Device and method for extracting medical information

Abstract

An apparatus and method for extracting medical information according to an embodiment are provided. An apparatus for extracting medical information according to an embodiment includes one or more processors; and a memory storing instructions for executing the one or more processors, and is an apparatus for extracting medical information, wherein the processor: detects a user input requesting extraction of medical information for the target image including the attribute data and the feature data of the target image, extracts the attribute data of the target image in response to the user input, classifies the feature data of the target image to extract the medical information, and identifies a region of interest in the target image corresponding to the target image and the medical information using a learned model.

Description

Technical Field

[0001] The disclosed embodiments relate to an apparatus and method for extracting medical information.

[0002] [Cross - Reference to Related Applications] This application claims priority based on Korean Provisional Application No. 10 - 2022 - 0071586, filed on June 13, 2022, and the entire contents of the said application are incorporated herein.

Background Art

[0003] Generally, medical information is managed by a PACS (Picture Archiving and Communication System; medical image storage and transmission system). However, due to patient personal information protection, PACS is generally restricted in processing.

[0004] Even if processing is possible, uploading and downloading medical information involves a complex procedure as it goes through multiple steps. Furthermore, due to personal information protection, patient consent may be required for the progress of the said procedure.

[0005] Moreover, medical information managed by PACS is derived from various medical equipment, and its format and protocol vary for each medical equipment. That is, when medical equipment is different, the compatibility between medical information is quite insufficient.

[0006] Therefore, in order to strengthen the protection of patients' personal information and the ease of processing medical information, there is a need for a system that can integrally execute a series of medical information extraction processes with just one execution.

Summary of the Invention

Problems to be Solved by the Invention

[0007] The disclosed embodiments are for continuously executing a series of multiple steps for extracting medical information in one instruction in one queue with one command.

Means for Solving the Problem

[0008] An apparatus for extracting medical information according to an embodiment includes one or more processors; and a memory for storing instruction words for executing the one or more processors. The apparatus for extracting medical information, wherein the processor: detects a user input requesting extraction of medical information for a target image including attribute data and feature data of the target image, extracts the attribute data of the target image in response to the user input, classifies the feature data of the target image to extract the medical information, and uses a learned model to identify a region of interest in the target image corresponding to the target image and the medical information.

[0009] The attribute data includes at least one of the resolution and the number of channels of the target image, and the feature data may include a classification prediction value for at least one of the modality, imaging direction, and number of signal channels of the target image.

[0010] The processor: uses a learned model to detect one or more candidate bounding boxes corresponding to each of one or more objects included in the target image, calculates a corresponding confidence for each of the one or more candidate bounding boxes, and may align the one or more candidate bounding boxes based on the confidence.

[0011] The processor: removes overlapping bounding boxes based on the confidence of each of the one or more candidate bounding boxes, and may identify a region of interest based on at least one of the remaining one or more candidate bounding boxes.

[0012] The processor may select a reference bounding box based on the confidence of each of the one or more candidate bounding boxes, remove overlapping bounding boxes from the one or more candidate bounding boxes based on the coincidence rate between the prediction region of the reference bounding box and the prediction regions of the remaining comparison target bounding boxes, and identify the region of interest based on the candidate bounding box having the highest confidence among the remaining one or more candidate bounding boxes.

[0013] When the processor detects user input including at least one of a predefined mouse gesture input, shortcut key input, icon input, and voice command, the processor may start extracting the medical information using a learned model.

[0014] The processor may identify the boundary of the region of interest based on a second input of a user designating the region of interest from the target image, and manually extract the medical information based on a third input of the user designating the attribute of the region of interest.

[0015] The processor may output a list of equipment including one or more medical equipment communicating with the device for extracting the medical information, and transmit at least one of the target image, the region of interest, the attribute data, and the feature data to the medical equipment selected based on a fourth user input in the list of equipment.

[0016] When the processor receives a fifth user input of dropping the target image or the region of interest into a preset region, the processor transmits it to a model that matches the target image based on at least one of the attribute data and the feature data among the one or more learned models, causes the matching model to perform a medical analysis on the target image, and when the medical analysis is completed, may output at least one of a visual notification signal and an auditory notification signal.

[0017] The processor may output a warning signal when at least one of the vertical resolution, horizontal resolution, and ratio of the vertical resolution to the horizontal resolution of the target image is less than or equal to a preset value.

[0018] A method for extracting medical information according to an embodiment is a method performed by an apparatus for extracting medical information including one or more processors; and a memory storing instruction words for executing the one or more processors, the method comprising: detecting a user input requesting extraction of medical information for a target image including attribute data and feature data of the target image; extracting the attribute data of the target image in response to the user input, classifying the feature data of the target image, and extracting the medical information; and identifying a region of interest in the target image corresponding to the target image and the medical information using a trained model.

[0019] The attribute data includes at least one of the resolution and the number of channels of the target image, and the feature data may include classification prediction values for at least one of the modality, imaging direction, and number of signal channels of the target image.

[0020] The step of extracting the medical information may include detecting one or more candidate bounding boxes corresponding to each of one or more objects included in the target image using a trained model; calculating a corresponding confidence level for each of the one or more candidate bounding boxes; and aligning the one or more candidate bounding boxes based on the confidence level.

[0021] The step of extracting the medical information may include removing overlapping bounding boxes based on the confidence level of each of the one or more candidate bounding boxes; and identifying a region of interest based on at least one of the remaining one or more candidate bounding boxes.

[0022] The step of extracting the medical information may include: selecting a reference bounding box based on the confidence of each of the one or more candidate bounding boxes; removing overlapping bounding boxes among the one or more candidate bounding boxes based on the coincidence rate between the prediction region of the reference bounding box and the prediction regions of the remaining comparison target bounding boxes; and identifying the region of interest based on the remaining one or more candidate bounding boxes.

[0023] The step of extracting the medical information may include starting to extract the medical information using a trained model when detecting user input including at least one of a predefined mouse gesture input, a shortcut key input, an icon input, and a voice command.

[0024] The step of extracting the medical information may include: identifying the boundary of the region of interest based on a second input of the user designating the region of interest from the target image; and manually extracting the medical information based on a third input of the user designating the attribute of the region of interest.

[0025] The method may include: outputting a list of equipment including one or more medical equipment communicating with the device for extracting the medical information; and transmitting at least one of the target image, the region of interest, the attribute data, and the feature data to the medical equipment selected based on a fourth user input from the list of equipment.

[0026] The step of transmitting may include: when receiving a fifth user input for dropping the target image or the region of interest into a preset region, transmitting to a model that matches the target image based on at least one of the attribute data and the feature data among one or more of the learned models; causing the matching model to perform a medical analysis on the target image; and when the medical analysis is completed, outputting at least one of a visual notification signal and an auditory notification signal.

[0027] The method may further include: outputting a warning signal when at least one of the vertical resolution of the target image, the horizontal resolution of the target image, and the ratio of the vertical resolution to the horizontal resolution is less than or equal to a preset value.

Advantages of the Invention

[0028] The disclosed embodiments can simply and continuously execute a series of multiple steps for extracting medical information in one command and in one queue.

Brief Description of the Drawings

[0029]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

[0030] Hereinafter, specific embodiments of an example will be described with reference to the drawings. The following detailed description is provided to assist in a comprehensive understanding of the invention disclosed herein. However, this is merely an example and the invention is not limited thereto.

[0031] When explaining an example, if it is determined that a detailed description of known techniques related to the present invention may unnecessarily obscure the gist of the example, the detailed description will be omitted.

[0032] The terms described below are terms defined in consideration of the operations in the present invention, which may vary depending on the intention or convention of the user or operator. Therefore, the definition should be determined based on the content throughout this specification. The terms used in the detailed description are merely for describing an example and should never be restrictive. Also, unless otherwise specified, singular expressions include plural meanings. In this description, expressions such as "including" or "comprising" are used to refer to a certain component, number, step, operation, element, part thereof, or combination, and should not be construed as excluding the existence or possibility of one or more other components, numbers, steps, operations, elements, part thereof, or combinations other than those described.

[0033] In addition, the embodiments described in this specification may have aspects that are entirely hardware, partly hardware and partly software, or entirely software. In this specification, "unit", "layer", "module", "device", "server", or "system", etc. refer to computer-related entities such as hardware, a combination of hardware and software, or software. For example, a unit, layer, module, device, server, or system may refer to hardware that constitutes part or all of a platform and / or software such as an application for driving the hardware. As a specific example, a unit, layer, module, device, server, or system may be implemented by a processor.

[0034] FIG. 1 is a block diagram for explaining an apparatus 100 for extracting medical information according to an embodiment.

[0035] The apparatus 100 for extracting medical information according to an embodiment is an apparatus that provides a simple data pipeline for an acquired target image. Specifically, the apparatus 100 for extracting medical information may mean an apparatus that detects a user input and extracts medical information in a queue.

[0036] Here, the medical information may include a region of interest selected as being suitable for being captured or cropped from the target image, the type and format of the target image, etc.

[0037] Referring to FIG. 1, the apparatus 100 for extracting medical information includes a processor 110 and a memory 120. The processor 110 detects a user input that requests extraction of medical information for a target image including attribute data and feature data of the target image.

[0038] Here, the target image may be at least one of an image including a patient's biological signal and the result of medical analysis of the biological signal. For example, the target image may include at least one of an electrocardiogram image and a chest radiograph image.

[0039] When the processor 110 detects a predefined user input, it may start extracting medical information using a learned model.

[0040] Here, the predefined user input may include at least one of, for example, a mouse gesture input mapped to start extracting medical information, a shortcut key input, a menu click input, an icon click input, and a voice command.

[0041] As a specific example, when the processor 110 detects a click input of the mouse on the target image, it may continuously perform a series of steps necessary for extracting medical information of the target image and specifying the region of interest.

[0042] As a specific example, when the processor 110 detects an input of a shortcut key mapped to a class predicted corresponding to the target image, it may continuously perform a region-of-interest specifying step corresponding to the shortcut key.

[0043] That is, while the processor 110 is waiting for the start of the extraction function in the background, when it detects a user input, it may be interpreted as calling the extraction function.

[0044] At this time, the processor 110 may change a visual signal including the shape, pattern, hue, and / or brightness of the screen or the user interface in order to let the user recognize that a user input has been detected and a request has been made.

[0045] On the other hand, although the user input for starting the extraction of medical information has been described as a mouse input, a shortcut key input, etc., this is exemplary and is not necessarily limited to the examples listed.

[0046] In response to a user input, the processor 110 extracts attribute data of a target image, classifies the feature data of the target image, and extracts medical information.

[0047] The attribute data is data for explaining the nature of the target image, and may include, for example, at least one of the resolution, number of channels, color space, bit depth, image format, contamination level, noise level, and compression method of the target image. At this time, the attribute data may be expressed in an array format.

[0048] The feature data may be data including at least one of the modality, imaging direction, and number of signal channels of the target image. At this time, the feature data may be expressed as a set of classes predicted by a learned model.

[0049] Specifically, the feature data may include a predicted value obtained by classifying what kind of medical image the target image is. As a specific example, the feature data may be expressed in a corresponding class by classifying the modality into an electrocardiogram or a chest radiograph.

[0050] As another example, the feature data may be expressed in a corresponding class by classifying the imaging direction of the target image into a posteroanterior (PA) direction, an anteroposterior (AP) direction, or a lateral (Lat) direction. As another example, when the feature data is signal data such as an electrocardiogram, at least one of the total number of channels and the type of channels constituting the signal data may be predicted and expressed in a corresponding class. For example, the feature data may be expressed in a corresponding class by predicting the number of signal channels of the electrodes used for the examination corresponding to the target image as 1 channel (e.g., lead II), 3 channels (e.g., lead I, II, III), or 12 channels.

[0051] The processor 110 uses a learned model to identify a region of interest in the target image corresponding to the target image and medical information.

[0052] According to one example, the processor 110 may identify a region of interest in the target image by associating the attribute data with the feature data using a learned model.

[0053] Here, the region of interest may mean a region in the target image that affects the result of medical diagnosis. In other words, the region of interest is a region that contains important objects that influence the medical diagnosis result, and there may be a plurality of such regions in one target image.

[0054] According to one example, the processor 110 may identify a region of interest by detecting one or more candidate bounding boxes corresponding to the objects included in the target image using a learned model.

[0055] According to one example, the processor 110 may identify at least one of the one or more candidate bounding boxes as the region of interest based on the confidence score of the one or more detected candidate bounding boxes in the target image.

[0056] Here, the confidence score may be the probability of the bounding box that defines a region including at least one of an electrocardiogram image or a chest radiograph image in a predetermined manner in the learned model and a statistical definition corresponding thereto, and may include the detailed types and regions of the electrocardiogram image and the detailed types and regions of the chest radiograph image.

[0057] Specifically, the processor 110 may define the class and region for the one or more detected candidate bounding boxes, and identify at least one of the candidate bounding boxes as the region of interest based on the confidence score of each bounding box.

[0058] The processor 110 selects, as the correct answer, the bounding box with the highest reliability prediction, calculates the coincidence rate between the prediction areas of one or more other candidate bounding boxes and the selected correct answer value, and may remove overlapping bounding boxes among the one or more candidate bounding boxes.

[0059] The processor 110 may then repeatedly perform the process of selecting the bounding box with the highest reliability among the remaining bounding boxes in the same manner and removing overlapping bounding boxes to specify the region of interest and the class for the target image.

[0060] The processor 110 may apply at least one of the attribute data and the specific data to a predefined candidate extraction function to select an arbitrary bounding box as the region of interest. Alternatively, the processor 110 may specify the candidate bounding box selected by the user as the region of interest. According to one example, the processor 110 may manually detect the region of interest based on user input.

[0061] Specifically, the processor 110 may identify the boundary and class of the region of interest based on a second input of the user who designates the region of interest from the target image.

[0062] For example, when the processor 110 detects a second input of the user that includes at least one of a drag input and a click input that specify the boundary of the region of interest, the processor 110 may manually identify the region of interest based on at least one of the drag input and the click input. A new window may be displayed so that the user can specify the class before, during, or after the start of the drag input and the click, or the user may be made aware that the class can be specified, including changing at least one of the shape, pattern, hue, and brightness of the bounding box. Also, the class may be specified by additional input.

[0063] Before identifying the region of interest, the processor 110 may determine the extraction suitability of the target image. In other words, the processor 110 may determine whether the target image is suitable for use as an extraction target for medical information.

[0064] Specifically, the processor 110 may determine the extraction suitability of the target image by using the attribute data of the target image to determine whether the target image meets preset qualitative or morphological conditions.

[0065] For example, if at least one of the vertical resolution, horizontal resolution, and the ratio of vertical resolution to horizontal resolution of the target image is less than or equal to a preset value, the processor 110 may exclude the target image from the extraction target.

[0066] The processor 110 may determine the extraction suitability of the target image by using the attribute data to determine whether the target image meets preset qualitative conditions.

[0067] For example, if the contamination level and noise level of the target image are greater than or equal to preset values, the processor 110 may exclude the target image from the extraction target. If the image format of the target image does not conform to a predefined format, the processor 110 may exclude the target image from the extraction target.

[0068] If the processor 110 determines that the target image does not meet at least one of the morphological conditions and qualitative conditions for using the target image, it may output a warning signal.

[0069] The processor 110 may qualitatively and / or quantitatively evaluate the extraction suitability based on at least one of the morphological condition result and qualitative condition result of the target image.

[0070] The memory 120 stores one or more instruction words executed by the processor 110.

[0071] Memory 120 may store various data used by processor 110. For example, memory 120 may include software (e.g., programs executed by processor 110 and / or instruction words related to the programs, libraries related to the instruction words, input data or output data for the instruction words).

[0072] Memory 120 may include volatile memory 120 or non-volatile memory 120.

[0073] FIG. 2 is an exemplary diagram for explaining menu 10 and icon 20 for calling the extraction function of device 100 for extracting medical information.

[0074] Referring to FIG. 2, the first image includes menu 10 and icon 20.

[0075] Processor 110 may display menu 10 at the upper part of the first image. Processor 110 may display one or more menus 10 aggregated in a toolbar for enhancing user convenience.

[0076] Processor 110 may display icon 20 in the content area of the first image. The content area is the lower area below the toolbar and may include the area where the user performs operations. For example, it may be an area for performing operations such as editing an image, uploading an image, etc.

[0077] FIG. 3 is an exemplary diagram for explaining menu 10, icon 20, and sub-menus 101, 102, 103 for calling the extraction function of device 100 for extracting medical information.

[0078] Referring to FIG. 3, the second image displays menu 10 and sub-menus 101, 102, 103. Specifically, processor 110 outputs one or more sub-menus 101, 102, 103 that are displayed when menu 10 is selected based on user input.

[0079] At this time, the processor 110 can change at least one of the shape, pattern, hue, and brightness of the menu 10 selected by the user input (e.g., clicked), and make the user recognize that the selected menu 10 has been executed.

[0080] After that, the processor 110 may output the sub-menus 101, 102, 103 included in the selected menu 10 near the selected menu 10 so that the user can execute the operations of the sub-menus 101, 102, 103. For example, the processor 110 may output the sub-menus 101, 102, 103 in a form in which the sub-menus 101, 102, 103 are expanded below the menu 10.

[0081] That is, the processor 110 may output the sub-menus 101, 102, 103 of the menu 10 by overlapping them on the same image where the menu 10 is output. For example, when the menu 10 is arranged in the first image, the processor 110 may output the sub-menus 101, 102, 103 in the first image. When the menu 10 is arranged in the second image, the processor 110 may output the sub-menus 101, 102, 103 in the second image.

[0082] At this time, at least one of the menu 10 and the sub-menus 101, 102, 103 may be visualized by at least one of intuitive symbols, patterns, and images that explain the corresponding operations.

[0083] For example, the menu 10 or the sub-menus 101, 102, 103 related to the capture operation may be visualized as an image including a logo, mark, symbol, etc. in the shape of a camera, and intuitively provide an explanation about the capture operation.

[0084] As another example, the menu 10 related to upload may be visualized as an image including a logo, mark, symbol, etc. in the shape of an upward arrow, and intuitively provide an explanation about the upload operation.

[0085] On the one hand, when the processor 110 receives a user input requesting the execution of the icon 20, it may immediately execute the sub-menus 101, 102, 103 corresponding to the icon 20 without querying one or more sub-menus 101, 102, 103 included in the menu 10.

[0086] The icon 20 may be visualized by at least one of an intuitive symbol, pattern, and image that explains the corresponding operation.

[0087] FIG. 4 is an exemplary diagram for explaining the process by which the apparatus 100 for extracting medical information extracts medical information.

[0088] The target image in FIG. 4 has been selected as the target image and is in a state of waiting for the call of the extraction function by user input.

[0089] Here, the input image is a medical image (e.g., an electrocardiogram image) acquired by the apparatus 100 for extracting medical information from the target of medical analysis, and may be, for example, an image acquired from an Electronic Medical Record.

[0090] When the processor 110 detects a predefined user input, it may extract the medical information of the target image to be uploaded or displayed on the screen.

[0091] Here, the processor 110 may identify the region of interest using a learned model in an automatic designation manner.

[0092] The processor 110 may identify the region of interest from the target image using a learned model mapped to the input shortcut key. For example, when the shortcut key "1" mapped to the same as "1" designated as the class of the electrocardiogram image is input, the processor 110 may identify the region of interest customized for the electrocardiogram image using the learned model.

[0093] Processor 110 may transmit the region of interest to an analysis model mapped based on at least one of attribute data, feature data, and user input, and cause the analysis model to analyze a medical diagnosis corresponding to the region of interest.

[0094] As another example, processor 110 may identify the region of interest based on user input in a manually specified manner.

[0095] Specifically, processor 110 may identify the region of interest based on a diagonal user drag input. Alternatively, processor 110 may also identify the region of interest based on a user click or touch input that selects coordinates.

[0096] As a specific example, processor 110 may manually identify the region of interest by a user input of dragging from a starting point to an ending point. Processor 110 may manually identify the region of interest based on a user input of clicking or touching four points.

[0097] After the region of interest is identified, processor 110 may generate a captured image based on the region of interest. Thereafter, after generating the captured image, processor 110 may display on the same screen a list of medical equipment capable of transmitting the captured image.

[0098] Specifically, when processor 110 detects a user right-click input or a click input on icon 20 that performs the operation, or it may display on the same screen a list of medical equipment that can be automatically transmitted.

[0099] Thereafter, when processor 110 receives a user input to drop the captured image in a preset region, it may cause a medical analysis corresponding to the biological signal type of the target image to be performed on the analysis model.

[0100] Thereafter, when the medical analysis is completed, processor 110 may cause device 100 that extracts medical information to output at least one of a visual notification signal and an auditory notification signal.

[0101] For example, the processor 110 may cause the device 100 for extracting medical information to output a notification signal including at least one of a pop-up message on the display, an animation effect, and a notification sound to the display.

[0102] FIG. 5 is an exemplary diagram illustrating a process in which the device 100 for extracting medical information uploads a target image.

[0103] When the processor 110 receives a user input for selecting the second sub-menu 102 or the second icon 20-2, the processor 110 may perform a file search and display a storage path of the target image.

[0104] Thereafter, when the processor 110 receives a user input for selecting the target image, the processor 110 may capture the selected target image. Here, the target image may be an image pre-stored in the device 100 for extracting medical information or a medical image that may be provided from the outside.

[0105] After uploading the image, the processor 110 may automatically perform specified operations required for extracting medical information without a user input. For example, after uploading the target image, the processor 110 may extract medical information and request a medical analysis without a user input.

[0106] After automatically or manually executing a specific operation, the processor 110 may display a specific icon 20 on the screen. For example, the specific icon 20 may be an icon 20 for executing the most frequently used operation after a specific operation. As a specific example, if requests for saving an area of interest from the user are frequent after extraction of medical information, the processor 110 may display an icon 20 for requesting saving of the area of interest on the screen after extraction of medical information.

[0107] FIG. 6 is an exemplary diagram of a screen in which the device 100 for extracting medical information provides the extracted medical information to the user.

[0108] The processor 110 may display the requested medical information on the screen in order to provide the medical information requested by the user.

[0109] Specifically, the processor 110 may display information related to the target image on the screen. For example, the information related to the target image may include the file name of the image, the upload time, the order of the query request, and the like.

[0110] Specifically, the processor 110 may display the result of the medical analysis of the target image on the screen. At this time, the result of the medical analysis may mean the result of the medical analysis analyzed based on the medical information using the learned model. Here, the result of the medical analysis may include at least one of the patient's name, the identification code of the patient number, and the diagnosis name.

[0111] FIG. 7 is another exemplary diagram of a screen in which the apparatus 100 for extracting medical information provides the extracted medical information to the user.

[0112] The processor 110 may transmit at least one of the information related to the target image and the result of the medical analysis to the server of the medical institution.

[0113] At this time, the processor 110 may transmit to the server of the medical institution without the user's consent only for the transmission request of the authenticated user.

[0114] FIG. 8 is a flowchart for explaining a method of providing a medical platform according to an embodiment.

[0115] Referring to FIG. 8, it may be performed by the apparatus 100 for extracting medical information according to an embodiment of FIG. 1.

[0116] First, the apparatus 100 for extracting medical information according to an embodiment detects (810) a user input for requesting extraction of medical information including the attribute data of the target image and the feature data of the target image from the target image.

[0117] After that, the apparatus 100 for extracting medical information according to one embodiment extracts attribute data of a target image in response to a user input, classifies the feature data of the target image, and extracts medical information (820).

[0118] After that, the apparatus 100 for extracting medical information according to one embodiment identifies (830) a region of interest in the target image corresponding to the target image and the medical information using a learned model.

[0119] On the other hand, an embodiment of the present invention may include a program for executing the method described in this specification on a computer and a computer-readable recording medium including the program. The computer-readable recording medium may include program instructions, local data files, local data structures, etc. alone or in combination. The medium may be specially designed and configured for the present invention, or may be one commonly used in the field of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and hardware devices specially configured to store and execute program instructions such as ROMs, RAMs, and flash memories 120. Examples of the program may include not only machine language code created by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

[0120] As described above, representative embodiments of the present invention have been described in detail. However, those having ordinary knowledge in the technical field to which the present invention pertains will understand that various modifications can be made to the above-described embodiments without departing from the scope of the present invention. Therefore, the scope of the rights of the present invention should not be determined by being limited to the described embodiments, but should be determined by not only the claims described later but also those equivalent to the claims.

Industrial Applicability

[0121] The apparatus and method for extracting medical information according to an embodiment can be used in the digital medical industry by executing a series of medical information extraction processes in one queue with a single user command.

Claims

1. One or more processors; and A memory for storing instructions for executing the one or more processors An apparatus for extracting medical information, comprising: The processor is: Detect a user input requesting extraction of medical information for the target image including the attribute data and the feature data of the target image, Extract the attribute data of the target image in response to the user input, classify the feature data of the target image, and extract the medical information, Identify a region of interest in the target image corresponding to the target image and the medical information using a trained model, An apparatus for extracting medical information.

2. The attribute data includes at least one of the resolution and the number of channels of the target image, The apparatus for extracting medical information according to claim 1, wherein the feature data includes a classification prediction value for at least one of the modality, the imaging direction, and the number of signal channels of the target image.

3. The processor is: Detect one or more candidate bounding boxes corresponding to each of the one or more objects included in the target image using a trained model, calculate a corresponding confidence level for each of the one or more candidate bounding boxes, and align the one or more candidate bounding boxes based on the confidence level. The apparatus for extracting medical information according to claim 1.

4. The processor is: Remove overlapping bounding boxes based on the confidence level of each of the one or more candidate bounding boxes, and identify a region of interest based on at least one of the remaining one or more candidate bounding boxes. The apparatus for extracting medical information according to claim 3.

5. The processor is: Select a reference bounding box based on the confidence level of each of the one or more candidate bounding boxes, remove overlapping bounding boxes from the one or more candidate bounding boxes based on the coincidence rate between the prediction region of the reference bounding box and the prediction regions of the remaining comparison target bounding boxes, and identify the region of interest based on the candidate bounding box having the highest confidence level among the remaining one or more candidate bounding boxes. The apparatus for extracting medical information according to claim 4.

6. The processor is: When detecting user input including at least one of predefined mouse gesture input, shortcut key input, icon input, and voice command, start extracting the medical information using a learned model, the apparatus for extracting medical information according to claim 1.

7. The processor: Specify the boundary of the region of interest based on a second input of the user who designates the region of interest from the target image, and manually extract the medical information based on a third input of the user who designates the attribute of the region of interest, the apparatus for extracting medical information according to claim 1.

8. The processor: Output a list of equipment including one or more medical equipment communicating with the apparatus for extracting the medical information, and transmit at least one of the target image, the region of interest, the attribute data, and the feature data to the medical equipment selected based on a fourth user input in the list of equipment, the apparatus for extracting medical information according to claim 1.

9. The processor: When receiving a fifth user input to drop the target image or the region of interest into a preset area, transmit to a model that matches the target image based on at least one of the attribute data and the feature data among one or more of the learned models, cause the target image to be processed by the matching model, and when the medical analysis is completed, output at least one of a visual notification signal and an auditory notification signal, the apparatus for extracting medical information according to claim 8.

10. The processor: Output a warning signal when at least one of the vertical resolution, horizontal resolution, and ratio of the vertical resolution to the horizontal resolution of the target image is less than or equal to a preset value, the apparatus for extracting medical information according to claim 1.

11. One or more processors; and A memory storing instruction words for executing the one or more processors An apparatus for extracting medical information including A method for extracting medical information performed by the apparatus for extracting medical information, comprising: The method includes: Detecting a user input requesting extraction of medical information for the target image including attribute data and feature data of the target image; Extracting the attribute data of the target image in response to the user input, classifying the feature data of the target image, and extracting the medical information; and A method for extracting medical information, comprising the step of identifying a region of interest in the target image corresponding to the target image and the medical information using a learned model.

12. The attribute data includes at least one of the resolution and the number of channels of the target image, The feature data includes a classification prediction value for at least one of the modality, imaging direction, and number of signal channels of the target image. The method for extracting medical information according to claim 11.

13. The step of extracting the medical information includes: Detecting one or more candidate bounding boxes corresponding to each of one or more objects included in the target image using a learned model; Calculating a corresponding confidence level for each of the one or more candidate bounding boxes; and Aligning the one or more candidate bounding boxes based on the confidence level. The method for extracting medical information according to claim 11.

14. The step of extracting the medical information includes: Removing overlapping bounding boxes based on the confidence level of each of the one or more candidate bounding boxes; and Identifying a region of interest based on at least one of the remaining one or more candidate bounding boxes. The method for extracting medical information according to claim 13.

15. The step of extracting the medical information includes: Selecting a reference bounding box based on the confidence level of each of the one or more candidate bounding boxes; Removing overlapping bounding boxes among the one or more candidate bounding boxes based on the coincidence rate between the prediction region of the reference bounding box and the prediction regions of the remaining comparison target bounding boxes; and Identifying the region of interest based on the remaining one or more candidate bounding boxes. The method for extracting medical information according to claim 14.

16. The step of extracting the medical information includes: When detecting user input including at least one of predefined mouse gesture input, shortcut key input, icon input, and voice command, starting the extraction of the medical information using a learned model. The method for extracting medical information according to claim 11.

17. The step of extracting the medical information includes: identifying a boundary of the region of interest based on a second input of a user who designates the region of interest from the target image; and the method for extracting medical information according to claim 11, further comprising manually extracting the medical information based on a third input of a user who designates an attribute of the region of interest.

18. The method includes: outputting a list of equipment including one or more medical devices communicating with the device for extracting the medical information; and transmitting at least one of the target image, the region of interest, the attribute data, and the feature data to a medical device selected based on a fourth user input from the list of equipment, the method for extracting medical information according to claim 11.

19. The step of transmitting includes when receiving a fifth user input of dropping the target image or the region of interest into a preset area, transmitting to a model that matches the target image based on at least one of the attribute data and the feature data among the one or more learned models; performing a medical analysis on the target image by the matching model; and when the medical analysis is completed, outputting at least one of a visual notification signal and an auditory notification signal, the method for extracting medical information according to claim 18.

20. The method includes further comprising outputting a warning signal when at least one of a vertical resolution, a horizontal resolution, and a ratio of the vertical resolution to the horizontal resolution of the target image is less than or equal to a preset value, the method for extracting medical information according to claim 11.

Images