WO2021075596A1 - System for processing medical images by using artificial intelligence-based curriculum learning method, system for intelligent medical diagnosis and treatment, and system for intelligent medical diagnosis and treatment by using block-based flexible ai model - Google Patents

Abstract

A system for processing medical images by using an artificial intelligence-based curriculum learning method according to the present invention comprises: an image acquisition unit for acquiring at least one image selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image; a potential disease input unit for inputting potential diseases; an image analysis unit for analyzing one of images acquired from the image acquisition unit, thereby extracting and analyzing at least one characteristic factor from among the position, length, size, and signal strength of a lesion, the image analysis unit analyzing a different image on the basis of the analyzed characteristic factor, re-extracting a characteristic factor of a lesion, and analyzing same; and a diagnosis result output unit for outputting at least one from among the area in which an analyzed lesion has occurred, the position thereof, the size thereof, the size of the lesion compared with the area in which same has occurred, and the signal strength thereof.

Description

A medical image processing system using an artificial intelligence-based curriculum learning method, an intelligent medical diagnosis and treatment system, and an intelligent medical diagnosis and treatment system using a block-based flexible AI model.

The present invention relates to a medical image processing system using an artificial intelligence-based curriculum learning method, an intelligent medical diagnosis and treatment system, and an intelligent medical diagnosis and treatment system using a block-based flexible AI model, and more specifically, an artificial intelligence-based curriculum learning method. Using a medical image processing system using an artificial intelligence-based curriculum learning method that can best detect lesions according to effective iterative learning, an intelligent medical diagnosis and treatment system, and a block-based flexible AI model are used to target specific diseases. It relates to an intelligent medical diagnosis and treatment system that can be applied to various diseases, not to the disease.

Currently, outpatient treatment in hospitals has a treatment system called '3-minute treatment' in general. In general, a university hospital sees about 4 patients every 15 minutes, and about 3 minutes of treatment time is given per patient. However, in reality, outpatient treatment hours are not enough time for patients, and are a limitation of not only Korea but also the global medical system.

In recent years, there is a growing movement to apply '15 minute treatment' in order to provide higher medical services to patients. '15-minute treatment' is a system that extends the treatment time, which is around 3 minutes on average, to close to 15 minutes so that patients can hear enough explanations about the disease, and is also called'in-depth examination'. This is an improvement plan that came out as patients' complaints increased, such as the patient's question about the disease did not go away even after the treatment due to the short treatment hours of Korean hospitals and clinics. In other words, the 15-minute treatment is aimed at improving the so-called 3-minute treatment, where the time to receive treatment from a doctor only stops within 3 minutes, and was derived from the results of a foreign study that the patient's understanding of the disease is high when the treatment is longer than 18 minutes. . The recent 15-minute treatment is being piloted at 19 large hospitals including Seoul National University Hospital, Severance Hospital, Seoul St. Mary's Hospital, Samsung Seoul Hospital, and Seoul Asan Hospital. The 15-minute treatment is to keep the treatment time at least 15 minutes, so that the doctor can make more accurate diagnosis and treatment, and from the patient's point of view, sufficient time and information regarding the disease can be secured.

From the point of view of the medical staff, the diagnosis results, the explanation of the disease and the treatment plan, etc., must be efficiently communicated to many patients during the prescribed treatment time, but the clinical medical staff are complaining that "there is not too much a way to explain to the patients." In fact, there is a very limited system for medical staff to explain to patients.

Currently, the best way to explain to most patients is to explain the current state through a medical image that has been examined. Patients and guardians want to know clearly the current condition of the patient and whether the disease can be improved through subsequent treatment plans, but the medical staff is sufficiently aware of these needs, but the situation is not resolved due to reasons such as treatment hours. to be.

When medical staff plan treatment and surgery, related departments gather and hold meetings from various angles about the patient. This is called multidisciplinary care, and the most important thing at this time is to review patients in the past similar to those in question together, minimize the likelihood of side effects, and establish a treatment plan that is the safest and fastest recoverable for the patient. In other words, the subjects of the past who are most similar to the patient in the past are gathered and discussed together, and a very effective treatment and treatment plan is created. However, the multidisciplinary treatment is not for general patients, but only for special patients, and it takes a lot of time and manpower to find similar patients in the past. In the end, unless in a special case, general outpatient treatment will eventually be made based on the experience of the medical staff.

Therefore, it can be applied to a variety of diseases, and results similar to multidisciplinary treatment can be obtained by providing the analysis results of medical images and clinical results and clinical, treatment, surgery and prognosis data of similar patients in advance prior to treatment by the medical staff in charge. However, there is a need to develop a system that enables efficient diagnosis and treatment within a short treatment time.

In addition, researches on medical image reading and processing systems using artificial intelligence have recently begun, but only the contents of detecting and analyzing lesions by analyzing captured images have been disclosed. Although the lesion is analyzed by comparative analysis of data, research on methods or systems for analyzing various types of images is lacking.

(Patent Document 1) Korean Registered Patent Publication No. 10-1623431

(Patent Document 2) Korean Patent Publication No. 10-1740464

(Patent Document 3) Korean Patent Application Publication No. 10-2018-0123810

The present invention is to solve the above problems, and an object of the present invention is to use a block-based flexible AI model, which can be applied to various diseases rather than targeting a specific disease. The purpose is to provide an intelligent medical diagnosis and treatment system capable of efficient diagnosis and treatment within a short treatment time by providing the analysis results of clinical results and the comparison with the lesions of similar patients, clinical, treatment, surgery and prognosis data in advance.

The present invention is to solve the above problems, and an object of the present invention is to use a block-based flexible AI model, which can be applied to various diseases rather than targeting a specific disease. The purpose is to provide an intelligent medical diagnosis and treatment system capable of efficient diagnosis and treatment within a short treatment time by providing the analysis results of clinical results and the comparison with the lesions of similar patients, clinical, treatment, surgery and prognosis data in advance.

In addition, it is an object of the present invention to analyze and learn from the image data that shows the best lesion characteristics according to the type of disease among a plurality of acquired images, and to store and transmit the analyzed and learned lesion information when analyzing the next image data. Thus, it is to provide a medical image processing system that can best detect lesions through effective repetitive learning using an artificial intelligence-based curriculum learning method that maximizes learning efficiency so that a target disease can be diagnosed more accurately.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned will be clearly understood by those skilled in the art from the following description.

According to an embodiment of the present invention for solving the above problem, an intelligent medical diagnosis and treatment system based on medical image data is provided. The intelligent medical diagnosis and treatment system includes an image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image, a suspect disease input unit that inputs a suspect disease, and a block-based artificial intelligence model. According to the suspicious disease input by using, the lesion is mapped by setting each analysis area of the lesion in the input image, and extracting and analyzing any one or more feature factors from the location, length, size, and signal intensity of the mapped lesion. An image analysis unit, a medical integrated database unit storing and providing disease-specific image factors, disease-specific clinical factors, disease-specific treatment and surgery or prognosis data in which feature factors extracted from images are expressed, and disease-specific data provided from the medical integrated database unit It provides an image factor and a clinical factor in a text format, and includes a diagnosis result output unit that provides a tracking result for the user's past and current state.

In addition, according to an embodiment of the present invention, there is provided an intelligent medical diagnosis and treatment system, wherein the suspected disease is a neurological disease.

In addition, according to an embodiment of the present invention, the neurological disease is an intelligent medical diagnosis and treatment system, characterized in that at least one selected from the group consisting of cerebral infarction, arteriosclerosis, cerebral aneurysm, Alzheimer's dementia, compression fracture, and disc. to provide.

In addition, according to an embodiment of the present invention, the MRI image is a T2 weighted image (T2), a fluid attenuated inversion recovery (FLAIR) image, a diffusion weighted image (DWI), Intelligent medical diagnosis comprising perfusion weighted imaging (PWI), magnetic resonance angiography (MRA) images, and high resolution T1 weighted 3D images (3DT1) and Provide a medical treatment system.

In addition, according to an embodiment of the present invention, the image analysis unit selects and analyzes an input image type according to an input suspicious disease.

In addition, when the input suspected disease is cerebral infarction, CT, T2-weighted image, fluid attenuation reversal image, diffusion-weighted image, and perfusion-weighted image are images to be analyzed, and when the input suspected disease is arteriosclerosis, CT and magnetic resonance An angiographic image is used as the image to be analyzed, and if the input suspected disease is a cerebral aneurysm, CT and magnetic resonance angiography images are used as the image to be analyzed, and if the input suspected disease is Alzheimer's dementia, a fluid attenuation inversion image, 3DT1 An intelligent medical diagnosis, characterized in that an emphasis image and a positron emission tomography image (PET) are used as an image to be analyzed, and when the input suspected disease is a compression fracture or a disc, an X-ray image is used as an image to be analyzed. And a medical treatment system.

In addition, according to an embodiment of the present invention, one or more clinical factors selected from the group consisting of a user's risk factor test, a clinical disease evaluation test, and a blood factor test are evaluated and provided to the integrated medical database. Provides an intelligent medical diagnosis and treatment system.

In addition, according to an embodiment of the present invention, in the diagnosis result output unit, an intelligent medical diagnosis and treatment system, characterized in that the condition for treatment and prognosis according to the user's tracking result is simultaneously displayed and provided to the user. to provide.

In addition, according to an embodiment of the present invention, by using the image factor for each disease as a search index, by searching for an image factor similar to the disease, clinical factors, treatment, surgery, and prognosis data of the similar image factor disease are obtained as a result of the diagnosis. It provides an intelligent medical diagnosis and treatment system comprising a clinical case search unit provided to the output unit.

In addition, according to an embodiment of the present invention, there is provided an intelligent medical diagnosis and treatment system, characterized in that the difference between the image factor for each disease and the searched image factor for a similar disease is provided.

According to an embodiment of the present invention for solving the above problem, an image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image, and a suspected disease inputting a suspected disease An input unit and an image obtained from the image acquisition unit are analyzed to extract and analyze any one or more of the location, length, size, and signal strength of the lesion, and other images based on the analyzed feature factors. An image analysis unit that analyzes and re-extracts and analyzes the characteristic factors of the lesion, and a diagnosis result output unit that outputs at least one of the occurrence region, location, size, size of the lesion relative to the occurrence region, and signal intensity of the analyzed lesion. A medical image processing system using an artificial intelligence-based curriculum learning method is provided.

In addition, according to an embodiment of the present invention, the image analysis unit recognizes and divides each image obtained from the image acquisition unit by using a reference image for each image, and a characteristic factor of the lesion for each segmented area Provides a medical image processing system using an artificial intelligence-based curriculum learning method characterized by extracting and analyzing.

In addition, according to an embodiment of the present invention, the image analysis unit analyzes the feature factor by selecting an image that facilitates lesion identification among the acquired images, and then selects another image to analyze the feature factor. At the end, it provides a medical image processing system using an artificial intelligence-based curriculum learning method, characterized in that the feature factor is further analyzed by selecting an image that is easy to confirm the lesion first analyzed.

In addition, according to an embodiment of the present invention, the MRI image is a diffusion weighted image (DWI), an apparent diffusion coefficient map (ADC map), and a fluid attenuated inversion recovery (FLAIR) image. , T2 weighted image (T2), Perfusion Weighted Imaging (PWI), Magnetic Resonance Angiography (MRA) image, and high resolution T1 Weighted 3D Image, 3DT1 It provides a medical image processing system using an artificial intelligence-based curriculum learning method, characterized in that it includes any one or more selected from the group consisting of ).

In addition, according to an embodiment of the present invention, when the suspected disease is a cerebral infarction, the image analysis unit first selects the diffusion-enhanced image (DWI) for easy identification of a cerebral infarct lesion among acquired images, After analyzing the factor, the ADC map is selected to analyze the characteristic factor, the fluid attenuation inversion (FLAIR) image is selected to analyze the characteristic factor, and the CT image is then selected to analyze the characteristic factor. And, it provides a medical image processing system using an artificial intelligence-based curriculum learning method, characterized in that the feature factor is analyzed by reselecting the firstly selected diffusion-weighted image (DWI).

In addition, according to an embodiment of the present invention for solving the above problem, an intelligent medical diagnosis and treatment system using an artificial intelligence-based curriculum learning method is provided. The intelligent medical diagnosis and treatment system includes an image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image, a suspect disease input unit that inputs a suspect disease, and a block-based artificial intelligence model. By analyzing any one of the input images according to the suspicious disease input by using, extract and analyze any one or more of the location, length, size, and signal strength of the lesion, and based on the analyzed feature factors. An image analysis unit that analyzes other images and re-extracts and analyzes the characteristic factors of the lesion, image factors for each disease, clinical factors for each disease, treatment and surgery or prognosis data for each disease, and And a diagnosis result output unit that provides an integrated medical database unit and image factors and clinical factors for each disease provided from the integrated medical database unit in a text format, and provides a tracking result of a user's past and present conditions.

In addition, according to an embodiment of the present invention, the suspected disease is an intelligent medical diagnosis and treatment system, characterized in that at least one selected from the group consisting of cerebral infarction, arteriosclerosis, cerebral aneurysm, Alzheimer's dementia, compression fracture, and disc. to provide.

In addition, according to an embodiment of the present invention, one or more clinical factors selected from the group consisting of a user's risk factor test, a clinical disease evaluation test, and a blood factor test are evaluated and provided to the integrated medical database. Provides an intelligent medical diagnosis and treatment system.

In addition, according to an embodiment of the present invention, in the diagnosis result output unit, an intelligent medical diagnosis and treatment system, characterized in that the condition for treatment and prognosis according to the user's tracking result is simultaneously displayed and provided to the user. to provide.

In addition, according to an embodiment of the present invention, the disease-specific image factor is used as a search index, and the image factor similar to the disease is searched to provide clinical factors, treatment, surgery, and prognosis data of the disease-like image factor, and the disease. It provides an intelligent medical diagnosis and treatment system comprising a clinical case search unit that provides a difference between the respective image factor and the searched similar disease image factor to the diagnosis result output unit.

According to the present invention, in an intelligent medical diagnosis and treatment system, it is possible to apply a block-based flexible AI model to various diseases rather than targeting a specific disease.

In addition, according to the present invention, prior to treatment by the medical staff in charge, various images and clinical results of the lesions of the user, which are the user, are analyzed to provide the analysis results to the medical staff, and feature factors of the analyzed images are extracted to By searching for data, it is possible to analyze and provide differences between similar patients and clinical, treatment, surgery and prognosis data of similar patients before treatment, enabling efficient diagnosis and treatment within a short treatment time.

According to the present invention, in an intelligent medical diagnosis and treatment system, it is possible to apply a block-based flexible AI model to various diseases rather than targeting a specific disease.

In addition, according to the present invention, prior to treatment by the medical staff in charge, various images and clinical results of the lesions of the user, which are the user, are analyzed to provide the analysis results to the medical staff, and feature factors of the analyzed images are extracted to By searching for data, it is possible to analyze and provide differences between similar patients and clinical, treatment, surgery and prognosis data of similar patients before treatment, enabling efficient diagnosis and treatment within a short treatment time.

In addition, according to the present invention, among a plurality of images acquired using artificial intelligence techniques, the image data that shows the most characteristic of the lesion according to the type of disease are analyzed and learned, and the analyzed and learned lesion information is converted to the next image data. By using the curriculum learning method based on artificial intelligence that maximizes the learning efficiency so that the target disease can be diagnosed more accurately by accumulating and transmitting during the analysis of, lesions can best be found through effective iterative learning.

1 is a flowchart illustrating a conventional medical diagnosis and treatment system.

FIG. 2 is a block diagram illustrating a medical image processing system using an artificial intelligence-based curriculum learning method, an intelligent medical diagnosis and treatment system, and an intelligent medical diagnosis and treatment system using a block-based flexible AI model according to an embodiment of the present invention. to be.

3 is a block diagram illustrating a medical image processing system using an artificial intelligence-based curriculum learning method according to an embodiment of the present invention.

4 is a block diagram illustrating an image analysis unit that extracts and analyzes a feature factor of an image by applying a block-based flexible artificial intelligence model according to an embodiment of the present invention.

According to an embodiment of the present invention, an intelligent medical diagnosis and treatment system based on medical image data is provided. The intelligent medical diagnosis and treatment system includes an image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image, a suspect disease input unit that inputs a suspect disease, and a block-based artificial intelligence model. According to the suspicious disease input by using, the lesion is mapped by setting each analysis area of the lesion in the input image, and extracting and analyzing any one or more feature factors from the location, length, size, and signal intensity of the mapped lesion. An image analysis unit, a medical integrated database unit storing and providing disease-specific image factors, disease-specific clinical factors, disease-specific treatment and surgery or prognosis data in which feature factors extracted from images are expressed, and disease-specific data provided from the medical integrated database unit It provides an image factor and a clinical factor in a text format, and includes a diagnosis result output unit that provides a tracking result for the user's past and current state.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments to be posted below, but may be implemented in a variety of different forms, and only these embodiments make the posting of the present invention complete, and common knowledge in the technical field to which the present invention pertains. It is provided to completely inform the scope of the invention to the possessor, and the invention is only defined by the scope of the claims. The same reference numerals refer to the same elements throughout the specification.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used with meanings that can be commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not interpreted ideally or excessively unless explicitly defined specifically. The terms used in the present specification are for describing exemplary embodiments and are not intended to limit the present invention. In this specification, the singular form also includes the plural form unless specifically stated in the phrase.

In addition, the terms "comprise" and/or "comprise" as used herein specify the presence of the mentioned shapes, numbers, steps, members, elements and or groups thereof, and are not mentioned. It is not intended to exclude the presence or addition of other shapes, numbers, actions, members, elements and/or groups.

In addition, a structure or shape arranged adjacent to another shape may have a portion disposed below or overlapping with the adjacent shape.

Relative terms such as below, above, upper, lower, horizontal or vertical in this specification are as shown in the figures, It may be used to describe the relationship one constituent member, layer, or region has with another constituent member, layer or region. These terms encompass not only the orientation indicated in the figures, but also other orientations of the device.

In the following, embodiments of the present invention are described with reference to cross-sectional views schematically showing ideal embodiments (and intermediate structures) of the present invention. In these drawings, for example, the size and shape of the members may be exaggerated for convenience and clarity of description, and in actual implementation, variations of the illustrated shape may be expected. Accordingly, embodiments of the present invention are not limited to the specific shape of the region shown in the present specification.

1 is a flowchart illustrating a conventional medical diagnosis and treatment system.

Taking the general outpatient treatment of the current university hospital, for example, first, after completing the tests such as necessary medical imaging tests and clinical symptom tests of the patient in advance, the doctor enters the clinic to hear the test results. Open the chart at this time and check the patient's condition.

In general, the patient may think that the doctor is aware of the test result and treatment plan in advance, but in reality, it is checked at the time of the patient's treatment. After that, the doctor explains the disease and treatment plan to the patient based on the test results. At this time, the test results are communicated orally with the clinical results or imaging test results, and if more detailed explanation is determined, the explanation is made through pictures on a memo pad. If the patient's condition deteriorates in the case of follow-up care, the doctor in charge should find the cause and change to an appropriate treatment method.

In conclusion, the medical staff in charge cannot provide a satisfactory explanation within a short treatment time, and the patient does not receive a satisfactory explanation within a short treatment time.

FIG. 2 is a block diagram illustrating a medical image processing system using an artificial intelligence-based curriculum learning method, an intelligent medical diagnosis and treatment system, and an intelligent medical diagnosis and treatment system using a block-based flexible AI model according to an embodiment of the present invention. to be. 3 is a block diagram illustrating a medical image processing system using an artificial intelligence-based curriculum learning method according to an embodiment of the present invention. 4 is a block diagram illustrating an image analysis unit that extracts and analyzes a feature factor of an image by applying a block-based flexible artificial intelligence model according to an embodiment of the present invention.

With reference to FIGS. 2 to 4, a medical image processing system using an artificial intelligence-based curriculum learning method, an intelligent medical diagnosis and treatment system, and an intelligent medical diagnosis and treatment system using a block-based flexible AI model will be described in detail.

The medical diagnosis and treatment system of the present invention is an intelligent medical diagnosis and treatment system based on medical image data, and includes an image acquisition unit 100, a suspected disease input unit 200, an image analysis unit 300, and a medical integrated database unit 400. And a diagnosis result output unit 500.

The image acquisition unit 100 acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image.

The suspected disease input unit 200 inputs a suspected disease. In the medical diagnosis and treatment system according to an embodiment of the present invention, the disease to be analyzed may specifically target a neurological disease. For example, the neurological disease may be any one or more selected from the group consisting of cerebral infarction, arteriosclerosis, cerebral aneurysm, Alzheimer's dementia, compression fracture, and disc.

When the disease to be analyzed is cerebral infarction, arteriosclerosis, cerebral aneurysm, or Alzheimer's dementia, the image acquired by the image acquisition unit 100 may be a CT image, an MRI image or a PET image, and the disease to be analyzed is a compression fracture or a disc. In the case of, the image acquired by the image acquisition unit 100 may be an X-ray image.

The image analysis unit 300 maps the lesions by respectively setting analysis regions of lesions in the input image according to the input suspicious disease using a block-based artificial intelligence model. The image analysis unit 300 extracts and analyzes one or more feature factors from among the location, length, size, and signal intensity of the mapped lesion.

The neurological disease may be any one or more selected from the group consisting of cerebral infarction, arteriosclerosis, cerebral aneurysm, Alzheimer's dementia, compression fracture, and disc.

2 and 4, the image analysis unit 300 selects and analyzes an input image type according to an input suspicious disease.

The image analysis unit 300 can be applied to various diseases rather than targeting a specific disease by using a block-based flexible AI model, and has excellent scalability.

By using the block-based flexible artificial intelligence model, the image analysis unit 300 recommends an optimal learning method according to the medical environment and purpose if only data on the disease is inputted, and a block-type algorithm combination technique Anyone can easily create a flexible artificial intelligence model by using. Since it is not a complete system for specific diseases, the expansion of additional diseases is very flexible and its versatility is very good. In addition, since all analysis is performed inside the medical institution, not outside the leak, there is no security problem of medical data.

In the medical diagnosis and treatment system of the present invention, the image analysis unit 300 selects and analyzes an input image type according to the input suspicious disease. For example, if the input suspicious disease is a cerebral infarction, the CT, The T2-weighted image, the fluid attenuated inversion image, the diffusion-weighted image, and the perfusion-weighted image can be used as an analysis target image.

For example, when the input suspicious disease is arteriosclerosis, CT and magnetic resonance angiography images may be used as an image to be analyzed. For example, when the input suspicious disease is a cerebral aneurysm, CT and magnetic resonance angiography images may be used as an image to be analyzed. For example, when the input suspicious disease is Alzheimer's dementia, a fluid attenuation inversion image, a 3DT1 weighted image, and a positron emission tomography image (PET) may be used as an image to be analyzed. For example, when the input suspicious disease is a compression fracture or a disc, an X-ray image may be used as an image to be analyzed.

The image analysis unit 300 analyzes any one of the images acquired from the image acquisition unit 100 to extract and analyze any one or more of the location, length, size, and signal strength of the lesion, and then Another image is analyzed based on the feature factors sequentially analyzed in, and the feature factors of the lesion are re-extracted and analyzed.

That is, the image first selected and analyzed by the image analysis unit 300 is learned by selecting an image technique in which a lesion is best identified according to the suspicious disease, and accordingly, learning is performed from the lowest learning difficulty. You can maximize the learning efficiency by proceeding.

Thereafter, an imaging technique with a lower learning difficulty level is selected and learning is performed sequentially, and the characteristics of the lesion identified according to the previously analyzed imaging technique are transmitted during the subsequent analysis, so that the target disease can be diagnosed more accurately. .

The last image analyzed by the image analysis unit 300 is selected for the first time, and the analyzed image is reselected to perform analysis once more. This is to minimize the loss of feature learning ability and data in the AI model by learning the imaging technique that can best diagnose the target disease once more, and learn and remember the most important features of the target disease lesion once more. It is to do. That is, it is possible to minimize a problem in which analysis results may be deformed or distorted according to repetitive learning.

For example, the MRI image is a diffusion weighted image (DWI), an apparent diffusion coefficient map (ADC map), a fluid attenuated inversion recovery (FLAIR) image, and a T2 weighted image (T2 weighted image). image, T2), Perfusion Weighted Imaging (PWI), Magnetic Resonance Angiography (MRA) image, and 3D T1 weighted image (high resolution T1 Weighted 3D Image, 3DT1). It may contain more than one.

For example, referring to FIG. 3, when the suspected disease is a cerebral infarction, the image analysis unit 300 is, first among the acquired images, the diffusion-weighted image (DWI) for easy identification of a cerebral infarction lesion. First, the feature factor is analyzed by selecting first, and then, secondly, the ADC map is selected to analyze the feature factor, and thirdly, the fluid attenuation inversion (FLAIR) image is selected and the feature factor is analyzed, Thereafter, fourthly, the CT image is selected to analyze the feature factor, and fifthly, the firstly selected diffusion-weighted image (DWI) may be reselected to analyze the feature factor.

(1) After selecting the DWI image for the first time, through “learned AI model #A”, (2) the ADC map selected afterwards is accumulated knowledge based on “learned AI model #A” and transferred to “learned AI model #A”. AI model #B" is formed, and the FLAIR video selected after (3) is accumulated knowledge based on "learned AI model #B" and transferred to form "learned AI model #C", and after (4) The selected CT image consists of “learned AI model #D” by accumulating knowledge based on “learned AI model #C”, and (5) the selected CT image is “learned AI model #D”. Knowledge is accumulated and delivered as a basis to form “learned AI model #E”.

“Learned AI model #A” is a model in which lesion features of DWI images are learned, inherits (transfers) the features of learned lesions seen in DWI images, and “Learned AI model #B” includes DWI images and ADC. The lesion features of the map are accumulated and learned. In "Learned AI Model #C", lesion features of DWI image, ADC map and FLAIR image are accumulated and learned. In "Learned AI Model #D", lesion features of DWI image, ADC map, FLAIR image and CT image are accumulated and learned.

Finally, in the “learned AI model #E”, in addition to the lesion characteristics of the accumulated and learned DWI image, ADC map, FLAIR image, and CT image, the DWI image that the characteristics of the cerebral infarct lesion is best confirmed is learned once more to be more accurate. The characteristics of the lesion can be analyzed.

The integrated medical database unit 400 stores and provides disease-specific image factors, disease-specific clinical factors, disease-specific treatment and surgery or prognosis data, in which characteristic factors extracted from images are expressed.

Currently, the medical database system is managed by PACS for images and EMR for clinical evaluation results for patients. Such a system cannot match specific symptoms, test results, and diseases of the patient, and thus, it is possible to check the test results, but it is impossible to search for similar diseases and patients.

In order to overcome these limitations, the intelligent medical diagnosis and treatment system according to an embodiment of the present invention converts the medical image into an index through artificial intelligence techniques, and uses a database through all data on clinical factors, treatment methods, and results matching the medical image. It is converted and stored in the medical integration database unit 400.

Therefore, it is possible to provide a reading system that quantitatively reflects the test result and reflects the continuity for the subject of tracking management.

The integrated medical database unit 400 stores data accumulated while providing services according to the intelligent medical diagnosis and treatment system according to an embodiment of the present invention, medical image data for each user, and medical care of similar users (patients). Image data, diagnostic information, clinical data, visual information and text data of a lesion area, tracking result data for a user's past and current state, treatment of a corresponding disease, surgery and prognosis data, and the like may be stored.

The diagnosis result output unit 500 provides disease-specific image factors and clinical factors provided from the medical integrated database unit 400 in text format, and provides tracking results of the user's past and present conditions.

The diagnosis result output unit 500 may display the image factors and clinical factors for each disease in a text format and provide them to a user and a medical staff.

In addition, it is possible to simultaneously display the results of tracking the user's past and present conditions, as well as treatment and prognosis according to the results, and provide them to the user.

The medical diagnosis and treatment system of the present invention is an intelligent medical diagnosis and treatment system based on medical image data, and includes an image acquisition unit 100, a suspected disease input unit 200, an image analysis unit 300, and a medical integrated database unit 400. And a diagnosis result output unit 500, and further includes a clinical factor input unit 600, a disease-specific treatment input unit 700, and a clinical case search unit 800.

The clinical factor input unit 600 evaluates one or more clinical factors selected from a group consisting of a risk factor test, a clinical disease evaluation test, and a blood factor test of a user, and provides and stores it in the medical integrated database unit 400. The integrated medical database unit 400 stores clinical factor data for each disease accumulated while providing services according to the intelligent medical diagnosis and treatment system according to an embodiment of the present invention.

The disease-specific treatment input unit 700 inputs disease-specific treatment, surgery, and prognosis data, and provides and stores it in the medical integrated database unit 400. The integrated medical database unit 400 stores treatment, surgery, and prognosis data for each disease accumulated while providing services according to the intelligent medical diagnosis and treatment system according to an embodiment of the present invention.

The clinical case search unit 800 uses the disease-specific image factor as a search index, searches for an image factor similar to the disease, and displays the clinical factor, treatment, surgery, and prognosis data of the disease-like image factor, and outputs the diagnosis result. Can be provided to.

In addition, the clinical case search unit 800 may provide a difference between the image factor for each disease and the searched image factor for a similar disease.

The clinical case search unit 800 interlocks with the medical integration database unit 400 to search and classify the data of the medical integration database unit 400, Provides data, visual information and text data of the lesion area, tracking result data for the user's past and current state, treatment of the disease, surgery and prognosis data, and the like, and the diagnosis result output unit 500, that is, the medical staff in charge Updates to automatic readings can be provided.

Accordingly, according to the intelligent medical diagnosis and treatment system of the present invention, it is possible to apply a block-based flexible AI model to various diseases rather than targeting a specific disease.

In addition, prior to treatment by the medical staff in charge, various images and clinical results of the patient's lesions are analyzed and the analysis results are provided to the medical staff, and feature factors of the analyzed images are extracted to search for similar patient data. It has the advantage of enabling efficient diagnosis and treatment within a short treatment time by analyzing and providing differences from patients and clinical, treatment, surgery and prognostic data of similar patients before treatment.

The foregoing description of the present application is for illustrative purposes only, and those of ordinary skill in the art to which the present application pertains will be able to understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and are not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present application is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present application.

Claims (15)

An image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image; A suspected disease input unit for inputting a suspected disease; Analyze any one of the images acquired from the image acquisition unit to extract and analyze any one or more of the location, length, size, and signal strength of the lesion, and analyze other images based on the analyzed feature factors. An image analysis unit that re-extracts and analyzes the characteristic factors of the lesion; And A medical image processing system using an artificial intelligence-based curriculum learning method comprising a diagnosis result output unit that outputs one or more of an occurrence region, a location, a size of the analyzed lesion, a size of a lesion compared to the occurrence region, and a signal strength. The method of claim 1, The image analysis unit, A curriculum learning method based on artificial intelligence, characterized in that each image acquired from the image acquisition unit is segmented by recognizing and segmenting major brain regions using a reference image for each image, and extracting and analyzing characteristic factors of lesions for each segmented region. Used medical image processing system. The method of claim 1, The image analysis unit analyzes the feature factor by selecting an image that is easy to identify a lesion among the acquired images, then selects another image to analyze the feature factor, and finally, it is easy to confirm the lesion first analyzed. A medical image processing system using an artificial intelligence-based curriculum learning method, characterized in that one image is selected and the feature factor is further analyzed. The method of claim 3, The MRI image is a diffusion weighted image (DWI), an apparent diffusion coefficient map (ADC map), a fluid attenuated inversion recovery (FLAIR) image, and a T2 weighted image (T2). , Perfusion Weighted Imaging (PWI), Magnetic Resonance Angiography (MRA) image, and 3D T1 weighted image (high resolution T1 Weighted 3D Image, 3DT1). Medical image processing system using an artificial intelligence-based curriculum learning method, characterized in that. The method of claim 4, When the suspected disease is a cerebral infarction, the image analysis unit first selects the diffusion-enhanced image (DWI) for easy identification of a cerebral infarct lesion among acquired images, analyzes the feature factor, and then selects the ADC map. To analyze the feature factor, then select the fluid attenuation inversion (FLAIR) image to analyze the feature factor, then select the CT image to analyze the feature factor, and finally, the first selected diffusion-weighted image ( A medical image processing system using an artificial intelligence-based curriculum learning method, characterized in that the characteristic factor is analyzed by reselecting (DWI). In an intelligent medical diagnosis and treatment system using an artificial intelligence-based curriculum learning method, An image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image; A suspected disease input unit for inputting a suspected disease; Analyzes any one of the input images according to the suspicious disease input using a block-based artificial intelligence model, extracts and analyzes any one or more of the location, length, size, and signal strength of the lesion, and analyzes An image analysis unit that analyzes another image based on the characteristic factor and re-extracts and analyzes the characteristic factor of the lesion; A medical integrated database unit for storing and providing disease-specific image factors, disease-specific clinical factors, disease-specific treatment and surgery or prognosis data in which feature factors extracted from images are expressed; And An intelligent medical diagnosis and treatment system including a diagnosis result output unit that provides image factors and clinical factors for each disease provided from the integrated medical database unit in text form, and provides tracking results for a user's past and present conditions. In the intelligent medical diagnosis and treatment system based on medical image data, An image acquisition unit that acquires one or more images selected from the group consisting of an X-ray image, a CT image, an MRI image, and a PET image; A suspected disease input unit for inputting a suspected disease; Using a block-based artificial intelligence model, each lesion is mapped by setting an analysis area of the lesion in the input image according to the input suspicious disease, and any one or more of the location, length, size, and signal intensity of the mapped lesion An image analysis unit that extracts and analyzes a factor; A medical integrated database unit for storing and providing disease-specific image factors, disease-specific clinical factors, disease-specific treatment and surgery or prognosis data in which feature factors extracted from images are expressed; And An intelligent medical diagnosis and treatment system including a diagnosis result output unit that provides image factors and clinical factors for each disease provided from the integrated medical database unit in text form, and provides tracking results for a user's past and present conditions. The method of claim 7, Intelligent medical diagnosis and treatment system, characterized in that the suspected disease is a neurological disease. The method of claim 8, The neurological disease is an intelligent medical diagnosis and treatment system, characterized in that at least one selected from the group consisting of cerebral infarction, arteriosclerosis, cerebral aneurysm, Alzheimer's dementia, compression fracture, and disc. The method of claim 9, The MRI image is a T2 weighted image (T2), a fluid attenuated inversion recovery (FLAIR) image, a diffusion weighted image (DWI), a perfusion weighted image (PWI), An intelligent medical diagnosis and treatment system comprising a magnetic resonance angiography (MRA) image and a 3D T1 weighted image (high resolution T1 Weighted 3D Image, 3DT1). The method of claim 10, The image analysis unit selects and analyzes the input image type according to the input suspicious disease, If the input suspected disease is cerebral infarction, CT, T2-weighted image, fluid attenuation reversal image, diffusion-weighted image, and perfusion-weighted image are the images to be analyzed, and if the input suspected disease is arteriosclerosis, CT and magnetic resonance angiography The image is used as the image to be analyzed, and if the input suspected disease is a cerebral aneurysm, CT and magnetic resonance angiography images are used as the image to be analyzed. And a positron emission tomography image (PET) as an image to be analyzed, and when the input suspicious disease is a compression fracture or a disc, an X-ray image is used as an image to be analyzed. system. The method of claim 7, An intelligent medical diagnosis and treatment system, characterized in that one or more clinical factors selected from a group consisting of a user's risk factor test, a clinical disease evaluation test, and a blood factor test are evaluated and provided to the integrated medical database. The method of claim 7, An intelligent medical diagnosis and treatment system, characterized in that the diagnosis result output unit simultaneously displays and provides the user with a state of treatment and prognosis according to the user's tracking result. The method of claim 7, Including a clinical case search unit that uses the disease-specific image factor as a search index, searches for image factors similar to the disease, and provides clinical factor, treatment, surgery, and prognosis data of the similar image factor disease to the diagnosis result output unit. Intelligent medical diagnosis and treatment system, characterized in that. The method of claim 14, An intelligent medical diagnosis and treatment system, characterized in that providing a difference between the disease-specific image factor and the searched similar disease image factor.

Images