077] FIG. 4 is a diagram showing experimental results obtained by applying an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention.

[0078] FIG. 5 is a diagram illustrating the flow of a method of using an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention.

[0079] <Explanation of Symbols>

[0080] 100: A processing system for an unstructured multi-medical image registration algorithm according to an embodiment of the present invention

[0081] 110: 1st image matching processing unit

[0082] 120: Second image matching processing unit

[0083] 130: Multi-resolution processing unit

[0084] S110: A step in which the first image matching processing unit performs image matching using a normalized cross-correlation process based on the Sobel operator.

[0085] S120: A step in which the second image matching processing unit performs image matching using a normalized cross-correlation process based on a logarithmic operator.

[0086] S130: A multi-resolution processing unit processes the respective matching images of the first image matching processing unit and the second image matching processing unit into multiple resolutions.

[0087] Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement the invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals.

[0088]

[0089] Throughout the specification, when a part is described as being "connected" to another part, this includes not only cases where they are "directly connected" but also cases where they are "indirectly connected" with other elements interposed between them. Furthermore, when a part is described as "including" a component, this means that, unless specifically stated otherwise, it does not exclude other components but rather allows for the inclusion of additional components; it should be understood that this does not preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0090]

[0091] The following examples are detailed descriptions to aid in understanding the present invention and are not intended to limit the scope of the present invention. Accordingly, inventions within the same scope that perform the same function as the present invention will also fall within the scope of the present invention.

[0092]

[0093] In addition, each component, process, procedure, or method included in each embodiment of the present invention may be shared within a scope that is not technically contradictory to one another.

[0094]

[0095] FIG. 1 is a diagram illustrating the configuration of an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention in functional blocks. As shown in FIG. 1, the unstructured multi-medical image registration algorithm processing system (100) according to an embodiment of the present invention may be configured to include a first image registration processing unit (110) that performs image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator, a second image registration processing unit (120) that performs image registration using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, and a multi-resolution processing unit (130) that processes the respective registered images of the first image registration processing unit (110) and the second image registration processing unit (120) into multiple resolutions. Hereinafter, the specific configuration of the unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention will be described in detail with reference to the attached drawings.

[0096]

[0097] FIG. 2 is a diagram illustrating the processing function configuration of the first image matching processing unit of an unstructured multi-medical image matching algorithm processing system according to an embodiment of the present invention, FIG. 3 is a diagram illustrating the processing function configuration of the second image matching processing unit of an unstructured multi-medical image matching algorithm processing system according to an embodiment of the present invention, and FIG. 4 is a diagram showing experimental results obtained by applying the unstructured multi-medical image matching algorithm processing system according to an embodiment of the present invention.

[0098]

[0099] The first image registration processing unit (110) is configured to perform image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator. As illustrated in FIG. 2, this first image registration processing unit (110) performs image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator, and can achieve registration by comparing the angle between the reference image and the DRR image. Here, the first image registration processing unit (110) registers by comparing the angle between the reference image and the DRR image, and the smaller the angle between the gradient vectors, the more similar the 2D reference image and the DRR image can become. At this time, the DICOM sequence obtained from the CT scan of the human brain model is used as a 3D plotted image in the registration experiment, and the projection image (DRR) of a specific CT parameter is used as a 2D reference image to simulate the actual X-ray image.

[0100]

[0101] Additionally, the first image matching processing unit (110) can function to make the matching result between the 2D reference image and the 3D floating image more accurate by comparing the angle between the reference image and the DRR image to match, and making the 2D reference image and the DRR image more similar as the angle between the gradient vectors becomes smaller. This first image matching processing unit (110) proposes a normalized cross-correlation based on the Sobel operator (NCCS) by combining the gradient vector angle and the normalized cross-correlation.

[0102]

[0103] The second image matching processing unit (120) is configured to perform image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator. This second image matching processing unit (120) performs image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, and can achieve matching by combining the normalized cross-correlation coefficient and image boundary information. Here, the second image matching processing unit (120) performs image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, and the LOG (Laplacian Of Gaussian) operator can extract image edge information.

[0104]

[0105] In addition, the second image matching processing unit (120) obtains a normalized cross-correlation (NCCL) based on a LOG operator, and the Laplacian image can be obtained by convolving the reference image and the DRR image with a LOG operator. This second image matching processing unit (120) proposes a method of combining the normalized cross-correlation coefficient and image boundary information to overcome the disadvantages of the normalized cross-correlation.

[0106]

[0107] The multi-resolution processing unit (130) is configured to process the respective matching images of the first image matching processing unit (110) and the second image matching processing unit (120) into multiple resolutions. This multi-resolution processing unit (130) processes the respective matching images of the first image matching processing unit (110) and the second image matching processing unit (120) into multiple resolutions, and may use Gaussian parameters for downsampling to obtain a low-resolution image from an original high-resolution image. Here, the multi-resolution processing unit (130) uses a multi-resolution strategy to achieve image matching and reduce the time consumed during matching.

[0108]

[0109] Additionally, the multi-resolution processing unit (130) can function to improve the signal-to-noise ratio by downsampling compared to wavelet transform. Additionally, the multi-resolution processing unit (130) can function to sample and store sampling information of each point signal, improve the matching speed, and increase the smoothing coefficient by using a Gaussian low-pass filter at the same level to smooth the image.

[0110]

[0111] The following describes experimental results applying an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention.

[0112]

[0113] 1) Experimental Procedure

[0114] Depending on the initial spatial transformation parameters, the CT image is transformed through the following process.

[0115] (1) A DRR image is generated by the projection of a CT image.

[0116] (2) To obtain a DRR image that is predominantly bone information, the projection threshold is set to 0.

[0117] (3) A similarity measurement between DRR and the reference image is calculated.

[0118] (4) Using the Powell-Brent optimization algorithm, the spatial transformation parameters are optimized up to the iteration stop condition after taking the similarity measure as the objective function.

[0119] (5) Stop matching and reach the output space conversion parameters.

[0120]

[0121] In the schematic diagram of the projection coordinate system, since the patient is located near the center of the X-ray irradiation when the X-ray image is taken, the center of the initial CT volume data can be set as the origin of the coordinate system, and the coordinate system can be established according to the orientation of the CT volume data. Additionally, the projection panel can be positioned on both sides of the CT volume data, and this projection coordinate system is similar to the actual imaging scene.

[0122]

[0123] 2) As an experimental result, the experiment is set up with three groups of experiments for original resolution image matching.

[0124]

[0125] The first row of Figure 4 shows the DRR image generated by CT projection after alignment, the second row shows the difference between the reference image and the aligned DRR image, the first column shows the normalized cross-correlation based on the Sobel operator, the second column shows the normalized cross-correlation based on the LOG operator, and the third column shows the original normalized cross-correlation.

[0126]

[0127] An intuitive comparison of these three experiments leads to the following conclusion. Given that the difference between images is greatest in the original normalized cross-correlation in the third column, it can be confirmed that the new normalized correlation is significantly improved compared to the original normalized cross-correlation. Furthermore, from the perspective of qualitative image analysis, the improved normalized cross-correlation based on the LOG operator is superior to the Sobel operator.

[0128]

[0129] In addition, statistical results show that when CT and DRR simulation X-ray images are aligned, the average values ​​of MTRE and MAE of the NCC based on the LOG operator are significantly improved compared to the original NCC, and the performance of the NCC based on the Sobel operator is improved compared to the original NCC in terms of the average values ​​of MTRE and MAE.

[0130]

[0131] In summary, introducing edge information through the LOG operator or angle information of the gradient vector through the Sobel operator can improve the matching accuracy and stability of the normalized cross-correlation matching indices, MAE and MTRE. Additionally, the edge information from the LOG operator can significantly improve normalized cross-correlation in terms of matching accuracy. However, the angle information of the gradient vector introduced by the Sobel operator can improve matching accuracy and extend the convergence range of the algorithm by increasing the sensitivity of the NCC measurement to rotational transformations.

[0132]

[0133] Figure 4 is a diagram showing experimental results, where (a) is the DRR based on NCCS registration, (b) is the DRR based on NCCL registration, and (c) is the DRR after registration based on NCC. Additionally, (d) is the difference map after registration based on NCCS, (e) is the difference map after registration based on NCCL, and (f) is the difference map after registration based on NCC.

[0134]

[0135] FIG. 5 is a diagram illustrating the flow of a method for using an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention. As shown in FIG. 5, the method for using an unstructured multi-medical image registration algorithm processing system according to an embodiment of the present invention may be implemented by including the step (S110) in which a first image registration processing unit performs image registration using a normalized cross-correlation process based on a Sobel operator, the step (S120) in which a second image registration processing unit performs image registration using a normalized cross-correlation process based on a logarithmic operator, and the step (S130) in which a multi-resolution processing unit processes the respective registration images of the first image registration processing unit and the second image registration processing unit into multiple resolutions.

[0136]

[0137] In step S110, the first image registration processing unit (110) performs image registration using a normalized cross-correlation (NCCS) process based on the Sobel operator. As illustrated in FIG. 2, the first image registration processing unit (110) in step S110 performs image registration using a normalized cross-correlation (NCCS) process based on the Sobel operator, and can achieve registration by comparing the angle between the reference image and the DRR image. Here, the first image registration processing unit (110) registers by comparing the angle between the reference image and the DRR image, and the smaller the angle between the gradient vectors, the more similar the 2D reference image and the DRR image can become. At this time, the DICOM sequence obtained from the CT scan of the human brain model is used as a 3D plotted image in the registration experiment, and the projection image (DRR) of a specific CT parameter is used as a 2D reference image to simulate the actual X-ray image.

[0138]

[0139] Additionally, the first image matching processing unit (110) can function to make the matching result between the 2D reference image and the 3D floating image more accurate by comparing the angle between the reference image and the DRR image to match, and making the 2D reference image and the DRR image more similar as the angle between the gradient vectors becomes smaller. This first image matching processing unit (110) proposes a normalized cross-correlation based on the Sobel operator (NCCS) by combining the gradient vector angle and the normalized cross-correlation.

[0140]

[0141] In step S120, the second image matching processing unit (120) performs image matching using a normalized cross-correlation (NCCL) process based on a log (LOG) operator. The second image matching processing unit (120) in step S120 performs image matching using a normalized cross-correlation (NCCL) process based on a log (LOG) operator, and can achieve matching by combining the normalized cross-correlation coefficient and image boundary information. Here, the second image matching processing unit (120) performs image matching using a normalized cross-correlation (NCCL) process based on a log (LOG) operator, and the LOG (Laplacian Of Gaussian) operator can extract image edge information.

[0142]

[0143] In addition, the second image matching processing unit (120) obtains a normalized cross-correlation (NCCL) based on a LOG operator, and the Laplacian image can be obtained by convolving the reference image and the DRR image with a LOG operator. This second image matching processing unit (120) proposes a method of combining the normalized cross-correlation coefficient and image boundary information to overcome the disadvantages of the normalized cross-correlation.

[0144]

[0145] In step S130, the multi-resolution processing unit (130) processes the respective matching images of the first image matching processing unit (110) and the second image matching processing unit (120) into multiple resolutions. In this step S130, the multi-resolution processing unit (130) processes the respective matching images of the first image matching processing unit (110) and the second image matching processing unit (120) into multiple resolutions, and may use Gaussian parameters for downsampling to obtain a low-resolution image from an original high-resolution image. Here, the multi-resolution processing unit (130) uses a multi-resolution strategy to achieve image matching and reduce the time consumed for matching.

[0146]

[0147] Additionally, the multi-resolution processing unit (130) can function to improve the signal-to-noise ratio by downsampling compared to wavelet transform. Additionally, the multi-resolution processing unit (130) can function to sample and store sampling information of each point signal, improve the matching speed, and increase the smoothing coefficient by using a Gaussian low-pass filter at the same level to smooth the image.

[0148]

[0149] As described above, the system for processing an unstructured multi-medical image registration algorithm and the method for using the same according to an embodiment of the present invention comprises a first image registration processing unit that performs image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator, a second image registration processing unit that performs image registration using a normalized cross-correlation (NCCL) process based on a logarithmic operator, and a multi-resolution processing unit that processes the registered images of the first image registration processing unit and the second image registration processing unit at multiple resolutions. By such a configuration, the accuracy of unstructured image registration can be improved using an unstructured multi-medical image registration algorithm that utilizes normalized cross-correlation based on a Sobel operator or normalized cross-correlation based on a logarithmic operator. In particular, by improving the accuracy of unstructured image registration using an unstructured multi-medical image registration algorithm that utilizes normalized cross-correlation based on a Sobel operator or normalized cross-correlation based on a logarithmic operator, the performance of the unstructured multi-medical image registration algorithm is improved. Consequently, along with the improvement in matching accuracy, it becomes possible to extend the convergence range of the algorithm.

[0150]

[0151] The foregoing description of the present invention is for illustrative purposes only, and those skilled in the art will understand that other specific forms can be easily modified without altering the technical spirit or essential features of the present invention. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. For example, each component described as a single unit may be implemented in a distributed manner, and components described as distributed may likewise be implemented in a combined form.

[0152]

[0153] The scope of the present invention is defined by the claims set forth below rather than by the detailed description above, and all modifications or variations derived from the meaning and scope of the claims and equivalent concepts thereof should be interpreted as being included within the scope of the present invention.

Claims

1. As an unstructured multi-medical image registration algorithm processing system (100), A first image matching processing unit (110) that performs image matching using a normalized cross-correlation (NCCS) process based on the Sobel operator; A second image matching processing unit (120) that performs image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator; and An unstructured multi-medical image matching algorithm processing system characterized by including a multi-resolution processing unit (130) that processes the matching images of the first image matching processing unit (110) and the second image matching processing unit (120) at multiple resolutions.

2. In paragraph 1, the first image matching processing unit (110) is, An unstructured multi-medical image registration algorithm processing system characterized by performing image registration using a normalized cross-correlation (NCCS) process based on the Sobel operator, and achieving registration by comparing angles between a reference image and a DRR image.

3. In paragraph 2, the first image matching processing unit (110) is, An unstructured multi-medical image registration algorithm processing system characterized by regulating the angle between a reference image and a DRR image by comparing the angles, wherein the smaller the angle between the gradient vectors, the more similar the 2D reference image and the DRR image become.

4. In paragraph 3, the first image matching processing unit (110) is, An unstructured multi-medical image registration algorithm processing system characterized by comparing the angle between a reference image and a DRR image to register them, and functioning to make the registration result between a 2D reference image and a 3D plotted image more accurate as the angle between the gradient vectors becomes smaller, thereby making the 2D reference image and the DRR image more similar.

5. In paragraph 1, the second image matching processing unit (120) is, An unstructured multi-medical image registration algorithm processing system characterized by performing image registration using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, and achieving registration by combining normalized cross-correlation coefficients and image boundary information.

6. In paragraph 5, the second image matching processing unit (120) is, An unstructured multi-medical image registration algorithm processing system characterized by performing image registration using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, wherein the LOG (Laplacian Of Gaussian) operator extracts image edge information.

7. In paragraph 6, the second image matching processing unit (120) is, An unstructured multi-medical image registration algorithm processing system characterized by obtaining a normalized cross-correlation (NCCL) based on a LOG operator and obtaining a Laplacian image by convolving a reference image and a DRR image with a LOG operator.

8. In any one of paragraphs 1 to 7, the multi-resolution processing unit (130) comprises, An unstructured multi-medical image matching algorithm processing system characterized by processing the matching images of the first image matching processing unit (110) and the second image matching processing unit (120) at multiple resolutions, and using Gaussian parameters for downsampling to obtain a low-resolution image from an original high-resolution image.

9. A method of using an unstructured multi-medical image registration algorithm processing system (100), (1) A first image matching processing unit (110) performs image matching using a normalized cross-correlation (NCCS) process based on a Sobel operator; (2) A step in which the second image matching processing unit (120) performs image matching using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator; and (3) A method of using an unstructured multi-medical image matching algorithm processing system, characterized in that the multi-resolution processing unit (130) includes the step of processing each of the matching images of the first image matching processing unit (110) and the second image matching processing unit (120) into multiple resolutions.

10. In claim 9, the first image matching processing unit (110) is, A method for using an unstructured multi-medical image registration algorithm processing system, characterized by performing image registration using a normalized cross-correlation (NCCS) process based on a Sobel operator, and achieving registration by comparing angles between a reference image and a DRR image.

11. In item 10, the first image matching processing unit (110) is, A method for using an unstructured multi-medical image registration algorithm processing system, characterized by comparing the angle between a reference image and a DRR image to register them, wherein the smaller the angle between the gradient vectors, the more similar the 2D reference image and the DRR image become.

12. In paragraph 11, the first image matching processing unit (110) is, A method for using an unstructured multi-medical image registration algorithm processing system, characterized by comparing the angle between a reference image and a DRR image to register them, and functioning to make the registration result between a 2D reference image and a 3D plotted image more accurate as the angle between the gradient vectors becomes smaller, thereby making the 2D reference image and the DRR image more similar.

13. In claim 9, the second image matching processing unit (120) is, A method for using an unstructured multi-medical image registration algorithm processing system, characterized by performing image registration using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, and achieving registration by combining normalized cross-correlation coefficients and image boundary information.

14. In paragraph 13, the second image matching processing unit (120) is, A method of using an unstructured multi-medical image registration algorithm processing system, characterized by performing image registration using a normalized cross-correlation (NCCL) process based on a logarithmic (LOG) operator, wherein the LOG (Laplacian Of Gaussian) operator extracts image edge information.

15. In paragraph 14, the second image matching processing unit (120) is, A method of using an unstructured multi-medical image registration algorithm processing system, characterized in that a normalized cross-correlation (NCCL) based on a LOG operator is obtained, and a Laplacian image is obtained by convolving a reference image and a DRR image with a LOG operator.

16. In any one of claims 9 to 15, the multi-resolution processing unit (130) is, A method of using an unstructured multi-medical image matching algorithm processing system, characterized by processing the matching images of the first image matching processing unit (110) and the second image matching processing unit (120) at multiple resolutions, and using Gaussian parameters for downsampling to obtain a low-resolution image from an original high-resolution image.

Images