WO2023101444A1 - Appareil et procédé de prise en charge de tumeur sur la base d'un modèle d'intelligence artificielle - Google Patents

Appareil et procédé de prise en charge de tumeur sur la base d'un modèle d'intelligence artificielle Download PDF

Info

Publication number
WO2023101444A1
WO2023101444A1 PCT/KR2022/019271 KR2022019271W WO2023101444A1 WO 2023101444 A1 WO2023101444 A1 WO 2023101444A1 KR 2022019271 W KR2022019271 W KR 2022019271W WO 2023101444 A1 WO2023101444 A1 WO 2023101444A1
Authority
WO
WIPO (PCT)
Prior art keywords
tumor
data
artificial intelligence
information
group
Prior art date
Application number
PCT/KR2022/019271
Other languages
English (en)
Korean (ko)
Inventor
박성걸
차병래
백성은
박소라
임선화
Original Assignee
(주)플라스바이오
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by (주)플라스바이오 filed Critical (주)플라스바이오
Publication of WO2023101444A1 publication Critical patent/WO2023101444A1/fr

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J11/00—Manipulators not otherwise provided for
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B45/00—ICT specially adapted for bioinformatics-related data visualisation, e.g. displaying of maps or networks
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00—ICT specially adapted for the handling or processing of medical images
    • G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the present disclosure relates to an apparatus for managing a tumor based on an artificial intelligence model and a method of operating the same. More specifically, it relates to a device and an operating method for measuring and managing the maturity of a tumor using machine learning.
  • the method of producing CDX, an existing cancer cell line transplantation model, is performed by a skilled animal researcher, and tumor cells are injected subcutaneously or intravascularly into an animal to check the size or volume of the tumor.
  • the CDX production method which is a conventional cancer cell line transplantation model, measures the size using vernier calipers or images, and since the maturity of tumors varies from individual to individual, the size of the tumor is determined when the test substance is administered, and the size is mainly used. Since the measurement method was used with a seawater instrument, there was a limit that could cause errors depending on the measurer.
  • an automatic administration device and an operation method thereof for analyzing a tumor and treating the tumor based on the analysis result may be provided.
  • an automatic administration device for managing a tumor based on an artificial intelligence model and an operation method thereof may be provided.
  • a method for an automatic administration device to manage a tumor based on an artificial intelligence model comprising: acquiring a tumor image of a tumor to be treated; obtaining maturity information from the artificial intelligence model by inputting the tumor image to an artificial intelligence model that outputs maturity information of a tumor to be treated included in the tumor image when the tumor image is input; determining drug administration information based on the maturity level information; and outputting the determined drug administration information.
  • a method may be provided.
  • an automatic administration device for managing a tumor based on an artificial intelligence model comprising: a display; camera; a robot arm that administers a drug for treating the tumor; and an edge computer processing image data of the tumor in real time based on a pre-learned weight file.
  • the edge computer includes a memory for storing one or more instructions; and at least one processor executing the one or more instructions; Including, the at least one processor acquires a tumor image of a tumor to be treated by executing the one or more instructions, and when the tumor image is input, the tumor to be treated included in the tumor image Obtaining maturity information from the artificial intelligence model by inputting the tumor image to an artificial intelligence model that outputs maturity information, determining drug administration information based on the maturity information, and outputting the determined drug administration information, An automatic dosing device may be provided.
  • a method for an automatic administration device to manage a tumor based on an artificial intelligence model comprising: acquiring a tumor image of a tumor to be treated; obtaining maturity information from the artificial intelligence model by inputting the tumor image to an artificial intelligence model that outputs maturity information of a tumor to be treated included in the tumor image when the tumor image is input; determining drug administration information based on the maturity level information; and outputting the determined drug administration information.
  • a computer-readable recording medium recording a program for executing the method on a computer, including, may be provided.
  • the automatic administration device can optimize the maturity of the tumor, the size of the tumor, and the timing of administering the therapeutic drug through image processing.
  • data on the tumor administration location and administration method can be constructed.
  • tumor recognition and maturity can be optimized.
  • an optimized dataset for tumor recognition improvement can be constructed.
  • it can be applied to various cancer cell line disease models and can be applied to various experiments.
  • FIG. 1 is a diagram for explaining a schematic process of managing a tumor by an automatic administration device according to an embodiment.
  • FIG. 2 is a flowchart of a method of managing a tumor based on an artificial intelligence model by an automatic administration device according to an embodiment.
  • FIG. 3 is a flowchart of a method for managing a tumor based on an artificial intelligence model by an automatic administration device according to another embodiment.
  • FIG. 4 is a diagram for explaining a process of constructing a tumor data set by an automatic administration device according to an embodiment.
  • FIG. 5 is a diagram for explaining a process of constructing a tumor data set by an automatic administration device according to an embodiment.
  • FIG. 6 is a diagram for explaining a process of analyzing a tumor data set by an automatic administration device according to an embodiment.
  • FIG. 7 is a diagram for explaining an artificial intelligence model used by an automatic administration device according to an embodiment.
  • FIG. 8 is a diagram for explaining a process of selecting one of a plurality of artificial intelligence models learned by an automatic administration device according to an embodiment.
  • FIG. 9 is a diagram for explaining a process of implementing an artificial intelligence model used by an automatic administration device for tumor detection according to an embodiment.
  • 10 is a diagram for explaining the accuracy of learning data used for learning an artificial intelligence model used by an automatic administration device according to an embodiment.
  • FIG. 11 is a block diagram of an automatic administration device according to an embodiment.
  • FIG. 1 is a diagram for explaining a schematic process of managing a tumor by an automatic administration device according to an embodiment.
  • Figures 102 and 104 of FIG. 1 show an embodiment in which the automatic administration device 1000 observes the tumor to be treated from the side and an embodiment in which the automatic administration device 1000 observes the tumor to be treated from the top.
  • the automatic administration device 1000 can optimize the standard size of a tumor model by learning the maturity level of cancer cells using machine learning, and also can optimize the standardized tumor model to determine the maturity and size of the tumor. can be accurately analyzed, and an optimal tumor treatment solution can be provided based on the accurately analyzed result.
  • the automatic administration device 1000 may include a robot arm 112, an edge computer 114, a CCD camera 116, an anesthesia respirator 118, and an animal calibration device 122, It is not limited thereto, and may include more components necessary for managing a tumor to be treated, or may be provided with fewer components.
  • the automatic administration device 1000 may obtain a tumor image of a tumor to be treated using a CCD camera. More specifically, the automatic administration device 1000 may use an upper tumor image obtained by photographing the tumor from the top of the target tumor to be treated and a side tumor image obtained by photographing the target tumor from the side of the target tumor. The automatic administration device 1000 may determine the maturity level of the tumor to be treated by applying a trigonometric method to the upper tumor image and the lateral tumor image.
  • the automatic administration device 1000 may acquire maturity information about the tumor to be treated by inputting the tumor image into a pre-learned artificial intelligence model.
  • the automatic administering device 1000 may determine maturity according to passing weights and define an optimal drug administration time by applying an artificial intelligence model to a tumor image obtained through a camera.
  • the maturity information acquired by the automatic administration device 1000 using an artificial intelligence model may include tumor type information on at least one of the size, shape, or color of a tumor to be treated.
  • the automatic administration device 1000 provides an artificial intelligence model-based automatic administration solution, it is possible to establish a method for producing a tumor disease model, to secure the reliability of experiments on next-generation anticancer drugs, and to provide an automatic administration solution for tumors. It can be used to establish a breast cancer tumor model manufacturing method and to build data on the location and method of tumor administration.
  • the automatic administration device 1000 can learn an artificial intelligence model capable of optimizing tumor recognition and maturity, determine the timing of drug administration using the optimized maturity, and optimize tumor recognition for improving tumor recognition. Datasets can be built, and it can be directly applied experimentally to tumor disease models using vision machine learning.
  • the automatic administration device 1000 can provide a future-oriented animal testing device manual and animal testing model that are more economical and efficient than conventional domestic laboratory animal facilities, and can be used to evaluate the effectiveness of next-generation anticancer drug research or to existing studies. can be utilized
  • the automatic administration device 1000 can check whether the experiment is possible by checking the size of the tumor through tumor recognition, and the clinical failure rate when evaluating the effectiveness of new drug candidates. has the effect of reducing
  • the automatic administration device 1000 reduces the factor of additional cost of the experimental facility, thereby enabling 'anti-cancer drug and new drug development period' and 'remarkable reduction in cost', and it takes 10 to 15 years, and phase 1 , It does not require a large number of experts, who account for the largest part of the period and cost of new drug development to go to phase 2 and 3, reducing the cost of professional manpower while maintaining productivity and accuracy. can be reduced to
  • FIG. 2 is a flowchart of a method of managing a tumor based on an artificial intelligence model by an automatic administration device according to an embodiment.
  • the automatic administration device 1000 may obtain a tumor image of a tumor to be treated.
  • the automatic administration device 1000 may acquire a plurality of tumor images by photographing a tumor to be treated from multiple angles using a camera.
  • the automatic administration device 1000 acquires a lateral tumor image and an upper tumor image by photographing a tumor to be treated, and may apply trigonometry to the obtained lateral tumor image and upper tumor image.
  • the automatic administration device 1000 when the tumor image is input, obtains maturity information from the artificial intelligence model by inputting the tumor image to an artificial intelligence model that outputs maturity information of the tumor to be treated included in the tumor image. can do.
  • the automatic administration device 1000 may learn a plurality of artificial intelligence models that output maturity information when a tumor image is input, and at least one selected from among the plurality of learned artificial intelligence models based on a performance evaluation result. By using one artificial intelligence model, tumor images can be analyzed.
  • the artificial intelligence model used by the automatic administration device 1000 may include a machine learning model, a deep learning network, and a neural network model that can be learned based on a machine learning algorithm.
  • the automatic administration device 1000 may determine drug administration information based on the maturity level information. For example, the automatic administration device 1000 administers a drug related to at least one of a type of drug to be applied to a tumor to be treated, a drug administration time, a drug administration cycle, and a drug administration method based on maturity information obtained from an artificial intelligence model. information can be determined. According to an embodiment, the maturity level information may further include tumor type information.
  • the automatic administration device 1000 determines drug type information regarding the type of drug for treating tumor based on the maturity level information and the tumor type information, and based on the maturity level information and the tumor type information, Drug administration timing information regarding a drug administration time point for treating a tumor may be determined, and drug administration dose information regarding a drug administration dose for tumor treatment may be determined based on the maturity level information and the tumor type information.
  • the drug administration information determined by the automatic administering device 1000 may include at least one of drug type information, drug administration timing information, and drug administration information.
  • the automatic administration device 1000 may output the drug administration information determined in S230. Although not shown in FIG. 2 , the automatic administration device 1000 may determine a robot arm control signal based on drug administration information and may control the robot arm based on the robot arm control signal. According to another example, of course, the automatic administration device 1000 may transmit drug administration information to other electronic devices or servers connected to the automatic administration device 1000 .
  • FIG. 3 is a flowchart of a method for managing a tumor based on an artificial intelligence model by an automatic administration device according to another embodiment.
  • the automatic administering device 1000 may determine a control signal for the robot arm within the automatic administering apparatus for administering a predetermined drug based on the output drug administration information.
  • the automatic administration device 1000 may control the robot arm within the automatic administration device based on the robot arm control signal.
  • the automatic administration device 1000 may determine drug administration information based on maturity information acquired based on an artificial intelligence model, and directly administer a drug according to the drug administration information to a target tumor through a robot arm.
  • FIG. 4 is a diagram for explaining a process of constructing a tumor data set by an automatic administration device according to an embodiment.
  • the automatic administering device may construct a tumor dataset for learning an artificial intelligence model prior to obtaining a tumor image (eg, before obtaining a tumor image to be applied to a learned and selected artificial intelligence model).
  • the automatic administration device 1000 may acquire tumor data and all unlabeled tumor data sets related to tumors.
  • the automatic administration device 1000 classifies the tumor data and the tumor data set of a portion (eg, 3% to 5%) of the entire unlabeled tumor data set into a first group tumor data set and a second group tumor data set.
  • the automatic administration device 1000 displays tumor data and tumor regions corresponding to the tumor data in n classes according to the tumor type for the first group of tumor data sets (3% to 5%). You can create labeled data. For example, the automatic administration device 1000 divides the first group of tumor data sets into groups into n number of tumor data sets according to tumor types using experts' knowledge (for example, based on experts' user input for the automatic administration device). Tumor data as a class and tumor region corresponding to the tumor data may be marked and labeled (for example, Class #1, Class #2, ... Class #n).
  • FIG. 5 is a diagram for explaining a process of constructing a tumor data set by an automatic administration device according to an embodiment.
  • Figures 510 and 520 show a process in which the automatic administration device 1000 constructs a tumor data set following the process of FIG. 4 .
  • the automatic administration device 1000 may first inspect the labeled data for the first group of tumor data sets based on the preset first group of tumor data.
  • the automatic administration device 1000 applies a semi-supervised learning technique to the unlabeled tumor dataset of the second group, and proceeds with labeling the tumor data with n groups and tumor regions.
  • the automatic administering device 1000 may perform classification learning of n classes by performing supervised learning based on the checked labeled data of the first group of tumor data sets.
  • the automatic administration device 1000 primarily uses n class classification information through supervised learning based on the first group of tumor data sets, thereby class information of the second group of tumor data sets by an unsupervised learning algorithm, tumor You can write zones and labeling.
  • the automatic administration device 1000 applies an unsupervised learning algorithm to the unlabeled tumor data set of the second group using n class information according to the classification learning result of the n classes, thereby determining the type of tumor.
  • Tumor data, tumor regions corresponding to the tumor data, and labeled data are generated with n classes according to the class.
  • the automatic administering device 1000 may secondarily check the labeled data for the second group of tumor data set based on the preset second group of tumor data. For example, the automatic administration device 1000 may randomly perform a second inspection on the tumor data of the second group prepared by experts, the tumor area, and the labeled data.
  • the automatic administering device 1000 according to the present disclosure may construct the entire tumor data including the first group tumor data set and the second group tumor data set, the tumor area, and the labeled data as a tumor data set.
  • FIG. 6 is a diagram for explaining a process of analyzing a tumor data set by an automatic administration device according to an embodiment.
  • Figure 610 of FIG. 6 shows a process in which the automatic administration device 1000 analyzes a tumor data set and classifies the built tumor data set into learning data, verification data, and test data.
  • the automatic administration device 1000 divides the data set into three data sets by using the entire tumor data obtained by integrating the tumor data set of the first group and the tumor data set of the second group, the tumor area, and labeling information. It can be classified into groups: training data, validation data, and test data.
  • the automatic administration device 1000 may determine hyperparameters for artificial intelligence model learning based on learning data, re-learn the artificial intelligence model based on learning data and verification data, or test data It can be used to evaluate the performance of artificial intelligence models learned based on
  • FIG. 7 is a diagram for explaining an artificial intelligence model used by an automatic administration device according to an embodiment.
  • the automatic administration device 1000 is an artificial intelligence model for tumor detection, and a process of designing a vision machine learning learning model is schematically shown.
  • the automatic administration device 1000 configures a candidate group of a vision machine learning learning model for tumor detection, and firstly sets hyperparameters using learning data in the candidate group of the learning model to proceed with learning, , as many models as the number of algorithms in the candidate group are constructed.
  • the automatic administration device 1000 determines hyperparameters for artificial intelligence model learning based on the learning data classified in FIG. 6 for tumor detection, and a plurality of artificial intelligence based on the determined hyperparameters. Models 712 may be trained.
  • FIG. 8 is a diagram for explaining a process of selecting one of a plurality of artificial intelligence models learned by an automatic administration device according to an embodiment.
  • the automatic administration device 1000 calculates a predicted value by using verification data of a plurality of learning models (eg, a plurality of artificial intelligence models), which are results of the first process.
  • the automatic administration device 1000 may evaluate the performance of a plurality of learned artificial intelligence models using the difference between the calculated prediction value and the label value of the verification data.
  • the automatic administration device 1000 may evaluate the performance of a plurality of artificial intelligence models and select one artificial intelligence model from among the plurality of artificial intelligence models based on the evaluation result.
  • the automatic administering device 1000 can accurately analyze the tumor to be treated by applying the tumor image to one artificial intelligence model showing the highest performance based on the performance evaluation result.
  • FIG. 9 is a diagram for explaining a process of implementing an artificial intelligence model used by an automatic administration device for tumor detection according to an embodiment.
  • the automatic administration device 1000 proceeds with re-learning of the selected artificial intelligence model using training data and verification data using the selected artificial intelligence model with excellent performance.
  • the automatic administration device 1000 may re-evaluate the performance of the retrained artificial intelligence model based on the test data. For example, the automatic administering device 1000 finally evaluates performance using a difference between a predicted value and a label value of test data using training data and a model using a retrained model.
  • the automatic administering device 1000 can train the artificial intelligence model retrained based on the built tumor data set when the performance score of the artificial intelligence model retrained based on the performance re-evaluation is identified as being above a predetermined threshold. there is. Referring to Figure 930, the automatic administering device 1000 can perform tumor analysis and treatment by using the artificial intelligence model selected as the final model after completing learning using the entire data set. The automatic administration device 1000 acquires maturity information from the artificial intelligence model by inputting the newly acquired tumor image to be treated into the finally selected artificial intelligence model.
  • Maturity information acquired by the automatic administration device 1000 according to the present disclosure from the artificial intelligence model further includes tumor type information regarding at least one of the size of the tumor, shape of the tumor, or color of the tumor in the tumor image from the artificial intelligence model. can do.
  • 10 is a diagram for explaining the accuracy of learning data used for learning an artificial intelligence model used by an automatic administration device according to an embodiment.
  • FIG. 1010 of FIG. 10 an example of a recognition result of an artificial intelligence model learned based on YOLO learning data among learning data used for learning an artificial intelligence model is shown.
  • FIG. 11 is a block diagram of an automatic administration device according to an embodiment.
  • the automatic administration device 1000 may include a display 1210, a robot arm 1500, an edge computer 1800, and a camera 1610. However, not all illustrated components are essential components. The automatic administration device 1000 may be implemented with more components than those shown, or the automatic administration device 1000 may be implemented with fewer components.
  • the edge computer 1800 may include a memory 1700 storing one or more instructions and at least one processor 1300 executing the one or more instructions.
  • the display 1210 may display an image of a tumor to be treated taken by a camera.
  • the display 1210 may display a tumor image photographed at a predetermined angle and may display maturity information obtained from an artificial intelligence model on a screen.
  • the display 1210 may display drug administration information determined based on the maturity level information on the screen.
  • the robot arm 1500 may administer drugs to a tumor to be treated under the control of a processor.
  • the robot arm 1500 may further include an injection module for administering a target drug, an end module to which the injection module is fastened, and a plurality of other driving modules for driving the end module and the injection module.
  • the edge computer 1800 may process image data of a tumor in real time based on a pre-learned weight file.
  • the automatic administration device 1000 can process image data using edge computing technology for real-time processing of tumor image data acquired through a camera, and the automatic administration device 1000 ) You can also control the functions of the overall components within.
  • the edge computer 1800 may determine the degree of maturity by applying an artificial intelligence model to the tumor image acquired through the camera, and determine the optimal drug administration time (tumor size, shape, color). In addition, the edge computer 1800 may determine the type of drug in the drop, the timing of drug administration, the cycle of drug administration, or other drug administration methods.
  • the camera 1610 may acquire a plurality of tumor images by capturing a tumor to be treated at a predetermined angle under the control of the edge computer 1800 . Also, the camera 1610 may acquire a tumor image in which a plurality of tumor images are arranged at predetermined frame intervals.
  • the processor 1300 acquires a tumor image of a tumor to be treated by executing one or more instructions stored in the memory 1700, and when the tumor image is input, the tumor image to be treated is included in the tumor image.
  • maturity information may be obtained from the artificial intelligence model, drug administration information may be determined based on the maturity information, and the determined drug administration information may be output. there is.
  • the processor 1300 determines a robot arm control signal in the automatic administering device for administering a predetermined drug based on the output drug administration information, and based on the determined robot arm control signal. It is possible to control the robot arm in the automatic dosing device.
  • the processor 1300 obtains tumor data and the entire unlabeled tumor data set prior to acquiring the tumor image, and the tumor data and a portion of the unlabeled tumor data set.
  • the data set is classified into a first group tumor data set and a second group tumor data set, and for the first group tumor data set, n classes according to the tumor type are used to classify the tumor data and the tumor corresponding to the tumor data
  • Generates region marking and labeled data first inspects the labeled data for the tumor data set of the first group based on the previously set tumor data of the first group, and inspects the first group Based on the labeled data for the tumor data set of, classification learning of n classes is performed by performing supervised learning, and n class information according to the classification learning result of the n classes is used to determine the unlabeled
  • an unsupervised learning algorithm to the tumor data set of the second group, tumor data and tumor regions corresponding to the tumor data and labeled data are generated in n classes according to tumor types, and the tumor of the second
  • the processor 1300 may classify the built tumor data set into training data, verification data, and test data prior to acquiring the tumor image.
  • the processor 1300 determines hyperparameters for artificial intelligence model learning based on the classified learning data prior to obtaining the tumor image, and based on the determined hyperparameters, a plurality of It trains artificial intelligence models, evaluates the performance of the plurality of learned artificial intelligence models, selects one artificial intelligence model from among the plurality of artificial intelligence models based on the evaluation result, and determines the maturity level from the selected artificial intelligence model. information can be obtained.
  • the processor 1300 re-learns the selected artificial intelligence model based on the training data and the verification data prior to obtaining the tumor image, and the re-learned based on the test data.
  • the performance of the artificial intelligence model is re-evaluated, and if the performance score of the re-learned artificial intelligence model is identified as being above a predetermined threshold based on the performance re-evaluation result, the re-learned based on the built tumor data set
  • the maturity level information may be obtained from the learned and re-learned artificial intelligence model based on the constructed tumor data set.
  • the memory 1700 may include a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (eg SD or XD memory, etc.), RAM (RAM, Random Access Memory) SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk , an optical disk, and at least one type of storage medium.
  • RAM Random Access Memory
  • SRAM Static Random Access Memory
  • ROM Read-Only Memory
  • EEPROM Electrical Erasable Programmable Read-Only Memory
  • PROM Programmable Read-Only Memory
  • magnetic memory magnetic disk , an optical disk, and at least one type of storage medium.
  • the memory 1700 stores tumor images of the tumor acquired by the automatic administration device 1000, artificial intelligence models, weight information for the artificial intelligence models, a tumor data set, learning data, verification data, and test data. can Also, according to an embodiment, the memory 1700 may store drug administration information determined to treat a tumor.
  • the method according to an embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer readable medium.
  • the computer readable medium may include program instructions, data files, data structures, etc. alone or in combination.
  • Program commands recorded on the medium may be specially designed and configured for the present disclosure, or may be known and usable to those skilled in computer software.
  • a computer program device including a recording medium in which a program for performing a different method according to the above embodiment is stored may be provided.
  • Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks.
  • - includes hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like.
  • program instructions include high-level language codes that can be executed by a computer using an interpreter, as well as machine language codes such as those produced by a compiler.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Biomedical Technology (AREA)
  • Physics & Mathematics (AREA)
  • Pathology (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Biophysics (AREA)
  • Biotechnology (AREA)
  • Evolutionary Biology (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Robotics (AREA)
  • Mechanical Engineering (AREA)
  • Chemical & Material Sciences (AREA)
  • Medicinal Chemistry (AREA)
  • Image Analysis (AREA)
  • Quality & Reliability (AREA)

Abstract

La présente divulgation concerne un procédé par lequel un appareil d'administration automatique prend en charge une tumeur sur la base d'un modèle d'intelligence artificielle et un appareil d'administration automatique mettant en œuvre ledit procédé. Selon un mode de réalisation, un procédé par lequel un appareil d'administration automatique prend en charge une tumeur sur la base d'un modèle d'intelligence artificielle peut comprendre les étapes consistant : à acquérir une image de tumeur d'une tumeur à traiter ; lorsque l'image de tumeur est entrée, à entrer l'image de tumeur dans un modèle d'intelligence artificielle qui émet des informations de maturité de la tumeur à traiter comprises dans l'image de tumeur et à acquérir les informations de maturité à partir du modèle d'intelligence artificielle ; à déterminer des informations d'administration de médicament sur la base des informations de maturité ; et à émettre les informations d'administration de médicament déterminées.
PCT/KR2022/019271 2021-11-30 2022-11-30 Appareil et procédé de prise en charge de tumeur sur la base d'un modèle d'intelligence artificielle WO2023101444A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020210169556A KR102417602B1 (ko) 2021-11-30 2021-11-30 인공 지능 모델 기반 종양 관리 장치 및 방법
KR10-2021-0169556 2021-11-30

Publications (1)

Publication Number Publication Date
WO2023101444A1 true WO2023101444A1 (fr) 2023-06-08

Family

ID=82398431

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2022/019271 WO2023101444A1 (fr) 2021-11-30 2022-11-30 Appareil et procédé de prise en charge de tumeur sur la base d'un modèle d'intelligence artificielle

Country Status (2)

Country Link
KR (1) KR102417602B1 (fr)
WO (1) WO2023101444A1 (fr)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102417602B1 (ko) * 2021-11-30 2022-07-07 (주)플라스바이오 인공 지능 모델 기반 종양 관리 장치 및 방법

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20170053617A (ko) * 2014-07-14 2017-05-16 알레그로 다이어그노스틱스 코포레이션 폐암 상태를 평가하는 방법
KR101818074B1 (ko) * 2017-07-20 2018-01-12 (주)제이엘케이인스펙션 인공지능 기반 의료용 자동 진단 보조 방법 및 그 시스템
WO2019111512A1 (fr) * 2017-12-05 2019-06-13 ソニー・オリンパスメディカルソリューションズ株式会社 Dispositif de traitement d'informations médicales et procédé de traitement d'informations
KR20200004204A (ko) * 2018-07-03 2020-01-13 (주) 프로큐라티오 빅데이터분석기반 암진단장치 및 방법
KR102417602B1 (ko) * 2021-11-30 2022-07-07 (주)플라스바이오 인공 지능 모델 기반 종양 관리 장치 및 방법

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102313661B1 (ko) 2020-12-30 2021-10-19 뉴로핏 주식회사 특성 정보를 고려한 의료 영상 분석 방법, 의료 영상 분석 장치 및 의료 영상 분석 시스템

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20170053617A (ko) * 2014-07-14 2017-05-16 알레그로 다이어그노스틱스 코포레이션 폐암 상태를 평가하는 방법
KR101818074B1 (ko) * 2017-07-20 2018-01-12 (주)제이엘케이인스펙션 인공지능 기반 의료용 자동 진단 보조 방법 및 그 시스템
WO2019111512A1 (fr) * 2017-12-05 2019-06-13 ソニー・オリンパスメディカルソリューションズ株式会社 Dispositif de traitement d'informations médicales et procédé de traitement d'informations
KR20200004204A (ko) * 2018-07-03 2020-01-13 (주) 프로큐라티오 빅데이터분석기반 암진단장치 및 방법
KR102417602B1 (ko) * 2021-11-30 2022-07-07 (주)플라스바이오 인공 지능 모델 기반 종양 관리 장치 및 방법

Also Published As

Publication number Publication date
KR102417602B1 (ko) 2022-07-07

Similar Documents

Publication Publication Date Title
Becker et al. Training and deploying a deep learning model for endoscopic severity grading in ulcerative colitis using multicenter clinical trial data
US11901077B2 (en) Multiple instance learner for prognostic tissue pattern identification
De Chaumont et al. Real-time analysis of the behaviour of groups of mice via a depth-sensing camera and machine learning
WO2017095014A1 (fr) Système de diagnostic d'anomalie cellulaire utilisant un apprentissage dnn, et procédé de gestion de diagnostic de celui-ci
KR102583103B1 (ko) 계산 검출 방법들을 위해 전자 이미지들을 처리하기 위한 시스템들 및 방법들
CN110853022B (zh) 病理切片图像的处理方法、装置、系统及存储介质
Schaumberg et al. Interpretable multimodal deep learning for real-time pan-tissue pan-disease pathology search on social media
WO2023101444A1 (fr) Appareil et procédé de prise en charge de tumeur sur la base d'un modèle d'intelligence artificielle
Dong et al. Minian, an open-source miniscope analysis pipeline
de Chaumont et al. Live Mouse Tracker: real-time behavioral analysis of groups of mice
WO2021075826A1 (fr) Procédé de diagnostic de maladie par prétraitement de données extraites par un moyen biologique et dispositif associé
BR112022013000B1 (pt) Método implementado por computador para processar imagens de patologia digitais empregando amostras in vitro, sistema para processar imagens de patologia digitais e meio legível por computador não transitório
CN111080624A (zh) 精子运动状态的分类方法、装置、介质及电子设备
van Albada et al. Bringing anatomical information into neuronal network models
Chierici et al. Automatically detecting Crohn’s disease and Ulcerative Colitis from endoscopic imaging
WO2023095989A1 (fr) Procédé et dispositif d'analyse d'images médicales à modalités multiples pour le diagnostic d'une maladie cérébrale
CN108205708A (zh) 一种新型可扩展的深度学习系统及数据识别方法
JP7315181B2 (ja) 探索方法及び情報処理システム
Chen et al. Application of artificial intelligence to clinical practice in inflammatory bowel disease–what the clinician needs to know
CN112883770A (zh) 一种基于深度学习的pd-1/pd-l1病理图片识别方法和装置
Wang et al. Automatic identification and analysis of multi-object cattle rumination based on computer vision
CN114283114A (zh) 图像处理方法、装置、设备及存储介质
König et al. Modeling visual exploration in rhesus macaques with bottom-up salience and oculomotor statistics
Orrù et al. A brief overview on the contribution of machine learning in systems neuroscience
CN109859813A (zh) 一种实体修饰词识别方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22901802

Country of ref document: EP

Kind code of ref document: A1