Improved versatility of AI-based medical image conversion method and device

Abstract

The present invention relates to a medical image conversion method using artificial intelligence, and more particularly to an image conversion method and apparatus for converting medical images using a generative adversarial network (GAN) with improved versatility. According to one embodiment of the present invention, a medical image conversion system with improved model versatility can be provided by reflecting diverse evaluation opinions of expert groups belonging to various institutions so that a learning model can provide consistent performance for tasks required by multiple institutions.

Description

[Technical Field]

[0001] One aspect of the present invention relates to a method for converting medical images using artificial intelligence, and more particularly, to an image conversion method and apparatus for converting medical images using a generative adversarial network (GAN) with improved versatility. [Background technology]

[0002] The material described in this section merely provides background information for embodiments of the present invention and may not constitute prior art.

[0003] In a medical image conversion system that reflects a learning model that performs machine learning, it is important to collect the learning data that serves as the basis for that learning.

[0004] Training data based on medical data is generally collected from hospitals and institutions, but such training data is very difficult to access due to security requirements for personal information.

[0005] Therefore, in most cases, only information from specific affiliated hospitals or institutions is secured and used as learning data.

[0006] However, learning models trained using training data processed using medical information provided only by specific hospitals or institutions are inevitably biased. In other words, AI models that are dependent on specific institutions are bound to provide inaccurate outputs for input images provided by other hospitals or institutions.

[0007] Therefore, there is an urgent need to develop medical image conversion software using an AI learning model that has the versatility to be applicable to various hospitals or institutions.

[0008] The above-mentioned background art is technical information that the inventor possessed for the purpose of deriving the embodiments of the present invention or that he acquired in the process of deriving the embodiments of the present invention, and therefore cannot necessarily be said to be publicly known art that was disclosed to the general public prior to the filing of the embodiments of the present invention. Summary of the Invention [Problem to be solved by the invention]

[0009] One aspect of the present invention has been proposed to solve the above-mentioned problems, and an object of the present invention is to provide a medical image conversion system with improved model versatility by reflecting diverse evaluation opinions of expert groups belonging to various institutions so that the learning model can provide consistent performance for tasks required by multiple institutions.

[0010] Another object of the present invention is to provide a medical image conversion system that enables medical staff to perform medical examinations, diagnoses, and treatments with accurate information through conversion between medical images, thereby enabling patients to receive appropriate medical services.

[0011] The technical problems that the present invention aims to solve are not limited to those mentioned above, and other technical problems not mentioned above will be clearly understood by those skilled in the art to which the present invention pertains from the following description. [Means for solving the problem]

[0012] In order to achieve the above-mentioned object, one aspect of the present invention provides an artificial intelligence learning method for improving the versatility of a learning model that is trained using a generative adversarial network (GAN), receives input data for medical image conversion, generates an output image, and is trained to improve accuracy of the medical image conversion result by comparing the output image with a target image, comprising:

[0013] The first stage involves providing the medical image conversion results to a group of assessors;

[0014] a second step of receiving an evaluation result corresponding to the medical image conversion result from the group of evaluators;

[0015] A third step of analyzing the evaluation tendency difference of the evaluator group based on the evaluation results; and

[0016] a fourth step of removing the evaluation propensity difference from the evaluation result to create a propensity-free evaluation result, and reflecting the difference-removed evaluation result in a learning model so that the learning model performs learning.

[0017] Here, the third step may include a step of analyzing a first evaluation tendency difference between different affiliated institutions and a step of analyzing a second evaluation tendency difference between different evaluators.

[0018] The different evaluators may belong to the same institution.

[0019] The third step may be characterized by analyzing the evaluation tendency difference using a statistical method. In some embodiments, the third step may be characterized by analyzing the evaluation tendency difference collected for a certain period of time or a certain number of times.

[0020] Another aspect of the present invention is a medical image conversion device that is trained using a generative adversarial network (GAN) and receives input data for medical image conversion and generates an output image, comprising:

[0021] an evaluator information providing unit that provides the medical image conversion result to the evaluator group;

[0022] an evaluator information receiving unit that receives an evaluation result corresponding to the medical image conversion result from the evaluator group;

[0023] a difference analysis unit that analyzes the difference in evaluation tendency of the evaluator group based on the evaluation results; and

[0024] A medical image conversion device may be provided, including a learning unit that removes the evaluation propensity difference from the evaluation result to create a propensity-free evaluation result, and reflects the difference-removed evaluation result in a learning model so that the learning model performs learning.

[0025] The difference analysis unit may include a first difference analysis unit that analyzes a first evaluation tendency difference between different affiliated institutions and a second difference analysis unit that analyzes a second evaluation tendency difference between different evaluators. In some embodiments, the different evaluators may belong to the same affiliated institution.

[0026] The difference analysis unit may analyze the evaluation tendency difference as a statistical method.

[0027] The difference analysis unit may analyze the evaluation tendency difference collected for a certain period of time or a certain number of times.

[0028] Another aspect of the present invention is a method for improving a conversion result of a learning model that performs medical image conversion using a generative adversarial network (GAN), comprising:

[0029] providing the medical image conversion result to a group of evaluators and receiving evaluation opinions on the medical image conversion result from the group of evaluators;

[0030] A difference extraction step of extracting a difference in evaluation tendency of the evaluator group based on the evaluation opinions; and

[0031] an opinion preparation step of removing the extracted evaluation tendency difference from the evaluation opinion to prepare a non-biased evaluation opinion from which the evaluation tendency difference has been removed;

[0032] It is possible to provide a method for improving the conversion result of a learning model including the above.

[0033] Another aspect of the present invention is a learning model that receives an input image and generates an output image having a domain different from the input image based on the input image;

[0034] a first feedback unit that compares the output image with a target image and feeds back the comparison result to the learning model; and

[0035] a second feedback unit that provides the output video to an evaluator, receives evaluation opinions from the evaluators, extracts evaluation tendency differences from the evaluators, removes the evaluation tendency differences from the evaluation opinions to derive unbiased evaluation opinions from which the evaluation tendency differences have been removed, and feeds back the unbiased evaluation opinions to the learning model;

[0036] A medical image conversion system including the above can be provided. [Effects of the Invention]

[0037] As described above, according to one embodiment of the present invention, a medical image conversion system with improved model versatility can be provided by reflecting diverse evaluation opinions from expert groups belonging to various institutions so that the learning model can provide consistent performance for tasks required by multiple institutions.

[0038] According to another embodiment of the present invention, there is provided a medical image conversion system that enables medical staff to perform medical examinations, diagnoses, and treatments with accurate information through conversion between medical images, thereby enabling patients to receive appropriate medical services.

[0039] The present invention has an effect of providing a medical image conversion method that enables medical staff to perform medical examinations, diagnoses, and treatments with accurate information through conversion between medical images, thereby enabling patients to receive appropriate medical services.

[0040] In addition, the present invention has various effects such as excellent versatility depending on the embodiment, and such effects can be clearly confirmed in the description of the embodiments below. [Brief explanation of the drawings]

[0041] The following drawings attached to this specification illustrate one embodiment of the present invention and, together with the detailed description of the invention described above, serve to further understand the technical concept of the present invention. Therefore, the present invention should not be interpreted as being limited to the matters shown in such drawings.

[0042] [Figure 1] FIG. 1 illustrates a medical image conversion method according to one embodiment of the present invention. [Figure 2] FIG. 2 illustrates an artificial intelligence learning model according to one embodiment of the present invention. [Figure 3] FIG. 3 shows an embodiment of calculating the match rate between the output image and the target image. [Figure 4] FIG. 4 illustrates an artificial intelligence learning method according to one embodiment of the present invention. [Figure 5] FIG. 5 shows a medical image conversion device according to another embodiment of the present invention. [Figure 6] FIG. 6 illustrates a training data preprocessing method according to an embodiment of the present invention. [Figure 7] FIG. 7 illustrates a medical image conversion system according to one embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0043] The advantages and features of the present invention, and methods for achieving them, will become clearer with reference to the following detailed description of the embodiments accompanied by the accompanying drawings. However, the present invention is not limited to the following embodiments, and may be embodied in various different forms, and includes all modifications, equivalents, and alternatives within the spirit and technical scope of the present invention. The following embodiments are provided to ensure a complete disclosure of the present invention and to fully convey the scope of the invention to those skilled in the art. In describing the present invention, if a detailed description of related prior art is deemed to obscure the gist of the present invention, such a detailed description will be omitted.

[0044] The terms used in this application are merely used to describe specific embodiments and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly dictates otherwise.

[0045] In this application, the terms "comprise" or "have" are intended to specify the presence of a specified feature, number, step, operation, component, part, or combination thereof, and should be understood as not precluding the presence or additional possibility of one or more other features, numbers, steps, operations, components, parts, or combinations thereof. Terms such as "first," "second," etc. may be used to describe various components, but the components are not limited by these terms. These terms are used only to distinguish one component from another.

[0046] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description with reference to the accompanying drawings, identical or corresponding components will be given the same drawing numbers, and duplicate descriptions thereof will be omitted.

[0047] FIG. 1 illustrates a medical image conversion method according to an embodiment of the present invention, and FIG. 2 illustrates an artificial intelligence learning model according to an embodiment of the present invention.

[0048] According to one embodiment of the present invention, there is provided a medical image conversion method for converting medical images using a generative adversarial network (GAN).

[0049] The medical image conversion method according to this embodiment includes a first step (S100) of receiving an input of a first image 110 selected from a paired data set including a first image 110 and a second image 120, and a second step (S110) of generating a third image 130 corresponding to the first image 110 and having a different domain from the first image 110 based on the first image 110.

[0050] The method may include a third step (S120) of comparing the third image 130 with the second image 120 and training a learning model taking into account the comparison result, and a fourth step (S130) of comparing the third image 130 with the first image 110 and training the learning model taking into account the comparison result.

[0051] The first image 110, the second image 120, and the third image 130 may represent medical images. Here, the medical images may include magnetic resonance (MR) images, such as 2D MR, 3D MR, 2D streaming MR, 4D MR, 4D volumetric MR, and 4D cine MR; functional MR images, such as fMR, DCE-MR, and diffusion MR; computed tomography (CT) images, such as 2D CT, conebeam CT, 3D CT, and 4D CT; ultrasound images, such as 2D ultrasound, 3D ultrasound, and 4D ultrasound; positron emission tomography (PET) images; X-ray images; fluoroscopic images; radiotherapy portal images; single-photon emission computed tomography (SPECT) images; and computer-generated synthetic images, such as pseudo-CT.

[0052] In addition, the medical image may include medical image data, such as a training image, a ground truth image, a contoured image, a dose image, and the like.

[0053] Depending on the embodiment, the medical image may include an image acquired from an image acquisition device, a computer-generated image, etc. The image acquisition device may include, for example, an MR imaging device, a CT imaging device, a PET imaging device, an ultrasound imaging device, a fluoroscopic device, a SPECT imaging device, an integrated linear accelerator and MR imaging device, etc. In addition, the image acquisition device is not limited to the above examples and may include various medical imaging devices within the scope of the technical concept for acquiring a medical image of a patient.

[0054] The paired data set may include a set of training data including images of an object, for example, a specific part of a patient.

[0055] For example, a CT imaging device and an MR imaging device can be used to take CT images and MR images of a specific part of a patient, such as the head, respectively, and in this case, the CT images and MR images can form a paired data set.

[0056] When non-enhanced CT images and MR images are taken of a specific area of ​​a patient, the non-enhanced CT images and MR images can form a paired data set, and when contrast-enhanced CT images and MR images are taken of a specific area of ​​a patient, the contrast-enhanced CT images and MR images can form a paired data set.

[0057] When T1 and T2 images are taken of a specific part of a patient, the T1 and T2 images can form a paired data set.

[0058] Here, the specific part of a patient may refer to a specific part of the same patient, or may refer to a specific part of different patients if the parts being imaged are the same.

[0059] For example, CT images and MR images of the same specific area of ​​a patient can form a paired data set, and CT images and MR images of the same area of ​​different patients can also form a paired data set.

[0060] What does it mean for two images to have different domains? For example, if the pattern contours contained in the two corresponding images differ between the two images, or if the intensity distribution or pattern between the two images is different, then the domains are different.

[0061] In addition, images taken with different imaging devices have different domains. For example, CT images and MR images are taken with CT imaging devices and MR imaging devices, respectively, so they have different domains. Similarly, there are different domains between ultrasound images and CT images, between ultrasound images and MR images, and between MR images and PET images.

[0062] Furthermore, non-contrast-enhanced images and contrast-enhanced images have different intensity distributions for the same region, and therefore have different domains.

[0063] In the case of MR images, T1 images and T2 images are the same MR images, but they have different intensity distributions, so they have different domains. In other words, if there is a difference in the signal, sequence, or image creation process that is the basis for image creation between two images created through reconstruction based on signals provided by an individual, the two images with the difference have different domains. In addition, the domains are different between the examples of medical images mentioned above.

[0064] To explain a medical image conversion method according to an embodiment of the present invention, an example will be described in which an image acquisition device acquires a first image 110 and a second image 120 of the same part of the same patient and obtains paired data sets. In this embodiment, the first image 110 is an original CT image, the second image 120 is an original MR image, and the third image 130 is a combined MR image.

[0065] The first step (S100) may include a step of receiving, as input to a learning model, a first image 110 selected from a paired dataset including a first image 110 and a second image 120. Here, the first image 110 may include an input image obtained by processing an original CT image using learning data.

[0066] The manufacturer of the software medical device that performs the medical image conversion method according to this embodiment can determine which of the first image 110 and the second image 120 to select as the input image in the paired data set including the first image 110 and the second image 120. In this case, the image to be selected can be automatically selected by a selection algorithm or manually selected by an operator.

[0067] A manufacturer can create a paired dataset by acquiring original CT images using a CT imaging device and original MR images using an MR imaging device, or can select the original CT images as input images in the paired dataset using an algorithm or manually.

[0068] The second step (S110) may include a step in which a learning model that receives an input of a first image 110 generates a third image 130 based on the first image 110, corresponding to the first image 110 and having a different domain from the first image 110.

[0069] For example, the learning model can receive an original CT image as input and generate a synthetic MR image based on the original CT image.

[0070] The process by which the learning model generates the third image 130 from the first image 110, which has a different domain from the first image 110, can be performed by the following two processes according to an embodiment. However, the process by which the learning model generates the third image 130 is not limited to the following two processes.

[0071] The first process is composed of one or more layers that receive a first image 110 acquired by an image acquisition device and perform convolution operations to extract high-dimensional features smaller than the first image 110, and this can be expressed as an artificial neural network that extracts features. Such an artificial neural network that extracts features receives input of 2D or 3D actual image data acquired by an image acquisition device and performs convolution operations to extract image features from the original image.

[0072] To exemplarily explain this process, a convolution layer extracts image features through a filter and a pooling layer strengthens the features and reduces the size of the image. Image features can be extracted by repeating convolution and pooling.

[0073] In a convolution layer configured with multiple layers according to an embodiment, the first convolution layer may extract features of an input image provided as an input and output a feature map as a result, the pooling layer may receive the feature map as an input and output a result in which the features are enhanced and the image size is reduced, and the output of the pooling layer may again be input to the second convolution layer.

[0074] The second step is composed of one or more layers that perform deconvolution operations to generate image data of different modalities from the result of the first step, and can be expressed as an artificial neural network that generates images of different modalities. Such an artificial neural network that generates images of different modalities can generate two-dimensional or three-dimensional image data of different modalities by performing deconvolution operations from the results of the convolution operation performed by the artificial neural network that extracts features in the first step.

[0075] The third step (S120) may include comparing the third image 130 with the second image 120 and training a learning model based on the comparison results. If the original CT image is determined as the input image from among the original CT image and the original MR image included in the paired dataset, the original MR image may be determined as the target image. Here, the second image 120 may correspond to ground truth. After generating the third image 130, the learning model may compare the generated third image 130 with the second image 120 and perform training to reduce the difference between the two. That is, the learning model may continuously compare the synthetic MR image with the original MR image to determine whether the synthetic MR image has been generated to be close to the original MR image, thereby training the synthetic MR image to be the same as the original MR image. According to an embodiment, the method may include applying a loss function to the algorithm to calculate a loss value between the third image 130 and the second image 120, and updating the parameters of the learning model based on the loss value.

[0076] According to an embodiment, in the step of calculating the loss value, the calculation of the loss value for updating the learning model may be performed for each iteration. That is, the learning model may calculate the loss value between the synthetic MR image and the original MR image for each iteration while performing learning.

[0077] When the number of repetitions includes a first repetition interval and a second repetition interval that are connected sequentially, depending on the embodiment, if the absolute value of the change between the first loss value calculated in the first repetition interval and the second loss value calculated in the second repetition interval is smaller than a reference value, the learning model may regard this as overfitting and terminate learning.

[0078] Once the learning model has sufficiently learned medical image conversion, it can complete the learning without further training. The learning model that has completed the learning in this way will now have the ability to generate medical images from third images 130 that have a different domain from the first images 110 for various first images 110 each having a different data distribution.

[0079] According to an embodiment of the present invention, the learning model 200 may be a GAN model and may include a generator and a classifier. The generator may perform image generation learning, and the classifier may perform image segmentation learning.

[0080] Referring to FIG. 2, in a learning model 200 to which a GAN model is applied, when an original CT image 110 is input as an input image to a generator 210, the generator 210 performs learning to generate a synthetic MR image 130 through the above-mentioned convolution operation, and a classifier 220 can perform learning to distinguish between the original MR image 120, i.e., the actual medical image, and the synthetic MR image 130.

[0081] According to an embodiment, a cycle-GAN model (not shown) may be applied to the learning model 200 according to this embodiment. In this case, the learning model 200 may be configured to include a first generator 210, a first division molecule 220, a second generator, and a second division molecule. When the original CT image 110 is input as an input image to the first generator 210, the first generator 210 performs learning to generate a synthetic MR image 130 through the convolution operation described above, and the first division molecule 220 may perform learning to distinguish between the original MR image 120, i.e., the actual medical image, and the synthetic MR image. Furthermore, when the synthetic MR image 130 generated by the first generator 210 is input to the second generator, the second generator may perform learning to generate a synthetic CT image based on the synthetic MR image 130, and the second division molecule may perform learning to distinguish the synthetic CT image from the original CT image.

[0082] Here, the original CT image 110 corresponds to the first image 110 in the above-mentioned medical image conversion method, the synthesized MR image 130 corresponds to the third image 130, and the original MR image 120 corresponds to the second image 120.

[0083] According to an embodiment, the third step (S120) may include calculating a match rate between the third image 130 and the second image 120 and feeding the calculated value back to the learning model.

[0084] FIG. 3 shows an embodiment of calculating the match rate between the output image and the target image.

[0085] In some embodiments, if a lesion area is displayed in the same location on the composite MR image and the original MR image, the matching rate can be calculated by comparing the size of the corresponding lesion area. For example, the positions of tumors and other lesions can be displayed on the composite MR image and the original MR image, respectively, and then the matching rate can be calculated by switching between binary images.

[0086] In this case, the match rate can be calculated using evaluation indices such as the Hausdorff distance or the Dice similarity coefficient, or can be quantitatively calculated based on the difference in the centers of mass of the two lesions. If the target and output images are CT images, the match rate between the two images can be determined by comparing the calculated radiation doses for the lesions. Depending on the embodiment, the difference between the synthesized and original MR images can be quantified using MAE, RMSE, SSIM, PSNR, etc. (hereinafter referred to as performance indices) for determination.

[0087] Referring to Figure 3, the process of aligning the lesion area between the synthesized MR image 130, which is the output image, and the original MR image 120, which is the target image, and then applying the Hausdorff distance measurement method to compare the contours of each lesion and calculate the matching rate is shown.

[0088] FIG. 4 illustrates an artificial intelligence learning method according to one embodiment of the present invention.

[0089] According to an embodiment of the present invention, there is provided an artificial intelligence learning method for improving the versatility of a learning model 200, which is trained using a generative adversarial network (GAN), receives input data for medical image conversion, generates an output image, and is trained to improve accuracy of the medical image conversion result by comparing the output image with a target image.

[0090] A first step (S200) of providing a medical image conversion result to a group of evaluators; and a second step (S210) of receiving an evaluation result corresponding to the medical image conversion result from the group of evaluators.

[0091] The third step (S220) analyzes the difference in evaluation tendency of the evaluator group based on the evaluation results, and

[0092] It may include a fourth step (S230) of removing the evaluation tendency difference from the evaluation result to create a non-tendency evaluation result, and reflecting the non-tendency evaluation result in the learning model 200 so that the learning model 200 performs learning.

[0093] Here, the evaluator may generally include a specialist, and in some embodiments, the evaluator may be a medical professional such as a doctor or nurse.

[0094] The evaluator group may include a group of experts who evaluate the medical image conversion results, or may refer to experts registered on a manageable platform, etc. Here, the platform may include a website or app that performs a function of providing the medical image conversion results to evaluators registered on the platform and receiving opinions on the conversion results from the evaluators.

[0095] In the second step (S210), the medical image conversion result may include an output image output by the learning model 200. The learning model 200 performs machine learning and, after the learning is completed, is provided to each hospital as a software medical device and can function as an auxiliary program for medical procedures at each hospital.

[0096] The evaluation results of the evaluator group may include evaluation results on the resolution, quality, etc. of the output image output by the learning model 200. The quality of the medical image may be generally given a score.

[0097] According to an embodiment, the third step (S220) may include a step of analyzing a first evaluation tendency difference between different affiliated institutions and a step of analyzing a second evaluation tendency difference between different evaluators, where the affiliated institutions may include institutions such as hospitals and the Health Insurance Review and Assessment Service.

[0098] The first evaluation tendency difference may include, for example, the perception of medical image conversion software between different affiliated institutions, i.e., Hospital A and Hospital B.

[0099] When there is Doctor I at Hospital A and Doctor R at Hospital B, when they evaluate the same image produced by the image conversion device, Hospital A's perception of the medical image conversion software program may differ from Hospital B's. Hospital A may have a good opinion of the medical image conversion software according to this embodiment and give it a high score, while Hospital B may have a bad reputation for the medical image conversion software according to this embodiment and give it an especially low score. This difference may be the first evaluation tendency difference.

[0100] The second evaluation tendency difference indicates the difference in evaluation tendency between different evaluators.

[0101] Depending on the embodiment, different evaluators may belong to the same institution. For example, Doctor I and Doctor R may both be doctors at Hospital A. In this case, even if the atmosphere and policies of Hospital A to which Doctor I and Doctor R belong are the same, their evaluations may differ depending on their personal tendencies. In other words, Doctor I may receive a relatively high evaluation, while Doctor R may receive a relatively low evaluation.

[0102] According to an embodiment, the third step (S220) may analyze the evaluation tendency difference using a statistical method. According to an embodiment, the third step (S220) may analyze the evaluation tendency difference collected for a certain period of time or a certain number of times.

[0103] Here, because the difference in evaluation tendency includes subjective opinions, it is preferable to analyze it statistically. That is, by analyzing a large number of evaluation results, it is possible to grasp the difference in evaluation tendency among doctors. In some embodiments, a certain period of time is set and the difference in evaluation tendency can be analyzed based on the evaluation results submitted during that period, or in other embodiments, the difference in evaluation tendency can be analyzed based on evaluation results collected a certain number of times.

[0104] The second evaluation tendency difference can be analyzed based on the identification information of the individual evaluator, and after collecting scores on the video quality for a certain period of time, the evaluation tendency difference can be extracted and analyzed, and the evaluation tendency difference derived from this analysis can be reflected as a score removed from the evaluation result and fed back to the artificial intelligence learning model 200.

[0105] The AI ​​learning model 200 can improve the accuracy of the output image by receiving feedback on the evaluator's evaluation results and reflecting them in its learning. However, if the evaluation results are subjective or narrow-minded, it will not be useful in improving the accuracy of the output image. Medical image conversion software trained based on subjective evaluation results provided by a specific hospital is not suitable for use by all hospitals in general. Therefore, it is preferable to remove evaluation tendencies from the evaluation results to prevent subjective or narrow-minded evaluation results from being reflected, and to feed back to the learning model 200 based on unbiased evaluation results having objective content.

[0106] FIG. 5 shows a medical image conversion device according to another embodiment of the present invention.

[0107] According to another embodiment of the present invention, a medical image conversion device 400 is trained using a generative adversarial network (GAN), receives input data for medical image conversion, and generates an output image.

[0108] an evaluator information providing unit 410 that provides the medical image conversion result to the evaluator group;

[0109] an evaluator information receiving unit 420 that receives an evaluation result corresponding to the medical image conversion result from the evaluator group;

[0110] a difference analysis unit 430 for analyzing the difference in evaluation tendency of the evaluator group based on the evaluation results; and

[0111] A medical image conversion device (400) may be provided that includes a learning unit (440) that removes the evaluation propensity difference from the evaluation result to create a non-propensity evaluation result, and reflects the non-propensity evaluation result in a learning model so that the learning model performs learning.

[0112] Here, the difference analysis unit 430 may include a first difference analysis unit that analyzes a first evaluation tendency difference between the different affiliated institutions and a second difference analysis unit that analyzes a second evaluation tendency difference between the different evaluators. In some embodiments, the different evaluators may belong to the same affiliated institution.

[0113] Depending on the embodiment, the difference analysis unit 430 may analyze the evaluation tendency difference using a statistical method, and depending on the embodiment, the difference analysis unit 430 may analyze the evaluation tendency difference collected for a certain period of time or a certain number of times.

[0114] The medical image conversion device 400 according to this embodiment may be substantially the same model as the model that performs the artificial intelligence learning method according to the previous embodiment.

[0115] FIG. 6 illustrates a training data preprocessing method according to an embodiment of the present invention.

[0116] According to another aspect of the present invention

[0117] A method for improving the conversion results of a learning model that performs medical image conversion trained using a generative adversarial network (GAN),

[0118] providing a medical image conversion result to a group of evaluators and receiving evaluation opinions on the medical image conversion result from the group of evaluators (S300);

[0119] A difference extraction step (S310) of extracting the difference in evaluation tendency of the evaluator group based on the evaluation opinions; and

[0120] an opinion preparation step (S320) of removing the extracted evaluation tendency difference from the evaluation opinion to prepare a non-biased evaluation opinion from which the evaluation tendency difference has been removed;

[0121] It is possible to provide a method for improving the conversion result of a learning model including the above.

[0122] FIG. 7 illustrates a medical image conversion system according to one embodiment of the present invention.

[0123] The medical image conversion system 500 according to an embodiment of the present invention includes a learning model 200 that receives a first image 110 and generates a third image 130 based on the first image 110, the third image 130 having a different domain from the first image 110;

[0124] a first feedback unit (300) that compares the third image (130) with the second image (120) and feeds back the comparison result to the learning model (200); and

[0125] The learning model 200 may further include a second feedback unit 310 that provides the third image 130 to an evaluator, receives the evaluator's evaluation opinion, extracts an evaluation tendency difference of the evaluator, removes the evaluation tendency difference from the evaluation opinion to derive a non-biased evaluation opinion from which the evaluation tendency difference has been removed, and feeds back the non-biased evaluation opinion to the learning model 200.

[0126] Here, the first image 110 displayed on the drawing corresponds to the input image in the above-described embodiment, the second image 130 corresponds to the target image, and the third image 120 corresponds to the output image.

[0127] The above-described embodiments of the present invention may be implemented in the form of a computer program that can be executed by various components on a computer, and such a computer program may be recorded on a computer-readable medium, which may include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program instructions, such as ROMs, RAMs, and flash memories.

[0128] Meanwhile, the computer program may be one specially designed and constructed for the present invention, or one that is known and available to those skilled in the art of computer software. Examples of the computer program include not only machine language code, such as that produced by a compiler, but also high-level language code that can be executed by a computer using an interpreter, etc.

[0129] In the present specification (particularly in the claims), the use of the term "said" and similar indicators may be applicable to both the singular and the plural. Furthermore, when a range is described in the present invention, it is considered to include inventions to which individual values ​​within the range are applied (unless otherwise specified), and is the same as if each individual value constituting the range were described in the detailed description of the invention.

[0130] Unless explicitly stated or stated to the contrary, steps constituting the method of the present invention may be performed in any suitable order. The present invention is not necessarily limited to the order of steps described above. The use of all examples or exemplary terms (e.g., etc.) in the present invention is merely for the purpose of describing the present invention in detail, and the scope of the present invention is not limited by the claims. Furthermore, those skilled in the art will recognize that various modifications, combinations, and variations can be made within the scope of the appended claims or their equivalents, depending on design conditions and factors.

[0131] Therefore, the concept of the present invention should not be limited to the above-described embodiments, and all scopes equivalent to or modified equivalently from the claims below, as well as the scope of the invention, can be said to fall within the scope of the concept of the present invention. [Explanation of symbols]

[0132] 110 First Image 120 Second Image 130 Third Image 200 Learning Models 300 First Feedback Section 310 Second Feedback Section 400 Medical Image Conversion Device 410 Evaluator Information Section 420 Evaluator Information Receiving Unit 430 Difference Analysis Department 440 Learning Department 500 Medical Image Conversion System

Claims

1. 1. An artificial intelligence learning method for improving the versatility of a learning model that is trained using a generative adversarial network (GAN), receives input data for medical image conversion, generates an output image, and compares the output image with a target image to improve accuracy of the medical image conversion result, A first step of providing the medical image conversion results to a group of evaluators; a second step of receiving an evaluation result corresponding to the medical image conversion result from the group of evaluators; A third step of analyzing the evaluation tendency difference of the evaluator group based on the evaluation results; and and a fourth step of removing the evaluation tendency difference from the evaluation result to generate a non-biased evaluation result, and reflecting the non-biased evaluation result in a learning model so that the learning model performs learning. An artificial intelligence learning method characterized by:

2. The third step includes a step of analyzing a first evaluation tendency difference between different affiliated institutions and a step of analyzing a second evaluation tendency difference between different evaluators. The artificial intelligence learning method according to claim 1 .

3. The different evaluators belong to the same institution. The artificial intelligence learning method according to claim 2.

4. The third step is to analyze the evaluation tendency difference using a statistical method. The artificial intelligence learning method according to claim 1 .

5. The third step is to analyze the evaluation tendency difference collected for a certain period or a certain number of times. The artificial intelligence learning method according to claim 4.

6. A medical image conversion device that is trained using a generative adversarial network (GAN) and receives input data for medical image conversion and generates an output image, an evaluator information providing unit that provides the medical image conversion result to the evaluator group; an evaluator information receiving unit that receives an evaluation result corresponding to the medical image conversion result from the evaluator group; a difference analysis unit that analyzes the difference in evaluation tendency of the evaluator group based on the evaluation results; and a learning unit that removes the evaluation tendency difference from the evaluation result to create a non-tendency evaluation result, and reflects the non-tendency evaluation result in a learning model so that the learning model performs learning. A medical image conversion device characterized by:

7. The difference analysis unit includes a first difference analysis unit for analyzing a first evaluation tendency difference between different affiliated institutions and a second difference analysis unit for analyzing a second evaluation tendency difference between different evaluators.

7. The medical image conversion device according to claim 6.

8. The different evaluators belong to the same institution. The medical image conversion device according to claim 7.

9. The difference analysis unit analyzes the evaluation tendency difference using a statistical method.

7. The medical image conversion device according to claim 6.

10. The difference analysis unit analyzes the evaluation tendency difference collected for a certain period or a certain number of times. The medical image conversion device according to claim 9.

11. A method for improving conversion results of a learning model that performs medical image conversion using a generative adversarial network (GAN), providing the medical image conversion result to a group of evaluators and receiving evaluation opinions on the medical image conversion result from the group of evaluators; A difference extraction step of extracting an evaluation tendency difference of the evaluator group based on the evaluation opinions; and and preparing an opinion that removes the extracted evaluation tendency difference from the evaluation opinion to prepare a non-biased evaluation opinion from which the evaluation tendency difference has been removed. A method for improving the conversion results of a learning model.

12. A learning model that receives an input image and generates an output image having a domain different from the input image based on the input image; a first feedback unit that compares the output image with a target image and feeds back the comparison result to the learning model; and a second feedback unit that provides the output video to an evaluator, receives evaluation opinions from the evaluators, extracts evaluation tendency differences from the evaluators, removes the evaluation tendency differences from the evaluation opinions to derive unbiased evaluation opinions from which the evaluation tendency differences have been removed, and feeds back the unbiased evaluation opinions to the learning model. A medical image conversion system characterized by:

Images