Automatic color conversion device for oral computed tomography images based on artificial intelligence and driving method thereof

Abstract

The present invention relates to an artificial intelligence-based automatic color conversion device for oral computed tomography (CT) scan images and a driving method thereof. According to an embodiment of the present invention, the artificial intelligence-based automatic color conversion device for oral CT scan images comprises: a storage unit for storing a user-prepared color conversion image related to alveolar bone density; and a control unit for, upon receiving an oral CT scan input image from a patient, converting the color of the oral CT scan input image based on preset parameter values, and automatically converting the color of a subsequently input oral CT scan input image by adjusting the preset parameter values ​​based on an error value generated between the converted oral CT scan input image and a (previously) stored color conversion image.

Description

Technical Field

[0001] The present invention relates to a device for automatic color conversion of oral computed tomography (CT) images based on artificial intelligence and a driving method thereof. More specifically, the present invention relates to an artificial intelligence-based automatic color conversion device for oral computed tomography (CT) images, wherein the device can automatically convert the color on a display according to the bone density of the alveolar bone (for example, when a dental implant is virtually placed in the alveolar bone) and a driving method of the device. Background Art

[0002] Dental implants originally refer to replacements used to restore lost human tissue. In the dental field, dental implants refer to a series of treatments to implant artificial teeth. To replace the lost tooth root (root), a root fixture made of a material that is not rejected by the human body, such as titanium, is implanted into the alveolar bone of the removed tooth. An artificial tooth is then fixed to restore the tooth's function. With general prostheses or dentures, nearby teeth or bones will be damaged over time, but implants do not harm nearby tooth tissue and do not cause tooth decay while having the same function or shape as natural teeth. Therefore, the semi-permanent use of implants is advantageous. While there are various treatment methods depending on the type of fixture, a general artificial tooth treatment (also known as dental implant or implant treatment) involves drilling a hole at the implant location using a specific drill, placing the fixture into the alveolar bone to osseointegrate it to the bone, coupling the abutment to the fixture, and finally crowning the prosthesis with the abutment to complete the treatment.

[0003] Dental implants can facilitate the restoration of a missing single tooth, enhance the function of dentures for partially or completely edentulous patients, improve the aesthetics of denture restorations, further distribute excess stress on the surrounding supporting bone tissue, and stabilize an entire row of teeth. A dental implant typically includes a fixture that serves as an artificial tooth root, a post coupled to the fixture, a post screw for securing the post to the fixture, and an artificial tooth coupled to the post. Before the post is coupled to the fixture (i.e., during osseointegration of the fixture into the alveolar bone), the healing post is coupled to the fixture and remains coupled.

[0004] The fixture, one of the implant components, is a part that is inserted into the hole formed in the alveolar bone with a drill bit and functions as an artificial tooth root. Therefore, the fixture is firmly placed in the alveolar bone. Implant placement varies from patient to patient, as the placement of the implant depends on many factors, such as the condition of the patient's teeth, the location of the teeth requiring implant treatment, and the condition of the patient's alveolar bone. In particular, alveolar bone density is a very important factor in implant placement, and the implant's placement position, depth, and direction are determined based on the patient's bone density, carefully tailored to the patient's characteristics.

[0005] As mentioned above, because implant placement varies significantly from patient to patient, imaging systems for dental implant diagnosis have been developed to help doctors accurately identify these differences. These imaging systems, based on related art, visualize the patient's oral area using, for example, computed tomography scans to assist in treatment simulations. However, existing issues are that indicators of bone density (which are crucial for determining the location, depth, and direction of implant placement) are inaccurate and often provided in a state that doctors cannot identify.

[0006] Specifically, conventional dental implant diagnostic image generation systems use only a colorless contrast to indicate alveolar bone density, making it difficult for doctors to quickly identify alveolar bone density. Furthermore, because only the overall alveolar bone density is displayed, without providing detailed information on the virtual locations where implants are to be placed, doctors need to spend a significant amount of time identifying the optimal implant placement location. Summary of the Invention

[0007] The present invention concept provides a device for automatic color conversion of oral computed tomography images based on artificial intelligence, which can automatically convert the color on the display according to the bone density of the alveolar bone (for example, when an implant is virtually placed in the alveolar bone), as well as a driving method of the device.

[0008] According to one aspect of the present inventive concept, a device for automatic color conversion of oral computed tomography images based on artificial intelligence includes: a storage unit configured to store a color conversion image pre-made by a user, wherein the color conversion image is related to the bone density of the alveolar bone; and a control unit configured to automatically convert the color of a subsequently input oral computed tomography input image, wherein the control unit automatically converts the color of the subsequently input oral computed tomography input image by: when an oral computed tomography input image of a patient is received, converting the color of the oral computed tomography input image based on a preset parameter value, and adjusting the preset parameter value by an error value, wherein the error value is generated from the converted oral computed tomography input image and the stored color conversion image.

[0009] The control unit may be further configured to: identify regions based on features extracted from the received oral computed tomography input image; and convert multiple colors of the identified multiple regions into different colors based on the preset parameter values.

[0010] The control unit may be further configured to calculate an error value after performing image pre-processing through morphology operations on the converted oral CT scan input image and the stored color-converted image.

[0011] The control unit may be further configured to convert the oral computed tomography input image based on an action selected by the user from a plurality of actions for adjusting the preset parameter value.

[0012] The control unit may be further configured to: quantize pixel values ​​of the oral computed tomography input image into multiple colors different from each other, and use the pixel values ​​of the oral computed tomography input image as the parameter values.

[0013] When the user places the virtual implant into the alveolar bone on the received oral CT input image of the patient, the control unit can automatically convert the color based on the action of placing the virtual implant into the alveolar bone and the bone density of the alveolar bone, and visualize and display the converted oral CT input image on the screen.

[0014] According to one aspect of the present inventive concept, a method for driving automatic color conversion of oral computed tomography images based on artificial intelligence includes: storing a color conversion image pre-made by a user in a storage unit, wherein the color conversion image is related to the bone density of the alveolar bone; and when an oral computed tomography input image of a patient is received, automatically converting the color of the subsequently input oral computed tomography input image by a control unit, wherein the control unit automatically converts the color of the subsequently input oral computed tomography input image by: converting the color of the received oral computed tomography input image based on a preset parameter value, and adjusting the preset parameter value by an error value, wherein the error value is generated from the converted oral computed tomography input image and the stored color conversion image.

[0015] In the automatic color conversion, regions may be identified based on features extracted from the received oral CT scan input image, and a plurality of colors of the identified plurality of regions may be converted into colors different from each other based on the preset parameter values.

[0016] In the automatic color conversion, after performing image pre-processing via morphological operations on the converted oral CT input image and the stored color-converted image, an error value may be calculated.

[0017] In the automatic color conversion, the oral CT scan input image may be converted by the control unit based on an action selected by the user from a plurality of actions for adjusting the preset parameter value.

[0018] In the automatic color conversion, the pixel values ​​of the oral computed tomography input image may be used as the parameter values ​​by quantizing the pixel values ​​of the oral computed tomography input image into a plurality of colors different from each other.

[0019] In automatic color conversion, when the user places the virtual implant into the alveolar bone on the received oral computed tomography input image of the patient, the color can be automatically converted based on the action of placing the virtual implant into the alveolar bone and the bone density of the alveolar bone, and can be visualized and displayed on the screen.

[0020] According to the present invention, by employing image preprocessing, artificial intelligence training, and a visualization method for automatic color segmentation of medical image regions, the accuracy of indicating alveolar bone density is improved, allowing doctors to easily identify alveolar bone density. Furthermore, according to one embodiment, the bone density of the virtual location where the implant is to be placed is visually and separately provided, enabling doctors to optimally place the implant. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Exemplary embodiments of the present inventive concept will be more clearly understood through the following detailed description taken in conjunction with the accompanying drawings, in which:

[0022] Figure 1 An artificial intelligence-based oral computed tomography color conversion system according to an embodiment of the present invention is illustrated;

[0023] Figure 2 yes Figure 1 A block diagram of the detailed structure of the oral computed tomography color conversion system is shown;

[0024] Figure 3 yes Figure 1 A block diagram of another detailed structure of the oral computed tomography color conversion system shown;

[0025] Figure 4 A number "N" of color conversion methods are depicted; and

[0026] Figure 5 yes Figure 1 The flowchart of the driving process of the oral computed tomography color conversion system is shown. DETAILED DESCRIPTION

[0027] The following describes embodiments of the present invention in detail with reference to the accompanying drawings. However, the present invention is not limited thereto. Furthermore, it should be understood that various changes in form and detail may be made to the present invention without departing from the spirit and scope of the appended patent applications. In other words, the specific structures or functions presented herein are merely illustrative of embodiments of the present invention.

[0028] Because the present invention allows for various variations and embodiments, multiple embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to a specific implementation mode. Furthermore, it should be understood that all variations, equivalents, and alternatives are included in the present invention without departing from the spirit and technical scope of the present invention.

[0029] Here, terms such as "first" and "second" are used only to describe various components, but these components are not limited by these terms. The purpose of using these terms is only to distinguish one component from another component. For example, a first component can be referred to as a second component, and vice versa, without departing from the correct scope of the present invention.

[0030] In this specification, when a component is “connected” or “connected” to another component, the component may be connected or connected to the other component directly or through at least one other component. Conversely, when a component is described as being “directly connected” or “directly connected” to another component, the component should be interpreted as being directly coupled to the other component without any other components in between. Other expressions, such as “between” and “directly between” to describe the relationship between components, should also be interpreted in the same manner.

[0031] The terms used in this specification are intended to explain a particular embodiment and are not intended to limit the inventive concept. Therefore, unless the context clearly indicates otherwise, expressions used in the singular in this specification also include expressions in the plural. Similarly, similar terms such as "comprise" or "include" may be interpreted as indicating a specific feature, quantity, step, operation, component, or combination thereof, but should not be interpreted as excluding the possibility of the presence of one or more additional features, quantities, steps, operations, components, or combinations thereof.

[0032] Unless otherwise defined, all terms (including technical or scientific terms) used herein have the same meaning as those commonly understood by those skilled in the art in the art to which the present invention may belong. Those terms defined in general dictionaries are interpreted as having meanings matching those in the context of the relevant technology, and should not be interpreted as ideally or excessively formal unless otherwise explicitly defined.

[0033] The present invention will be described in detail below by explaining preferred embodiments of the present invention with reference to the accompanying drawings, in which the same reference numerals represent the same elements.

[0034] Figure 1 An artificial intelligence-based oral computed tomography color conversion system according to an embodiment of the present inventive concept is illustrated.

[0035] like Figure 1 As shown, the artificial intelligence-based oral CT color conversion system 90 according to an embodiment of the present inventive concept may include a portion or all of the oral imaging device 100 and the oral CT color conversion device 110 .

[0036] The expression "including some or all" may indicate that the artificial intelligence-based intra-oral computed tomography color conversion system 90 is configured by omitting some components (e.g., the intra-oral imaging device 100 ), or may indicate that some components (e.g., the intra-oral imaging device 100 ) are integrally formed with the intra-oral computed tomography color conversion device 110 . To facilitate a thorough understanding of the present invention, the following description will use the expression "including all."

[0037] The intraoral imaging device 100 may include a device for capturing oral images of a patient or dental implant recipient. For example, the intraoral imaging device 100 may include various devices such as a computed tomography (CT) scanner or a magnetic resonance imaging (MRI) system. Furthermore, any device capable of capturing images of the alveolar bone within the patient's oral cavity may be used. For example, while CT scanners offer the advantage of accurately capturing the patient's bone shape, images may be distorted by various dentures and implants present in the patient's oral cavity.

[0038] Therefore, the intraoral imaging device 100 according to an embodiment of the present inventive concept may further include a three-dimensional (3D) scanner that can acquire an image by scanning the dental plaster mold G. The image data acquired by the scanner may include stereolithography (STL) data. The STL data may be in ASCII or binary format and may represent the surface of a three-dimensional object as polygons, thereby enabling the modeling data of the three-dimensional object in a three-dimensional program to be recognized in other three-dimensional programs.

[0039] Furthermore, the oral CT color conversion device 110 can include various devices such as desktop computers, laptop computers, tablet computers, smartphones, and smart TVs. The oral CT color conversion device 110 can pre-store color conversion images, which are pre-created by experts and correlated with the color conversion image (i.e., alveolar bone density). This configuration allows the user to easily understand each region at a glance. For example, a color conversion image is generated by an expert with extensive prior knowledge in related medical fields and image processing by segmenting the alveolar bone into a region. Using the first and second images captured by the aforementioned CT scanner and 3D scanner, the average value (M) of a plurality of conversion parameters and the range (R) of the HU value of a color are adjusted for each region. Upon receiving an oral CT scan input image from a user, the received CT input image is converted based on preset parameter values, a color conversion result state of the converted CT input image is determined, and an error value is reflected in the preset parameter value (the error value is generated based on the converted CT input image and a pre-stored color conversion image), thereby achieving automatic color conversion of the subsequently input CT input image.

[0040] In other words, the oral CT color conversion device 110 may include a program to assist with, for example, implant placement treatment planning. When an implant is virtually placed into the alveolar bone through execution of the program, the multiple colors on the display change based on the bone density of the alveolar bone, enabling the physician to perform implant placement. Specifically, to automatically convert the color of the alveolar bone density, the oral CT color conversion device 110 may pre-store previously generated image data related to the bone density of the alveolar bone (i.e., color-converted images) and train the image data through deep learning, etc. using artificial intelligence. Furthermore, the oral CT color conversion device 110 may pre-process each of the pre-stored image data and newly input image data, and then compare the differences between the image data. If the comparison results in an error value, the status of the preset parameters is adjusted based on the comparison result, and training is performed. During training, automatic parameter selection is possible, thereby improving accuracy.

[0041] For example, in an embodiment of the present invention, when an oral computed tomography image is input as a parameter, the region is segmented, and approximately "N" colors are used to represent the segmented regions. During the color presentation process, "N" colors are determined and presented because quantization is performed. The aforementioned "quantization" can be defined as a method of segmenting the range of a variable into a finite number of local ranges or blocks, wherein the local ranges or blocks do not overlap with each other. In this case, the conversion factor can be a parameter. The oral computed tomography color conversion device 110 according to an embodiment of the present invention can reflect an error value based on pre-stored image data (e.g., image data previously manually presented and stored by experts in the relevant field) to convert and output a newly input or subsequently input computed tomography input image by adjusting the conversion factor (i.e., parameter). Further detailed description will be provided later.

[0042] In this way, the oral CT color conversion device 110 can use the specified parameter values ​​to convert the oral CT input image, and then compare the image with the pre-stored image to determine whether an error value has occurred. This process can be performed for each action (for example, virtual implant placement). By training this process, the color conversion of the oral CT input image can be automated and the accuracy of the color conversion can be improved at the same time. For example, noise may appear during the color conversion process, and due to the limitations of the CT medical device, irregular conversion results may be produced. In summary, the oral CT color conversion device 110 performs color conversion by adjusting the initially set parameter values, thereby improving the accuracy of the color conversion.

[0043] Figure 2 yes Figure 1 The block diagram of the detailed structure of the oral computed tomography color conversion system is shown.

[0044] like Figure 2 As shown, the oral computed tomography color conversion device 110 according to an embodiment of the present inventive concept may include a communication interface unit 200, a control unit 210, an oral computed tomography color conversion unit 220, a user interface unit 230, a portion or all of a storage unit 240, and may also include a display unit.

[0045] In this case, the expression "including a portion or all of..." may indicate that the oral computed tomography color conversion device 110 is configured by omitting some components (e.g., the storage unit 240), or that some components (e.g., the oral computed tomography color conversion unit 220) are integrally formed with other components (e.g., the control unit 210). To facilitate a thorough understanding of the present invention, the following expression will be described as "including all of."

[0046] For example, the communication interface unit 200 may receive an oral CT scan image, wherein the oral CT scan image is obtained by photographing the oral cavity of the patient. Figure 1 The intraoral imaging device 100 shown in FIG. 1 is provided with a communication interface unit 200 that can transmit received intraoral CT scan images to the control unit 210. For example, when the communication interface unit 200 and the intraoral imaging device 100 are integrally formed, the intraoral CT scan images can be received and provided to the control unit 210 in an uncompressed state.

[0047] As described above, the communication interface unit 200 can receive the compressed image and perform operations such as decoding, or may not use the compressed image, depending on how the system designer configures the oral computed tomography color conversion device 110. Therefore, while various configurations are possible within the embodiments of the present inventive concept, the present disclosure is not limited to any particular form.

[0048] The control unit 210 can execute Figure 1The control unit 210 of the oral CT color conversion device 110, including the communication interface unit 200, the oral CT color conversion unit 220, the user interface unit 230, and the storage unit 240, is shown. For example, the control unit 210 may store training image data (i.e., pre-generated color-converted images in the storage unit 240) so that the oral CT color conversion unit 220 can perform color conversion operations based on artificial intelligence, and then provide the image data to the oral CT color conversion unit 220. In this case, the training image data stored in the storage unit 240 may be image data configured to facilitate user understanding of each region at a glance. This training image data is manually adjusted by an expert with extensive prior knowledge, using conversion parameters such as the average (M) and range (R) of pixel values ​​or HU values, to divide an input oral CT image into a plurality of regions (e.g., regarding the alveolar bone) and represent each region with a different color. This data may be obtained from an external medical organization, for example, for use. Collecting and using big data in related fields is helpful for deep learning in artificial intelligence. In other words, using large amounts of data can improve the objectivity or accuracy of the results.

[0049] Furthermore, the control unit 210 can receive an action to virtually place an implant into the alveolar bone via the user interface unit 230. The control unit 210 can receive a user interface signal. This action can be one of various events extracted from deep learning. The control unit 210 can provide the received interface signal to the oral CT color conversion unit 220, which can then provide the corresponding color change in the alveolar bone image to a display, etc. The color of the converted oral CT input image according to one embodiment can be visualized and displayed on a screen of the display.

[0050] The oral CT color conversion unit 220 may include a program to assist, for example, a dentist in implant placement treatment planning, and execute the program under the control of the control unit 210. When the oral CT color conversion unit 220 receives an input CT image of a patient through the communication interface unit 200, the input CT image may be stored in the storage unit 240.

[0051] Furthermore, for example, when the oral CTO color conversion unit 220 receives an action to place a virtual implant into the alveolar bone, the oral CTO color conversion unit 220 is driven to apply the action to the CTO input image stored in the storage unit 240. During this process, reference image data (i.e., previously generated and stored color-converted images) pre-stored in the storage unit 240 can be referenced and trained. For example, the oral CTO color conversion unit 220 can use specified parameter values ​​to perform color conversion on the received oral CTO input image of a specified area. However, due to noise and limitations of CTO medical devices, irregular conversion results may be generated on the color-converted oral CTO input image.

[0052] As described above, the oral CT color conversion unit 220 references and trains image data to modify specified parameter values. This image data is obtained by manually converting oral CT images of multiple patients through manual manipulation by an expert with extensive prior knowledge in the relevant field. This parameter value modification is performed for each different action. Thus, the oral CT color conversion unit 220 adjusts the error value to improve the accuracy of color conversion, and can then automatically display the color-converted image data on a display, etc.

[0053] The user interface unit 230 can receive various instructions from a user (e.g., a dentist). For example, the user interface unit 230 can represent a computer keyboard or mouse. The user interface unit 230 can also include a display unit (e.g., a monitor). With respect to the display unit, if it includes a touch panel, touch control of the screen can be possible to receive user instructions.

[0054] For example, a dentist may obtain an oral CT scan image of a patient, then execute a program according to an embodiment of the present invention and place a virtual implant into the alveolar bone displayed on a monitor to check the color of the bone density of the alveolar bone. Therefore, by determining the placement strength or the exact position, the accuracy of implant placement can be improved. To this end, the oral CT color conversion unit 220 can use image data generated by an expert through existing manual operations, and can correct error values ​​or limitations of the CT medical device by comparing the image data generated by the expert and performing training based on artificial intelligence in the processing, thereby improving accuracy.

[0055] Under the control of the control unit 210, the storage unit 240 can store various information and data. In this case, the information may refer to interface signals set by the user (i.e., the dentist), and the data may refer to image data of captured images, etc. However, embodiments of the present invention are not limited to these terms.

[0056] The storage unit 240 can store training data obtained by a specialist through manual manipulation of color conversion of oral CT scan images obtained from a plurality of patients. Under the control of the control unit 210, the training data can be provided to the oral CT color conversion unit 220.

[0057] Figure 3 yes Figure 1 The block diagram of another detailed structure of the oral computed tomography color conversion system is shown. Figure 4 A number "N" of color conversion methods are depicted.

[0058] like Figure 3 As shown, an oral computed tomography color conversion device 110' according to another embodiment of the present inventive concept can perform image preprocessing, artificial intelligence training, and visualization operations for automatic color differentiation of medical image regions, and can include a feature extraction unit 300, an agent unit 310, and part or all of a color conversion unit 320.

[0059] In this context, the expression "including a portion or all of..." may indicate that the oral computed tomography color conversion device 110 is configured by omitting some components (e.g., the feature extraction unit 300), or that some components (e.g., the color conversion unit 320) are integrated with other components (e.g., the proxy unit 310). To facilitate a thorough understanding of the present invention, the following expression will be described as "including all of." Furthermore, the aforementioned components may be configured with software (S / W), hardware (H / W), or a combination of both.

[0060] The feature extraction unit 300 can extract features from the input oral computed tomography image 295. For example, the computed tomography image can be received in units of frames or macroblocks (i.e., obtained by dividing the image of a unit frame into multiple blocks). Therefore, the feature extraction unit 300 can extract states for reinforcement learning, for example, through feature extraction. In detail, the feature extraction unit 300 can use various feature extraction algorithms to extract states for reinforcement learning from the input computed tomography image. For example, various algorithms such as U-Net, AutoEncoder, and convolutional neural network (CNN) can be used.

[0061] The agent unit 310 can train the agent to have appropriate color conversion by the following actions: allowing the color-converted image generated based on the result of reinforcement learning to undergo image preprocessing (for example, morphological operations, etc.), and then defining the difference "E(i-i_)^2" between the color-converted image and the color-converted image produced by an expert and subjected to the same image preprocessing as an error value, and defining the inverse of the error value as a reward for reinforcement learning.

[0062] First, the process of manually generating image data by the agent unit 310 will be described. Features of medical images using CBCT, etc., have different Hounsfield Unit (HU) values ​​for different regions. For CBCT scans, HU values ​​may have different scales depending on the physical characteristics of each region (e.g., teeth, alveolar bone, etc.). Furthermore, using CBCT, etc. to effectively visualize medical images, it becomes possible to segment parameters based on color. In order to convert (quantize) the pixel values ​​of each different region into different colors, "N" colors can be determined, and the "N" HU averages and ranges can be manually adjusted by the user. When the actual pixel value (HU) is included in "N" different quantized regions, the pixel values ​​of each different region are quantized into representative color values, thereby obtaining a color quantity or result. Figure 4 A number "N" of color conversion methods are depicted.

[0063] According to an embodiment of the present invention, in a manual method for generating a pre-generated color conversion image, in order to segment into different colors, an expert with rich prior knowledge may include the mean value (M) of the HU values ​​of the colors ( Figure 4 The "HU average value" in the color) and the range of HU values ​​(R) ( Figure 4The conversion parameters of the image (the "HU value range") are adjusted to manual expression. Therefore, it is possible for the user to understand the configuration of each area at a glance. Conversion (quantization) by filtering in units of pixel values ​​(HU values) may cause the color conversion result to be irregular (although it is in the same area). In other words, due to noise and limitations of computed tomography medical devices, even in the same area, the conversion image using HU values ​​(which is a conversion by linear filtering using the range of pixel values) may cause the conversion result to be irregular. In order to solve the above problem, it is possible to reconstruct a regular color conversion result by applying image processing techniques (such as morphological operations, etc.) to the conversion image prepared by experts.

[0064] Therefore, in an embodiment of the present invention, to implement an expert-level automatic color conversion algorithm based on reconstructed color conversion results, various parameters (M, R) and the state of the color conversion result for manual conversion are defined based on a reinforcement learning method (e.g., a deep Q-network, etc.), and the behavior of the conversion parameter values ​​is defined, so that expert-level color conversion is automatically performed. Through image preprocessing (e.g., morphological operations and similar image processing), the conversion image produced by an expert using existing manual methods is corrected to a regular color-converted image. This image is used as output (target) data, which is reconstructed from a computed tomography scan (using the current parameters (M, R) as input data). The difference between the reconstructed image and the result of image preprocessing (e.g., morphological operations, etc.) is defined as an error value, and the inverse of the error value is used as a reward for the deep Q-network (DQN) structure to train the conversion parameter state. As training continues, the automatic selection of parameters is transformed to be similar to that of an expert.

[0065] According to one embodiment, agent unit 310 can reconstruct color-converted images created by experts through morphological operations, etc., into regular color conversion results based on image preprocessing used to purify training data, thereby purifying training data and improving the efficiency of artificial intelligence training. Parameters (p) are defined as (N, m_n, m_r). Parameters (p) are the parameters required for the expert to manually perform image conversion based on the DQN structure (a type of reinforcement learning model, which serves as a background model).

[0066] The action unit (not shown) can update the color conversion result based on the result selected from the three result definitions of reinforcement learning (e.g., increase, decrease, and maintain) (i.e., "3*N" reinforcement learning results) to adjust "N" conversion parameter (P) values.

[0067] The color conversion unit 320 (environment) may perform a color conversion operation using the parameters determined by the agent unit 310. For example, the color conversion unit 320 may convert a color based on parameters associated with a segmented region of a computed tomography input image.

[0068] Furthermore, the color conversion unit 320 may perform a feedback operation on the agent unit 310. In one embodiment, the feedback operation may correspond to a reward operation. In this case, the reward is the inverse of the difference (mean square error) between the parameters of the determined action and the actual parameters, which may correspond to the inverse of the difference (pixel-wise cross-entropy) between an image and a color conversion result, where the image is converted to color using the parameters determined by the determined action and image pre-processing (e.g., morphology) is performed, and the color conversion result is manually created by an expert through image pre-processing (e.g., morphology).

[0069] The agent unit 310 can define a parameter (p), which is a parameter required for an expert to manually perform image conversion based on the DQN structure (which is a type of reinforcement learning model and serves as a background model). The agent unit 310 can also reconstruct the color conversion image into a regular color conversion result through an image preprocessing model used to purify training data (such as morphological operations, etc.), thereby purifying the training data and improving the training efficiency of artificial intelligence.

[0070] Figure 5 yes Figure 1 The flowchart of the driving process of the oral computed tomography color conversion system is shown.

[0071] For ease of explanation, please also refer to Figure 5 and Figure 1 , according to an embodiment of the inventive concept, Figure 1 The illustrated oral CT color conversion device 110 stores a color-converted image pre-created by a user or an expert in a storage unit. The color-converted image is related to the bone density of the alveolar bone or teeth (S500). In this case, the pre-created color-converted image may be a reference image or a standard image used when converting the color of an oral CT image input by photographing the patient's oral cavity.

[0072] In addition, when the oral CT color conversion device 110 receives an oral CT input image of a patient, it converts the received oral CT input image based on a preset parameter value, and reflects an error value in the preset parameter value (the error value is generated based on the converted oral CT input image and the pre-stored color conversion image), thereby achieving automatic color conversion of the subsequently input oral CT input image (S510).

[0073] In addition to the above, Figure 1 The oral computed tomography color conversion device 110 shown can perform various operations. Since other details have been described in detail above, redundant descriptions are not repeated here.

[0074] Although it has been described above that all the components in the embodiments of the present disclosure are coupled to act as a single unit or are coupled to be operated as a single unit, the present disclosure is not necessarily limited to such embodiments. That is, for the purpose of the present disclosure, one or more components in the plurality of components can be selectively coupled to operate as one or more units. Moreover, although each component can be implemented as independent hardware, some or all components can be selectively combined with each other so that they can be implemented as a computer program of one or more program modules, some or all of which functions can be executed in one or more hardware. The program code and program code fragments that make up the computer program can be easily thought of by those with ordinary knowledge in the technical field to which the present disclosure belongs. Such a computer program can be stored in a non-transitory computer-readable medium and read and executed by a computer to implement the embodiments of the present disclosure.

[0075] A non-transitory readable recording medium is not a medium that can only store data for a short period of time (such as registers, caches, and memory). Instead, it refers to a medium that can semi-permanently store data and can be read by a device. Specifically, the program can be provided by storing it on a non-transitory readable recording medium such as a CD, DVD, hard drive, Blu-ray disc, USB, memory card, or read-only memory (ROM).

[0076] Thus, the present disclosure has been particularly shown and described with reference to the preferred embodiments thereof, and those skilled in the art will understand that various changes in form and details may be made without departing from the spirit and scope of the inventive concept as defined by the appended claims. The preferred embodiments should be considered in a descriptive sense only and not for purposes of limitation. Therefore, the scope of the inventive concept is defined not by the detailed description of the inventive concept but by the appended claims, and all differences within the scope will be construed as being included in the inventive concept.

[0077] The concept of the present invention can be used in the artificial intelligence industry and the dental medical industry.

Claims

1. An artificial intelligence-based automatic color conversion device for oral computed tomography images, comprising: a storage unit configured to store a color conversion image pre-made by a user, wherein the color conversion image is related to the bone density of the alveolar bone; and A control unit configured to automatically convert the color of a subsequently input oral computed tomography input image, wherein the control unit automatically converts the color of the subsequently input oral computed tomography input image by: converting the color of the received oral computed tomography input image based on a preset parameter value when the oral computed tomography input image of the patient is received, and adjusting the preset parameter value by an error value, wherein the error value is generated from the converted oral computed tomography input image and the stored color-converted image; wherein The oral computed tomography input image inputted later refers to the oral computed tomography input image inputted after adjusting the preset parameter value; The control unit is further configured to calculate an error value after performing image pre-processing by morphological operations on the converted oral computed tomography input image and the stored color conversion image respectively.

2. The device according to claim 1, wherein The control unit is further configured to: identify regions based on features extracted from the received oral computed tomography input image; and convert multiple colors of the identified multiple regions into different colors based on the preset parameter values.

3. The device according to claim 1, wherein The control unit is further configured to convert the received oral computed tomography input image based on an action selected by the user from a plurality of actions for adjusting the preset parameter value, wherein the selected action is an action of the user interface unit virtually placing the implant into the alveolar bone.

4. The device according to claim 1, wherein The control unit is further configured to: quantize the pixel values ​​of the received oral computed tomography input image into multiple colors different from each other, and use the pixel values ​​of the received oral computed tomography input image as the preset parameter values.

5. The device according to claim 1, wherein When the user places the virtual implant into the alveolar bone on the received oral CT input image of the patient, the control unit automatically converts the color based on the action of placing the virtual implant into the alveolar bone and the bone density of the alveolar bone, and visualizes and displays the converted oral CT input image on the screen.

6. A method for driving a device for automatically converting the color of an oral computed tomography image based on artificial intelligence, the method comprising: storing a color conversion image pre-made by a user in a storage unit, wherein the color conversion image is related to the bone density of the alveolar bone; and When an oral computed tomography input image of a patient is received, the control unit automatically converts the color of the subsequently input oral computed tomography input image, wherein the control unit automatically converts the color of the subsequently input oral computed tomography input image by: converting the color of the received oral computed tomography input image based on a preset parameter value, and adjusting the preset parameter value by an error value, wherein the error value is generated from the converted oral computed tomography input image and the stored color-converted image; wherein The oral computed tomography input image inputted later refers to the oral computed tomography input image inputted after adjusting the preset parameter value; In the automatic color conversion, after image pre-processing is performed on the converted oral CT input image and the stored color-converted image through morphological operations, an error value is calculated.

7. The method according to claim 6, wherein In the automatic color conversion, regions are identified based on features extracted from the received oral CT scan input image, and a plurality of colors of the identified plurality of regions are converted into colors different from each other based on the preset parameter values.

8. The method according to claim 6, wherein In automatic color conversion, the received oral computed tomography input image is converted by the control unit based on an action selected by the user from a plurality of actions for adjusting the preset parameter value, wherein the selected action is an action of the user interface unit virtually placing the implant into the alveolar bone.

9. The method according to claim 6, wherein In the automatic color conversion, the pixel values ​​of the received oral computed tomography input image are quantized into multiple colors different from each other, and the pixel values ​​of the received oral computed tomography input image are used as the preset parameter values.

10. The method according to claim 6, wherein In automatic color conversion, when the user places the virtual implant into the alveolar bone on the received oral computed tomography input image of the patient, the color is automatically converted based on the action of placing the virtual implant into the alveolar bone and the bone density of the alveolar bone and is visualized and displayed on the screen.

Images