WO2020082638A1 - 医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置、计算机可读存储介质 - Google Patents

医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置、计算机可读存储介质 Download PDF

Info

Publication number
WO2020082638A1
WO2020082638A1 PCT/CN2019/073339 CN2019073339W WO2020082638A1 WO 2020082638 A1 WO2020082638 A1 WO 2020082638A1 CN 2019073339 W CN2019073339 W CN 2019073339W WO 2020082638 A1 WO2020082638 A1 WO 2020082638A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
medical image
annotation
result
labeling
Prior art date
Application number
PCT/CN2019/073339
Other languages
English (en)
French (fr)
Inventor
车拴龙
罗丕福
刘栋
刘斯
梁贺华
李映华
邱伟松
苏钜铭
Original Assignee
广州金域医学检验中心有限公司
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
Priority claimed from CN201811260183.5A external-priority patent/CN109473160A/zh
Priority claimed from CN201811261521.7A external-priority patent/CN109509541B/zh
Priority claimed from CN201811260195.8A external-priority patent/CN109461147B/zh
Priority claimed from CN201811260198.1A external-priority patent/CN109446370A/zh
Application filed by 广州金域医学检验中心有限公司 filed Critical 广州金域医学检验中心有限公司
Priority to US16/628,683 priority Critical patent/US11094411B2/en
Publication of WO2020082638A1 publication Critical patent/WO2020082638A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/20ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00ICT specially adapted for the handling or processing of medical images
    • G16H30/20ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00ICT specially adapted for the handling or processing of medical images
    • G16H30/40ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning

Definitions

  • the application number is 2018112601958
  • the application name is the pathological annotation method and device for FOV pictures applied to mobile terminals
  • application number For 2018112601835 the application name is the pathological annotation qualification method and device for medical images
  • the application number is 2018112601981
  • the application name is the medical image pathology annotation method and device
  • the computer-readable storage medium the application number is 2018112615217
  • the application name is based on Image report issuing method and device, and computer readable storage medium.
  • the invention relates to the field of intelligent medical treatment, in particular to a pathological annotation method and device for medical images, and a method and device for issuing reports based on medical images.
  • the embodiments of the present invention provide a pathological labeling method and device for medical images, and a computer-readable storage medium, which can effectively solve the problem of the existing manual labeling work place is limited, improve the efficiency of labeling work, and provide high quality and A large number of learning samples.
  • An embodiment of the present invention provides a pathological annotation method for medical images, including the steps of:
  • the medical image to be labeled is randomly called and displayed on the display interface of the mobile terminal;
  • the medical image is generated by the following steps:
  • the pathological annotation method for medical images disclosed in the embodiments of the present invention determines whether the user is qualified for pathological annotation when receiving a request from any user to enter the annotation mode entered on the mobile terminal ;
  • receive the first annotation result of the medical image by the user and save the
  • the first annotation result of the medical image by the user the medical image is generated by the following steps: after dividing the original scanned image of the pathological slice into n FOV pictures, calculating each of the FOV pictures in the n FOV pictures Pathological index; where, 10 ⁇ 10 ⁇ n ⁇ 10; the storage space occupied by the FOV picture is less than a preset threshold; the pathological index of each FOV picture in the n small pictures, obtain the n
  • the top m FOV pictures with the highest pathological index among the FOV pictures are used as the medical images; where 50 ⁇ m ⁇ 5, the labeling worker does
  • the method further includes the steps of:
  • the first labeling results of the two users are used as the reference labeling results of the medical image, and the correct labeling numbers of the two users are increased by 1 respectively .
  • the method further includes the steps of:
  • the method further includes the steps of:
  • the second annotation result of the medical image by the other user or expert user is used as the reference annotation result of the medical image And add 1 to the number of erroneous labels for users whose first and second labeling results are inconsistent, and add 1 to the number of correct labels for users whose first and second labeling results are consistent .
  • the method further includes the steps of:
  • the user When it is detected that the number of erroneous annotations of any user is greater than the preset first threshold, the user is disqualified for pathological annotation.
  • the method further includes the steps of:
  • the medical image is added to the difficult case database.
  • the method further includes the steps of:
  • the AI annotation result of the medical image When displaying the medical image, obtain the AI annotation result of the medical image; wherein, the AI annotation result of the medical image includes a positive result and a negative result;
  • the AI annotation result of the medical image is a negative result, and the first annotation result of the medical image by any user is consistent with the AI annotation result of the medical image, the AI annotation result of the medical image is used as The reference annotation result of the medical image;
  • the AI annotation result of the medical image is a positive result, and the first annotation result of the medical image by any two users is consistent with the AI annotation result of the medical image, the AI annotation result of the medical image is used As the reference annotation result of the medical image;
  • the first annotation result of the medical image serves as a reference annotation result of the medical image
  • the AI annotation result of the medical image is a positive result, and the first annotation result of the medical image by one of any two users is inconsistent with the AI annotation result of the medical image, randomly send to any other
  • the user or expert user sends a request for annotation of the medical image or a request for organization and discussion;
  • the second annotation result of the medical image by the other user or expert user is received, the second annotation result of the medical image by the other user or expert user is used as the reference annotation result of the medical image ;
  • the third annotation result is used as the reference annotation result of the medical image.
  • the determining whether the user is qualified for pathological labeling specifically includes:
  • the determining whether the user is qualified for pathological annotation further includes:
  • the labeling accuracy rate is greater than a preset third threshold
  • a training interface is displayed, and x medical images used to train the user are called;
  • the medical images used to train the user all have correct annotation results; where x ⁇ 5.
  • the determining whether the user is qualified for pathological annotation further includes:
  • the determining whether the user is qualified for pathological annotation further includes:
  • the user's training request and labeling qualification request are re-accepted.
  • the receiving the first annotation result of the medical image by the user and saving the first annotation result of the medical image by the user includes:
  • obtaining the first annotation result of the medical image by the user includes the steps of:
  • the first annotation result of the medical image by the user is obtained.
  • the second option button includes an agree button and an object button
  • the result of obtaining the first annotation of the medical image by the user is specifically:
  • the result of the labeling of the medical image by the other user or machine is used as the first labeling result of the user;
  • the annotation result re-entered by the user is used as the first annotation result of the user.
  • the method further includes the steps of:
  • a voice input mark is displayed at the same time; wherein the display state of the voice input mark includes a voice input state being detected and a voice input stop detection state;
  • the display state of the voice input mark is modified to the state where the voice input is being detected, and the annotation result of the user input is obtained through the voice device as the user's The first annotation result.
  • the voice input mark is a voice input progress bar
  • the voice progress bar in a full state is displayed on the display interface ;
  • the display state of the voice input mark is the state of detecting the voice input, the voice progress bar that is gradually shortened within the preset time period on the display interface.
  • the embodiment of the present invention also correspondingly provides a medical image-based report issuance method, including the steps of:
  • the AI interpretation result of the medical image When receiving a report issuance instruction on any medical image, obtain the AI interpretation result of the medical image; wherein, the AI interpretation result of the medical image includes a negative result and a positive result; wherein, the AI interpretation of the medical image
  • the result is obtained after being interpreted by a preset classifier, and the classifier is trained by the reference annotation result of the medical image as described in any one of the above;
  • the method further includes the steps of:
  • the Or an expert doctor user sends a request for interpretation of the medical image or a request for organization and discussion.
  • the method further includes the steps of:
  • the method further includes the steps of:
  • Another embodiment of the present invention provides a pathological annotation apparatus for medical images, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor executing the The computer program implements the pathological annotation method of the medical image described in any of the above embodiments of the invention.
  • Another embodiment of the present invention provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the above
  • the pathological annotation method for medical images according to any embodiment of the invention.
  • Another embodiment of the present invention provides a pathological annotation apparatus for medical images, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor executing the The computer program implements the medical image-based report issuance method described in any of the above embodiments of the invention.
  • Another embodiment of the present invention provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the above
  • the medical image-based report issuance method described in any embodiment of the invention is not limited to any embodiment of the invention.
  • FIG. 1 is a schematic flowchart of a medical image pathology labeling method according to an embodiment of the present invention.
  • FIG. 2 is a schematic diagram of cutting an original scan image of a pathological slice provided by an embodiment of the present invention.
  • Fig. 3 (a) is a small picture of cervical cytology single cell cutting
  • Fig. 3 (b) is a picture of cervical cytology cell cutting.
  • FIG. 4 is a schematic diagram of a display interface of a mobile terminal provided by an embodiment of the present invention.
  • FIG. 5 is a schematic diagram of a display interface of a mobile terminal provided by another embodiment of the present invention.
  • FIG. 6 is a schematic flowchart of confirming whether the user is qualified for pathological labeling according to an embodiment of the present invention.
  • FIG. 7 is a schematic flowchart of a medical image-based report issuance method provided by an embodiment of the present invention.
  • FIG. 8 is a schematic structural diagram of a medical image pathology labeling device according to an embodiment of the present invention.
  • FIG. 9 is a schematic structural diagram of a terminal device provided by an embodiment of the present invention.
  • FIG. 1 is a schematic flowchart of a pathological annotation method for medical images according to an embodiment of the present invention, which is used in a mobile terminal and includes the steps of:
  • step S1 the user needs to undergo a special test to grant his pathological annotation qualification.
  • s sheets of medical images for testing the user need to be randomly called to obtain the annotation result of each medical image by the user.
  • the user compares the labeling result of each medical image with the correct labeling result, and obtains the labeling correct rate of the medical image with correct labeling result of the s sheets (s ⁇ 10) by the user.
  • the labeling correct rate is greater than a preset first threshold, it is determined that the user is qualified for pathological labeling. Screening the labeling workers through a unified test can ensure a high medical level, consistency and stability of labeling quality, can ensure labeling quality, and can obtain objective scientific evaluation indicators, which is conducive to subsequent machine learning model construction .
  • step S1 preferably, the medical image is generated in the following manner:
  • the pathological index of each of the FOV pictures in the n FOV pictures is calculated; wherein, 10 ⁇ 10 ⁇ n ⁇ 10 ;
  • the storage space occupied by the FOV picture is less than a preset threshold;
  • artificial intelligence can prioritize and present 20 images from a large number of images, which can greatly reduce the workload of the doctor.
  • the original scanned image is converted into a digital image of X400 times by a cell pathology slide through a digital slice scanner, and its size can reach several G to several tens of G.
  • the labeling workers label the whole film, the workload is very large, and they can only work on a large-capacity computer, and the location is limited. Therefore, the original labeling work has missing labeling, cumbersome labeling steps, and huge workload. problem. Therefore, by cutting the original digital image, the labeling worker can perform labeling anytime and anywhere. The cut image does not need to be loaded on a large-capacity computer, which reduces the workload of the labeling worker.
  • Figure 3 (a) it is a small picture of cervical cytology single cell cutting
  • Figure 3 (b) it is a small picture of cervical cytology cell cutting.
  • the small pictures can be easily downloaded to different receiving platforms on the Internet, which can facilitate quick response in the cloud to the mobile terminal device to achieve fast operation, thereby completing the labeling process. Therefore, the labeling workers can use mobile terminals (such as mobile phones, etc.) for testing and labeling in their spare time (eg, in a car, waiting for a car), or use a general PC at home or office for labeling, which can be used on different devices. Synchronized operation greatly increases convenience, and maximizes the user's fragmented time operation.
  • a first option button including ASC, LSIL, AGC, HSIL, etc.
  • the system first selects LSIL by default, and adjusts the position of various label options according to the characteristics of the probability distribution of the lesion, which improves the friendliness and convenience of system operation.
  • the first annotation result of the medical image by the user is obtained.
  • the second option button includes an agree button and an object button
  • the result of obtaining the first annotation of the medical image by the user is specifically:
  • the result of the labeling of the medical image by the other user or machine is used as the first labeling result of the user;
  • the annotation result re-entered by the user is used as the first annotation result of the user.
  • the display interface of the mobile terminal in addition to displaying medical images, the previous manual labeling results or machine labeling results are displayed, and whether the user ’s or the machine ’s labeling results of the medical images are agreed
  • the second option button (that is, the "YES” and "NO” buttons in the figure) is correct. Click YES to save and enter the next one. If wrong, click NO and re-enter the new annotation result. This way of reviewing improves the speed of correcting errors and minimizes the complexity of human operations.
  • a voice input mark is displayed at the same time; wherein the display state of the voice input mark includes a voice input state being detected and a voice input stop detection state;
  • the display state of the voice input mark is modified to the state where the voice input is being detected, and the annotation result of the user input is obtained through the voice device as the user's The first annotation result.
  • the voice input mark is a voice input progress bar, and when the display state of the voice input mark is to stop detecting the voice input state, the voice progress bar in a full state is displayed on the display interface;
  • the display state of the voice input mark is the voice progress bar that is being detected and the voice input bar is gradually shortened within a preset time period on the display interface.
  • the first labeling results of the two users are used as the reference labeling results of the medical image, and the correct labeling numbers of the two users are increased by 1 respectively ;
  • the second annotation result of the medical image by the other user or expert user is used as the reference annotation result of the medical image And add 1 to the number of erroneous labels for users whose first and second labeling results are inconsistent, and add 1 to the number of correct labels for users whose first and second labeling results are consistent .
  • the double review system also avoids the systematic errors in the design of artificial intelligence and ensures the reliability of the labeling. And accuracy.
  • the user when it is detected that the number of erroneous annotations of any user is greater than the preset first threshold, the user is disqualified from the pathological annotation. Join the work supervision mechanism and elimination mechanism to avoid the loopholes that may exist due to personal discretionary operation and insufficient diagnostic capabilities.
  • the user When it is detected that the number of correct labels of any user is greater than the preset second threshold, the user is added to the professional labeling worker database; when it is detected that the number of accurate labels of any user is greater than the preset third threshold, and When it is judged that the user has a licensed physician certificate, the user is granted the qualification to issue a report.
  • the ranking of users with the correct number of annotations can also be displayed on the display interface.
  • the method further includes steps:
  • the AI annotation result of the medical image When displaying the medical image, obtain the AI annotation result of the medical image; wherein, the AI annotation result of the medical image includes a positive result and a negative result;
  • the AI annotation result of the medical image is a negative result, and the first annotation result of the medical image by any user is consistent with the AI annotation result of the medical image, the AI annotation result of the medical image is used as The reference annotation result of the medical image;
  • the AI annotation result of the medical image is a positive result, and the first annotation result of the medical image by any two users is consistent with the AI annotation result of the medical image, the AI annotation result of the medical image is used As the reference annotation result of the medical image;
  • the first annotation result of the medical image serves as a reference annotation result of the medical image
  • the AI annotation result of the medical image is a positive result, and the first annotation result of the medical image by one of any two users is inconsistent with the AI annotation result of the medical image, randomly send to any other
  • the user or expert user sends a request for annotation of the medical image or a request for organization and discussion;
  • the second annotation result of the medical image by the other user or expert user is received, the second annotation result of the medical image by the other user or expert user is used as the reference annotation result of the medical image ;
  • the third annotation result is used as the reference annotation result of the medical image.
  • the labeling workers can label the medical images anytime and anywhere, the working place is not limited, the inefficient working mode is improved, and the medical service is significantly improved Quality and medical work efficiency provide artificial intelligence with high-quality and abundant learning samples.
  • the acquired medical image is random, and for each medical image, the assigned user is also random, which maximizes the objectivity in the interpretation of medical natural images. It reduces the differences in diagnosis and treatment in different regions and different hospitals, ensuring more patients benefit and more medical fairness.
  • confirming whether the user is qualified for pathological labeling specifically includes:
  • step S12 the acquisition correctness rate of the user's annotation of the a medical images with correct labeling results is obtained:
  • the accuracy rate of the labeling is greater than 0.95, it can be considered that the medical level of the labeling doctor is high and the consistency is relatively stable. It is determined that the user is qualified for labeling and can successfully join the membership as a labeling worker who trains machine learning.
  • the labeling workers are screened through a unified test, which can ensure a high medical level, consistency and stability of labeling quality.
  • the selection of labeling workers should be based on quality rather than seniority, to avoid subjective deviations based on the selection criteria of highly qualified doctors as labeling workers, such as senior labeling doctors. Often focusing on one type of sub-specialty, there may be a certain blind spot for other sub-specialties. Therefore, the scheme proves the quality of labeling and can obtain objective scientific evaluation indicators, which is conducive to the subsequent construction of machine learning models.
  • the labeling accuracy rate is greater than a preset third threshold
  • a training interface is displayed, and x medical images used to train the user are called;
  • the medical images used to train the user all have correct annotation results; where x ⁇ 5.
  • the user's test labeling accuracy rate is less than 0.95 and greater than 0.50, it may be due to operational errors, misunderstandings about individual lesions, etc., you can apply for online test questions to learn, and then you can apply for the preliminary test for preliminary screening . If the correct rate is greater than 0.95, it can be determined that the user is qualified for labeling. If it is still less than 0.95 and greater than 0.50, the user must still enter the training mode for training before re-initiating the labeling qualification request.
  • the training request and labeling qualification request of the user are rejected within a preset time period, that is, the user's account is blocked, and the learning interface is displayed.
  • the system knowledge about annotations is displayed on the learning interface for the user to conduct comprehensive theoretical knowledge training. After a preset time period, the user's training request and labeling qualification request are accepted again.
  • the correct rate of the labeling is less than 0.50, it may be due to the lack of diagnostic level and other reasons. It is recommended to conduct a systematic study of the sub-specialty. After the account is banned for a period of time (for example, 24-48h), users can apply for the entry test for online test question learning and preliminary screening again.
  • an embodiment of the present invention also correspondingly provides a medical image-based report issuance method, including steps:
  • a doctor user is a user who is qualified to issue a certificate, and generally has a licensed medical certificate.
  • the medical image is a small picture after cutting, it can be loaded on a mobile terminal held by a doctor user for display. Therefore, doctor users can use mobile terminals (for example, mobile phones, etc.) for testing and labeling in their spare time (for example, in a car, waiting for a car), or use a general PC at home or office for interpretation, and can be used on different devices.
  • the synchronous operation greatly increases the convenience, and uses the doctor's user's fragmentary time to operate the maximum time.
  • the small pictures after screening through artificial intelligence can greatly reduce the workload of doctor users and reduce the risk of doctor users being prone to errors due to fatigue.
  • step S73 since the probability of a negative result is much higher than that of a positive result, the error rate is low, and the order can be issued without review by multiple doctor users.
  • step S74 by randomly arranging the results of AI interpretation to any two doctor users to ensure the objectivity of the positive result, to avoid the bias of one doctor user, etc., leading to wrong results. And in the medical field, the probability of a positive result is low, and it needs to be treated with caution. Therefore, two doctors need to review to make an order.
  • the AI interpretation result has high objectivity and accuracy, and a report form regarding the medical image can be issued.
  • the AI interpretation result of the medical image is displayed on the display interface, and at the same time, whether to agree to the medical
  • the third option button of the AI interpretation result of the image wherein, the second option button includes an agree button and an object button;
  • artificial intelligence can be applied to the issuance of clinical practice reports, which is conducive to the development of medical intelligence, improves the efficiency of report issuance, ensures medical safety under artificial intelligence-assisted diagnosis, and increases the number of patients At the same time, the quality of medical care is improved without decreasing.
  • another embodiment further includes steps:
  • the Or an expert doctor user sends a request for interpretation of the medical image or a request for organization and discussion;
  • the labeling user in this disclosure is different from the doctor user.
  • the labeling user is only qualified for pathology labeling, while the doctor user is also qualified for report issuance and is responsible for the content of the report. Users need to undergo special tests to grant their pathology labeling qualifications. For example, a random number (a ⁇ 10) is required to test the user ’s medical images to obtain the user ’s labeling results for each of the medical images.
  • the user compares the labeling result of each medical image with the correct labeling result to obtain the labeling correct rate of the user for the a medical image with the correct labeling result, when it is judged that the labeling is correct When the rate is greater than the preset second threshold, it is determined that the user is qualified for pathology labeling. Screening the labeling workers through a unified test can ensure a high medical level, consistency and stability of labeling quality, can ensure labeling quality, and can obtain objective scientific evaluation indicators, which is conducive to subsequent machine learning model construction .
  • the method further includes the steps of:
  • the labeling user is added to the doctor user library.
  • FIG. 8 is a schematic structural diagram of a medical image pathology labeling device according to an embodiment of the present invention, including:
  • the pathology labeling qualification acquisition module 801 is configured to determine whether the user is qualified for pathology labeling when receiving a request to enter a labeling mode input by any user on the mobile terminal;
  • the medical image display module 802 is used to randomly call the medical image to be annotated and display it on the display interface of the mobile terminal when it is judged that the user is qualified for pathological annotation;
  • the first annotation result saving module 803 is configured to receive the first annotation result of the medical image by the user, and save the first annotation result of the medical image by the user.
  • Another embodiment of the present invention provides a pathological annotation apparatus for medical images, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor executing the The computer program implements the medical image pathology labeling method or the medical image-based report issuance method described in any of the above embodiments of the invention.
  • Another embodiment of the present invention provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the above
  • the method for annotating a medical image pathology or the method for issuing a report based on a medical image according to any embodiment of the invention.
  • the terminal device includes: at least one processor 11, such as a CPU, at least one network interface 14 or other user interface 13, a memory 15, at least one communication bus 12, and the communication bus 12 is used to implement connection and communication between these components.
  • the user interface 13 may optionally include a USB interface, other standard interfaces, and a wired interface.
  • the network interface 14 may optionally include a Wi-Fi interface and other wireless interfaces.
  • the memory 15 may include a high-speed RAM memory, or may also include a non-volatile memory (non-volatile memory), for example, at least one magnetic disk memory.
  • the memory 15 may optionally include at least one storage device located away from the foregoing processor 11.
  • the memory 15 stores the following elements, executable modules or data structures, or their subsets, or their extensions:
  • the operating system 151 includes various system programs, such as a battery management system, etc., for implementing various basic services and processing hardware-based tasks;
  • the processor 11 is used to call the program 152 stored in the memory 15 to execute the pathological labeling method of the medical image or the report issuing method based on the medical image described in the foregoing embodiment, for example, step S1 shown in FIG. 1.
  • the processor 11 executes the computer program, the functions of each module / unit in the foregoing device embodiments are realized, for example, the medical image display module 802.
  • the computer program may be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention.
  • the one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.
  • the terminal device may include, but is not limited to, the processor 11 and the memory 15.
  • the schematic diagram is only an example of a terminal device, and does not constitute a limitation on the terminal device, and may include more or fewer components than those illustrated, or combine certain components, or different components,
  • the apparatus for annotating pathology of medical images may further include input and output devices, network access devices, buses, and the like.
  • the so-called processor 11 may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Ready-made programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.
  • the processor 11 is a control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.
  • the memory 15 may be used to store the computer program and / or module.
  • the processor 11 implements the computer by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. Describe the various functions of the terminal device.
  • the memory 15 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, at least one function required application programs (such as sound playback function, image playback function, etc.), etc .; the storage data area may Store data created according to the use of mobile phones (such as audio data, phone book, etc.), etc.
  • the memory 15 may include a high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart, Media, Card, SMC), and a secure digital (SD) Card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device.
  • non-volatile memory such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart, Media, Card, SMC), and a secure digital (SD) Card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device.
  • the module / unit integrated in the terminal device may be stored in a computer-readable storage medium.
  • the present invention can implement all or part of the processes in the methods of the above embodiments, and can also be completed by a computer program instructing relevant hardware.
  • the computer program can be stored in a computer-readable storage medium. When the program is executed by the processor, the steps of the foregoing method embodiments may be implemented.
  • the computer program includes computer program code, and the computer program code may be in a source code form, an object code form, an executable file, or some intermediate form, etc.
  • the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM, Read-Only Memory) , Random Access Memory (RAM, Random Access Memory), and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Epidemiology (AREA)
  • Medical Informatics (AREA)
  • Primary Health Care (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Biomedical Technology (AREA)
  • General Business, Economics & Management (AREA)
  • Business, Economics & Management (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

本发明实施例公开的医学图像的病理标注方法,通过当接收到任一用户在所述移动终端输入的进入标注模式的请求时,判定所述用户是否具有病理标注资格;当判断所述用户具有病理标注资格中,随机调用待标注的医学图像并在所述移动终端的显示界面进行显示;接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果,所述医学图像通过以下步骤生成:对病理切片的原始扫描图像分割成n个FOV图片后,计算所述n个FOV图片中每一所述FOV图片的病理指数;其中,10^10≥n≥10;所述FOV图片所占用的存储空间小于预设的阈值;所述n个小图片中每一所述FOV图片的病理指数,获取所述n个FOV图片中病理指数最高的前m个FOV图片作为所述医学图像;其中,50≥m≥5,标注工作者无需面对全片整体性标注,仅需对分割后的小图片进行标注,减少标注工作者的工作量,降低其疲劳感,提高标注准确率,且在移动终端上操作,可随时随地进行标注,不受地点的限制,有利于提高标注工作的效率,为人工智能提供质量高且数量丰富的学习样本。能在保证标注质量的前提下,使得标注工作者可随时随地对医学图像进行标注,工作地点不受限制,能大大提高标注工作的效率,为人工智能提供质量高且数量丰富的学习样本。

Description

医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置、计算机可读存储介质
本申请要求2018年10月26号提交到中国专利局的四篇中国专利的优先权,具体如下:申请号为2018112601958、申请名称为应用于移动终端的FOV图片的病理标注方法及装置,申请号为2018112601835、申请名称为医学图像的病理标注资格确定方法及装置,申请号为2018112601981、申请名称为医学图像的病理标注方法及装置、计算机可读存储介质,申请号为2018112615217、申请名称为基于医学图像的报告签发方法及装置、计算机可读存储介质。其全部内容通过引用结合在本申请中。
技术领域
本发明涉及智能医疗领域,尤其涉及一种医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置。
背景技术
机器学习用于医学图像分析前,需要大量的高质量的标注数据。如何确保标注质量结果的可靠性及医生如何在人工智能辅助下签发医学报告,都是未来人工智能应用于临床实践中所要面临和解决的问题。在不断验证和调试人工智能的准确性方面也缺乏系统性的方法。将科学研究中人工智能辅助诊断系统,落地到临床医疗实践中,确保诊断结果的可靠性及准确性,需要医生和人工智能之间有良好的工作流程。发现人工智能现在已存在或者未来可能遇到的问题时,有不断学习和调整的相关机制。
人工智能在前期标注时,一般选择全片整体性标注、大至中等FOV图片的标注,则只能在大容量的计算机上工作,地点受限,无法实现大量医学图像的标注工作,则无法为人工智能提供学习样本。
发明内容
本发明实施例提供一种医学图像的病理标注方法及装置、计算机可读存储介质,能有效解决现有的人工标注工作地点受限的问题,提高标注工作的效率,为人工智能提供质量高且数量丰富的学习样本。
本发明一实施例提供一种医学图像的病理标注方法,包括步骤:
当接收到任一用户在所述移动终端输入的进入标注模式的请求时,判定所述用户是否具有病理标注资格;
当判断所述用户具有病理标注资格中,随机调用待标注的医学图像并在所述移动终端的显示界面进行显示;
接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果;其中,
所述医学图像通过以下步骤生成:
对病理切片的原始扫描图像分割成n个FOV图片后,计算所述n个FOV图片中每一所述FOV图片的病理指数;其中,10^10≥n≥10;所述FOV图片所占用的存储空间小于预设的阈值;
所述n个小图片中每一所述FOV图片的病理指数,获取所述n个FOV图片中病理指数最高的前m个FOV图片作为所述医学图像;其中,50≥m≥5。
与现有技术相比,本发明实施例公开的医学图像的病理标注方法,通过当接收到任一用户在所述移动终端输入的进入标注模式的请求时,判定所述用户是否具有病理标注资格;当判断所述用户具有病理标注资格中,随机调用待标注的医学图像并在所述移动终端的显示界面进行显示;接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果,所述医学图像通过以下步骤生成:对病理切片的原始扫描图像分割成n个FOV图片后,计算所述n个FOV图片中每一所述FOV图片的病理指数;其中,10^10≥n≥10;所述FOV图片所占用的存储空间小于预设的阈值;所述n个小图片中每一所述FOV图片的病理指数,获取所述n个FOV图片中病理指数最高的前m个FOV图片作为所述医学图像;其中,50≥m≥5,标注工作者无需面对全片整体性标注,仅需对分割后的小图片进行标注,减少标注工作者的工作量,降低其疲劳感,提高标注准确率,且在移动终端上操作,可随时随地进行标注,不受地点的限制,有利于提高标注工作的效率,为人工智能提供质量高且数量丰富的学习样本。
作为上述方案的改进,所述方法还包括步骤:
当检测到任一所述医学图像具有两个用户的第一标注结果时,判断所述两个用户的第一标注结果是否一致;
当判断所述两个用户的第一标注结果一致时,以所述两个用户的第一标注结果作为所述医学图像的参考标注结果,并分别为两个所述用户的正确标注数加1。
作为上述方案的改进,所述方法还包括步骤:
当判断所述两个用户的第一标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求;
作为上述方案的改进,所述方法还包括步骤:
当判断所述两个用户的第一标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求;
当接收到所述其他用户或专家用户对所述医学图像的第二标注结果时,将该所述其他用户或专家用户对所述医学图像的第二标注结果作为所述医学图像的参考标注结果,并将所述第一标注结果与所述第二标注结果不一致的用户的错误标注数加1,将所述第一标注结果与所述第二标注结果相一致的用户的正确标注数加1。
作为上述方案的改进,所述方法还包括步骤:
当检测到任一用户的错误标注数大于预设的第一阈值时,取消所述用户的病理标注资格。
作为上述方案的改进,所述方法还包括步骤:
当接收到所述其他用户或专家用户对所述医学图像加入疑难病例库的请求时,将所述医学图像加入疑难病例库。
作为上述方案的改进,所述方法还包括步骤:
在显示所述医学图像时,获取所述医学图像的AI标注结果;其中,所述医学图像的AI标注结果包括阳性结果和阴性结果;
当所述医学图像的AI标注结果为阴性结果时,且任一用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果一致时,以所述医学图像的AI标注结果作为所述医学图像的参考标注结果;
当所述医学图像的AI标注结果为阳性结果时,且任两个用户对所述医学图像的第一标注结果与所述医学图像 的AI标注结果一致时,以所述医学图像的AI标注结果作为所述医学图像的参考标注结果;
当所述医学图像的AI标注结果为阳性结果时,且任两个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果均不一致时,以所述任两个用户对所述医学图像的第一标注结果作为所述医学图像的参考标注结果;
当所述医学图像的AI标注结果为阳性结果时,且任两个用户中其中一个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求或组织讨论请求;
当接收到所述其他用户或专家用户对所述医学图像的第二标注结果时,将该所述其他用户或专家用户对所述医学图像的第二标注结果作为所述医学图像的参考标注结果;
当接收到所述任两个用户和其他用户或专家用户经过讨论后对所述医学图像的第三标注结果时,将所述第三标注结果作为所述医学图像的参考标注结果。
作为上述方案的改进,所述判定所述用户是否具有病理标注资格具体包括:
当接收到所述用户输入的标注资格请求时,显示测试界面,随机调用a张用于测试所述用户的医学图像,获取所述用户对每一所述医学图像的标注结果;其中,所述a张用于测试所述用户的医学图像均具有正确标注结果;其中,a≥5;
将所述用户对每一所述医学图像的标注结果与所述正确标注结果进行比对,获取所述用户对所述a张具有正确标记结果的医学图像的标注正确率;
当判断所述标注正确率大于预设的第二阈值时,则确定所述用户具有标注资格。
作为上述方案的改进,所述判定所述用户是否具有病理标注资格还包括:
当判断所述标注正确率小于预设的第二阈值时,判断所述标注正确率是否大于预设的第三阈值;
当判断所述标注正确率大于预设的第三阈值时,若接收到所述用户的训练请求,显示训练界面,调用x张用于训练所述用户的医学图像;其中,所述x张用于训练所述用户的医学图像均具有正确标注结果;其中,x≥5。
作为上述方案的改进,所述判定所述用户是否具有病理标注资格还包括:
当判断所述标注正确率小于所述第三阈值时,显示学习界面,在所述学习界面上显示关于标注的系统知识。
作为上述方案的改进,所述判定所述用户是否具有病理标注资格还包括:
当判断所述标注正确率小于所述第三阈值时,在预设的时间段内拒绝所述用户的训练请求和标注资格请求;
当判断所述标注正确率小于所述第三阈值时,在预设的时间段后,重新接受所述用户的训练请求和标注资格请求。
作为上述方案的改进,所述接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果包括:
在显示所述医学图像时,在所述显示界面上显示是否有病变细胞和病变细胞类型的第一选项按钮;
当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果。
作为上述方案的改进,所述当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果包括步骤:
在显示所述医学图像时,在所述显示界面上显示其他用户或机器对所述医学图像的标注结果,并同时显示是否同意其他用户或机器对所述医学图像的标注结果的第二选项按钮;
当接收到所述用户对任一所述第二选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果。
作为上述方案的改进,所述第二选项按钮包括同意按钮和反对按钮;
所述当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果具体为:
当接收到所述用户对所述同意按钮的点击操作时,将所述其他用户或机器对所述医学图像的标注结果作为所述用户的第一标注结果;
当接收到所述用户对所述反对按钮的点击操作时,将所述用户重新输入的标注结果作为所述用户的第一标注结果。
作为上述方案的改进,所述方法还包括步骤:
在显示所述第二选项按钮时,同时显示语音输入标记;其中,所述语音输入标记的显示状态包括正在检测语音输入状态和停止检测语音输入状态;
所述当接收到所述用户对所述反对按钮的点击操作时,将所述用户重新输入的标注结果作为所述用户的第一标注结果具体为:
当接收到所述用户对所述反对按钮的点击操作时,将所述语音输入标记的显示状态修改为正在检测语音输入状态,并通过语音设备获取所述用户输入的标注结果作为所述用户的第一标注结果。
作为上述方案的改进,所述语音输入标记为语音输入进度条,当所述语音输入标记的显示状态为停止检测语音输入状态时,在所述显示界面显示处于满格状态的所述语音进度条;当所述语音输入标记的显示状态为正在检测语音输入状态,在所述显示界面在预设的时间段内逐渐缩短的所述语音进度条。
本发明实施例还对应提供了一种基于医学图像的报告签发方法,包括步骤:
当接收到关于任一医学图像的报告签发指令时,获取所述医学图像的AI判读结果;其中,所述医学图像的AI判读结果包括阴性结果和阳性结果;其中,所述医学图像的AI判读结果通过一预设的分类器进行判读后获得,所述分类器通过如上述任一项所述的所述医学图像的参考标注结果进行训练;
当判断所述医学图像的AI判读结果为阴性结果时,向任一医生用户请求所述医学图像的人工判读结果;
当判断所述医学图像的AI判读结果为阴性结果,且所述医生用户对所述医学图像的人工判读结果与所述AI判读结果一致时,签发关于所述医学图像的报告单,并在所述报告单中标明阴性结果;
当判断所述医学图像的AI判读结果为阳性结果时,向任两个医生用户请求所述医学图像的人工判读结果;
当判断所述医学图像的AI判读结果为阳性结果,且所述两个医生用户对所述医学图像的人工判读结果均与所述AI判读结果一致时,签发关于所述医学图像的报告单,并在所述报告单中标明阳性结果。
作为上述方案的改进,所述方法还包括步骤:
当判断所述医学图像的AI判读结果为阳性结果,且所述两个医生用户中存在至少一个医生用户对所述医学图像的人工判读结果均与所述AI判读结果不一致时,向其他医生用户或专家医生用户发送所述医学图像的判读请求或组织讨论请求。
作为上述方案的改进,所述方法还包括步骤:
当接收到所述其他医生用户或专家医生用户对所述医学图像的人工判读结果时,签发关于所述医学图像的报告单,并在所述报告单中标明所述其他用户或专家用户对所述医学图像的人工判读结果。
作为上述方案的改进,所述方法还包括步骤:
当接收到所述两个医生用户和所述其他医生用户或专家医生用户经过讨论后发送的最终人工判读结果时,签发关于所述医学图像的报告单,并在所述报告单中标明所述最终判读结果。
本发明另一实施例提供了一种医学图像的病理标注装置,包括处理器、存储器以及存储在所述存储器中且被配置为由所述处理器执行的计算机程序,所述处理器执行所述计算机程序时实现上述任一发明实施例所述的医学图像的病理标注方法。
本发明另一实施例提供了一种计算机可读存储介质,所述计算机可读存储介质包括存储的计算机程序,其中,在所述计算机程序运行时控制所述计算机可读存储介质所在设备执行上述任一发明实施例所述的医学图像的病理标注方法。
本发明另一实施例提供了一种医学图像的病理标注装置,包括处理器、存储器以及存储在所述存储器中且被配置为由所述处理器执行的计算机程序,所述处理器执行所述计算机程序时实现上述任一发明实施例所述的基于医学图像的报告签发方法。
本发明另一实施例提供了一种计算机可读存储介质,所述计算机可读存储介质包括存储的计算机程序,其中,在所述计算机程序运行时控制所述计算机可读存储介质所在设备执行上述任一发明实施例所述的基于医学图像的报告签发方法。
附图说明
图1是本发明一实施例提供的一种医学图像的病理标注方法的流程示意图。
图2是本发明一实施例提供的病理切片的原始扫描图像的切割示意图。
图3(a)是宫颈细胞学单细胞切割小图片,图3(b)是宫颈细胞学细胞团切割小图片。
图4是本发明一实施例提供的移动终端的显示界面示意图。
图5是本发明另一实施例提供的移动终端的显示界面示意图。
图6是本发明实施例提供的确认所述用户是否具有病理标注资格的流程示意图。
图7是本发明实施例提供的一种基于医学图像的报告签发方法的流程示意图。
图8是本发明一实施例提供的一种医学图像的病理标注装置的结构示意图。
图9是本发明实施例提供的终端设备的结构示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
参见图1,是本发明一实施例提供的一种医学图像的病理标注方法的流程示意图,其使用于一移动终端,包括步骤:
S1、当接收到任一用户在所述移动终端输入的进入标注模式的请求时,判定所述用户是否具有病理标注资格;
在步骤S1中,用户需经过专门的测试才能授予其病理标注资格,例如需随机调用s张用于测试所述用户的医学图像,获取所述用户对每一所述医学图像的标注结果,将所述用户对每一所述医学图像的标注结果与所述正确标注结果进行比对,获取所述用户对所述s张(s≥10)具有正确标记结果的医学图像的标注正确率,当判断所述标注正确率大于预设的第一阈值时,则确定所述用户具有病理标注资格。通过统一的测试对标注工作者进行筛选,可保证了较高的医疗水平,标注质量的一致性和稳定性,能够保证标注质量,能获得客观性的科学评价指标,利于后续的机器学习模型构建。
S2、当判断所述用户具有病理标注资格中,随机调用待标注的医学图像并在所述移动终端的显示界面进行显示;
在步骤S1中,优选地,所述医学图像通过以下方式生成:
首先,如图2所示,对病理切片的原始扫描图像分割成n个FOV图片后,计算所述n个FOV图片中每一所述FOV图片的病理指数;其中,10^10≥n≥10;所述FOV图片所占用的存储空间小于预设的阈值;
所述n个小图片中每一所述FOV图片的病理指数,获取所述n个FOV图片中病理指数最高的前m个FOV图片作为所述医学图像;其中,50≥m≥5。
上述计算所述n个小图片中每一所述小图片的病理指数由人工智能完成,例如通过人工智能可从大量图像中优先筛选排序呈现20张图像,可大幅度降低医生的工作量。
一般来说,原始扫描图像通过将细胞病理玻片,通过数字切片扫描仪,转化为X400倍的数字图像,其大小可达到几个G到几十个G。如果标注工作者通过对全片进行标注,工作量非常大,而且只能在大容量的计算机上工作,地点受限,因此,原有的标注工作存在漏标注、标注步骤繁琐和工作量巨大等问题。因此,通过对原始数字图像进行切割后,可使标注工作者能随时随地进行标注,切割后的图像无需在大容量的计算机才能加载,降低标注工作者的工作量。如图3(a)所示,为宫颈细胞学单细胞切割小图片,如图3(b)所示,为宫颈细胞学细胞团切割小图片。
另外,小图片可以方便在互联网上下载至不同的接收平台下,可便于在云端快速响应到移动端设备中,实现快速的操作,从而完成标注的过程。因此,标注工作者可在闲暇的空余时间(例如,在坐车、等车过程)使用移动终端(例如,手机等)进行测试标注,也可在家或办公室使用一般的PC进行标注,可在不同设备上同步性操作,大大增加了便捷性,最大时间利用用户的零碎时间操作。
S3、接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果。
例如,如图4所示,在移动终端的显示界面上,除了显示医学图像外,还在该医学图像周围显示是否有恶性细胞和恶性细胞类型的第一选项按钮(包括ASC、LSIL、AGC、HSIL等),当接收到用户对任一第一选项按钮的选择操作时,保存该操作记录,并自动前往下一张医学图像。一般来说,标注结果为LSIL的概率最高,因此,系统首先默认用户选择LSIL,并按照病变概率分布的特点调节多种标签选项的位置,提高了系统操作的友好型和便捷性。
在显示所述医学图像时,在所述显示界面上显示其他用户或机器对所述医学图像的标注结果,并同时显示是否同意其他用户或机器对所述医学图像的标注结果的第二选项按钮;
当接收到所述用户对任一所述第二选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果。
所述第二选项按钮包括同意按钮和反对按钮;
所述当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果具体为:
当接收到所述用户对所述同意按钮的点击操作时,将所述其他用户或机器对所述医学图像的标注结果作为所述用户的第一标注结果;
当接收到所述用户对所述反对按钮的点击操作时,将所述用户重新输入的标注结果作为所述用户的第一标注结果。
例如,如图5所示,在移动终端的显示界面上,除了显示医学图像外,示有前期人工标注结果或者机器标注结果,同时显示是否同意其他用户或机器对所述医学图像的标注结果的第二选项按钮(即图中的“YES”和“NO”按钮),正确,点击YES,保存并进入下一张。错误,点击NO,重新输入新的标注结果。通过这种复核的方式,提高了纠正错误的快捷性,最大限度的降低人操作的复杂性。
由图4图5显示界面可知,该方案还兼顾了标注界面的友好性及游戏性,可降低标注病理医师的不适感及疲劳性。
在显示所述第二选项按钮时,同时显示语音输入标记;其中,所述语音输入标记的显示状态包括正在检测语音输入状态和停止检测语音输入状态;
所述当接收到所述用户对所述反对按钮的点击操作时,将所述用户重新输入的标注结果作为所述用户的第一标注结果具体为:
当接收到所述用户对所述反对按钮的点击操作时,将所述语音输入标记的显示状态修改为正在检测语音输入状态,并通过语音设备获取所述用户输入的标注结果作为所述用户的第一标注结果。
优选地,所述语音输入标记为语音输入进度条,当所述语音输入标记的显示状态为停止检测语音输入状态时,在所述显示界面显示处于满格状态的所述语音进度条;当所述语音输入标记的显示状态为正在检测语音输入状态,在所述显示界面在预设的时间段内逐渐缩短的所述语音进度条。
例如,如图5示,当标注工作者点击“NO”时,显示5-10s倒计时语音纠错功能,进度条处于逐渐缩短的状态。例如一张LSIL的片子,之前机器标注ASC-US,标注工作者说出LSIL的发音,移动终端则自动纠正。
在一优选实施例中,当检测到任一所述医学图像具有两个用户的第一标注结果时,判断所述两个用户的第一标注结果是否一致;
当判断所述两个用户的第一标注结果一致时,以所述两个用户的第一标注结果作为所述医学图像的参考标注结果,并分别为两个所述用户的正确标注数加1;
当判断所述两个用户的第一标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求;
当接收到所述其他用户或专家用户对所述医学图像的第二标注结果时,将该所述其他用户或专家用户对所述医学图像的第二标注结果作为所述医学图像的参考标注结果,并将所述第一标注结果与所述第二标注结果不一致的用户的错误标注数加1,将所述第一标注结果与所述第二标注结果相一致的用户的正确标注数加1。
通过以上方案,对于同一医学图像,随机分配两个不同标注者,对同一病例进行在线评价标注,双复核制度,也避免了因人工智能在设计上存在的系统性误差,确保了标注的可靠性及准确性。
另外,当检测到任一用户的错误标注数大于预设的第一阈值时,取消所述用户的病理标注资格。加入工作监督机制和淘汰机制,避免了由于个人随意性操作及诊断能力不足可能存在的漏洞。当检测到任一用户的正确标注数大于预设的第二阈值时,将该用户加入专业标注工作者库中;当检测到任一用户的准确标注数大于预设的第三阈值时,且判断该用户具备执业医师证时,授予其签发报告资格。另外,还可在显示界面显示正确标注数的用户排名。通过以上激励策略,有利于提高用户标注的效率,实现大量的医学图像标注。
除请求他用户或专家用户对该医学图像进行标注外,还可组织更多的专家讨论达成共识。另外,当接收到所述其他用户或专家用户对所述医学图像加入疑难病例库的请求时,将所述医学图像加入疑难病例库作为非典型案例的研究使用。
而在另一实施例中,所述方法还包括步骤:
在显示所述医学图像时,获取所述医学图像的AI标注结果;其中,所述医学图像的AI标注结果包括阳性结果和阴性结果;
当所述医学图像的AI标注结果为阴性结果时,且任一用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果一致时,以所述医学图像的AI标注结果作为所述医学图像的参考标注结果;
当所述医学图像的AI标注结果为阳性结果时,且任两个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果一致时,以所述医学图像的AI标注结果作为所述医学图像的参考标注结果;
当所述医学图像的AI标注结果为阳性结果时,且任两个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果均不一致时,以所述任两个用户对所述医学图像的第一标注结果作为所述医学图像的参考标注结果;
当所述医学图像的AI标注结果为阳性结果时,且任两个用户中其中一个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求或组织讨论请求;
当接收到所述其他用户或专家用户对所述医学图像的第二标注结果时,将该所述其他用户或专家用户对所述医学图像的第二标注结果作为所述医学图像的参考标注结果;
当接收到所述任两个用户和其他用户或专家用户经过讨论后对所述医学图像的第三标注结果时,将所述第三标注结果作为所述医学图像的参考标注结果。
综上,通过实施本发明的方案,能在保证标注质量的前提下,使得标注工作者可随时随地对医学图像进行标注,工作地点不受限制,改善了低效的工作模式,明显提升医疗服务质量及医疗工作效率,为人工智能提供质量高且数量丰富的学习样本。除此之外,对于每一用户,其获取到的医学图像是随机的,而对于每一医学图像,分配给的用户也是随机的,在最大程度的实现了医学自然图像判读上的客观化,降低了不同地区和不同医院的诊疗差异性,保证了更多的患者受益,更多的医疗公平。
在一优选实施例中,参见图6,确认所述用户是否具有病理标注资格具体包括:
S11、当接收到所述用户输入的标注资格请求时,显示测试界面,随机调用a张用于测试所述用户的医学图像,获取所述用户对每一所述医学图像的标注结果;其中,所述a张用于测试所述用户的医学图像均具有正确标注结果;其中,a≥5;
S12、将所述用户对每一所述医学图像的标注结果与所述正确标注结果进行比对,获取所述用户对所述a张具有正确标记结果的医学图像的标注正确率;
在实际操作过程中,测试数量a可不做具体的限制,当超过一定数量时,可以理解的,用户在越小的测试数量下正确个数越多越好。
在步骤S12中,所述获取所述用户对所述a张具有正确标记结果的医学图像的标注正确率:
将所述用户对所述医学图像的标注结果与所述正确标注结果相一致的个数与a相除,获取所述用户对所述a张具有正确标记结果的医学图像的标注正确率。
S13、当判断所述标注正确率大于预设的第二阈值时,则确定所述用户具有标注资格。
例如,当所述标注正确率大于0.95,可认为标注医生的医疗水平较高,一致性较为稳定,确定该用户具有标注资格,可成功作为训练机器学习的标注工作者入会成功。
因此,在本实施例中,通过统一的测试对标注工作者进行筛选,可保证了较高的医疗水平,标注质量的一致性和稳定性。另外,通过实施本方案,使得标注工作者的筛选,以质量为优先,而非以资历为优先原则,避免根据高资历医生作为标注工作者筛选标准存在的主观偏差,例如高年资的标注医生往往专注于一类亚专科,对于其他亚专科可能存在一定的盲区。因此,该方案证标注质量,能获得客观性的科学评价指标,利于后续的机器学习模型构建。
在另一优选实施例中,当判断所述标注正确率小于预设的第二阈值时,判断所述标注正确率是否大于预设的第三阈值;
当判断所述标注正确率大于预设的第三阈值时,若接收到所述用户的训练请求,显示训练界面,调用x张用于训练所述用户的医学图像;其中,所述x张用于训练所述用户的医学图像均具有正确标注结果;其中,x≥5。
例如,当用户测试的标注正确率小于0.95,大于0.50,可能因为操作上的失误、对个别病变存在认识上的误区等原因,可先申请在线测试题学习,然后可申请再次初步筛选的入门测试。如果正确率大于0.95,则可确定该用户具有标注资格,如果仍然小于0.95,大于0.50则仍需进入训练模式进行训练后方能重新发起标注资格请求。
另外,当判断所述标注正确率小于所述第三阈值时,在预设的时间段内拒绝所述用户的训练请求和标注资格请求,即封禁该用户的账号,并显示学习界面,在所述学习界面上显示关于标注的系统知识,以供该用户进行全面的理论知识培训。而在预设的时间段后,重新接受所述用户的训练请求和标注资格请求。
例如,当所述标注正确率小于0.50,可能是由于诊断水平上不足等原因所致,建议进行该亚专科的系统性学习后。账号禁止后一段时间(例如,24-48h)后,用户可再次申请在线测试题学习及初步筛选的入门测试。
参见图7,本发明实施例还对应提供了一种基于医学图像的报告签发方法,包括步骤:
S71、任一医学图像的报告签发指令时,获取所述医学图像的AI判读结果;其中,所述医学图像的AI判读结果包括阴性结果和阳性结果;其中,所述医学图像的AI判读结果通过一预设的分类器进行判读后获得,所述分类 器通过如所述提供的医学图像的的参考标注结果进行训练;
S72、当判断所述医学图像的AI判读结果为阴性结果时,向任一医生用户请求所述医学图像的人工判读结果;
具体的,医生用户即具有签发资格的用户,一般其具有执业医师证。
由于所述医学图像为切割后的小图片,其可加载至医生用户所持的移动终端进行显示。因此,医生用户可在闲暇的空余时间(例如,在坐车、等车过程)使用移动终端(例如,手机等)进行测试标注,也可在家或办公室使用一般的PC进行判读,可在不同设备上同步性操作,大大增加了便捷性,最大时间利用医生用户的零碎时间操作。而且,通过人工智能筛选过后的小图片,能大大降低医生用户的工作量,并降低医生用户由于疲劳而容易出错的风险。
S73、当判断所述医学图像的AI判读结果为阴性结果,且所述医生用户对所述医学图像的人工判读结果与所述AI判读结果一致时,签发关于所述医学图像的报告单,并在所述报告单中标明阴性结果;
在步骤S73中,由于阴性结果的概率远远高于阳性结果,因此,其出错率较低,无需多个医生用户复核即可出单。
S74、当判断所述医学图像的AI判读结果为阳性结果时,向任两个医生用户请求所述医学图像的人工判读结果;
在步骤S74中,由通过随机向任两个医生用户符合AI判读结果,保证了阳性结果的客观性,避免一个医生用户的偏见等导致结果出错。且在医学领域,阳性结果的概率较低,需谨慎对待,因此,需要两个医生进行复核才可出单。
S75、当判断所述医学图像的AI判读结果为阳性结果,且所述两个医生用户对所述医学图像的人工判读结果均与所述AI判读结果一致时,签发关于所述医学图像的报告单,并在所述报告单中标明阳性结果。
可以理解的,当两个医生用户均同意AI判读结果时,则证明AI判读结果具有较高的客观性和准确性,可签发关于所述医学图像的报告单。
对于医生用户所持的移动终端,当接受到请求人工判读的请求时,在显示所述医学图像时,在所述显示界面上显示所述医学图像的AI判读结果,并同时显示是否同意所述医学图像的AI判读结果的第三选项按钮;其中,所述第二选项按钮包括同意按钮和反对按钮;
当所接收到所述医生用户对所述同意按钮的点击操作时,则确定该医生用户的人工判读结果与所述医学图像的AI判读结果一致;
当接收到所述医生用户对所述反对按钮的点击操作时,则确定该医生用户的人工判读结果与所述医学图像的AI判读结果不一致。
本发明实施例在实施过程中,能将人工智能应用于临床实践的报告签发中,利于医学智能化的发展,提高报告签发的效率,确保在人工智能辅助诊断下的医疗安全,在患者人数增加时,医疗质量不降低而提升。
此外,在上述实施例的基础上,另一实施例还包括步骤:
当判断所述医学图像的AI判读结果为阳性结果,且所述两个医生用户中存在至少一个医生用户对所述医学图像的人工判读结果均与所述AI判读结果不一致时,向其他医生用户或专家医生用户发送所述医学图像的判读请求或组织讨论请求;
当接收到所述其他医生用户或专家医生用户对所述医学图像的人工判读结果时,签发关于所述医学图像的报告单,并在所述报告单中标明所述其他用户或专家用户对所述医学图像的人工判读结果;
当接收到所述两个医生用户和所述其他医生用户或专家医生用户经过讨论后发送的最终人工判读结果时,签发关于所述医学图像的报告单,并在所述报告单中标明所述最终判读结果。
需要说明的是,本公开中的标注用户和医生用户有所区别,标注用户仅具有病理标注资格,而医生用户还具有报告签发资格,其对报告的内容负责。用户需经过专门的测试才能授予其病理标注资格,例如需随机调用a张(a≥10)用于测试所述用户的医学图像,获取所述用户对每一所述医学图像的标注结果,将所述用户对每一所述医学图像的标注结果与所述正确标注结果进行比对,获取所述用户对所述a张具有正确标记结果的医学图像的标注正确率,当判断所述标注正确率大于预设的第二阈值时,则确定所述用户具有病理标注资格。通过统一的测试对标注工作者进行筛选,可保证了较高的医疗水平,标注质量的一致性和稳定性,能够保证标注质量,能获得客观性的科学评价指标,利于后续的机器学习模型构建。
在另一优选实施例中,所述方法还包括步骤:
当任一标注用户对任一医学图像的人工标注结果和所述参考标注结果相一致时,则为所述用户的正确标注数加1;
当所述标注用户的正确标注数大于预设的阈值时,将所述标注用户加入医生用户库中。
将优秀的标注用户加入医生用户库中,当其具有执业医师证时,可申请正式的医生用户,具有报告签发资格,能克服现有医生用户严重不足和超负荷工作的问题,大大缓解现有的报告签发压力,提高报告签发的效率。
参见图8,为本发明实施例提供的一种医学图像的病理标注装置的结构示意图,包括:
病理标注资格获取模块801,用于当接收到任一用户在所述移动终端输入的进入标注模式的请求时,判定所述用户是否具有病理标注资格;
医学图像显示模块802,用于当判断所述用户具有病理标注资格中,随机调用待标注的医学图像并在所述移动终端的显示界面进行显示;
第一标注结果保存模块803,用于接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果。
本发明实施例的医学图像的病理标注装置的实施过程和工作原理可参考上述任一项对医学图像的病理标注方法的描述,在此不再赘述。
本发明另一实施例提供了一种医学图像的病理标注装置,包括处理器、存储器以及存储在所述存储器中且被配置为由所述处理器执行的计算机程序,所述处理器执行所述计算机程序时实现上述任一发明实施例所述的医学图像的病理标注方法或基于医学图像的报告签发方法。
本发明另一实施例提供了一种计算机可读存储介质,所述计算机可读存储介质包括存储的计算机程序,其中,在所述计算机程序运行时控制所述计算机可读存储介质所在设备执行上述任一发明实施例所述的医学图像的病理标注方法或基于医学图像的报告签发方法。
参见图9,是本发明实施例提供的终端设备的示意图。所述终端设备包括:至少一个处理器11,例如CPU,至少一个网络接口14或者其他用户接口13,存储器15,至少一个通信总线12,通信总线12用于实现这些组件之间 的连接通信。其中,用户接口13可选的可以包括USB接口以及其他标准接口、有线接口。网络接口14可选的可以包括Wi-Fi接口以及其他无线接口。存储器15可能包含高速RAM存储器,也可能还包括非不稳定的存储器(non-volatilememory),例如至少一个磁盘存储器。存储器15可选的可以包含至少一个位于远离前述处理器11的存储装置。
在一些实施方式中,存储器15存储了如下的元素,可执行模块或者数据结构,或者他们的子集,或者他们的扩展集:
操作系统151,包含各种系统程序,如电池管理系统等等,用于实现各种基础业务以及处理基于硬件的任务;
程序152。
具体地,处理器11用于调用存储器15中存储的程序152,执行上述实施例所述医学图像的病理标注方法或基于医学图像的报告签发方法,例如图1所示的步骤S1。或者,所述处理器11执行所述计算机程序时实现上述各装置实施例中各模块/单元的功能,例如医学图像显示模块802。
示例性的,所述计算机程序可以被分割成一个或多个模块/单元,所述一个或者多个模块/单元被存储在所述存储器中,并由所述处理器执行,以完成本发明。所述一个或多个模块/单元可以是能够完成特定功能的一系列计算机程序指令段,该指令段用于描述所述计算机程序在所述终端设备中的执行过程。
所述终端设备可包括,但不仅限于,处理器11、存储器15。本领域技术人员可以理解,所述示意图仅仅是终端设备的示例,并不构成对终端设备的限定,可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件,例如所述医学图像的病理标注装置还可以包括输入输出设备、网络接入设备、总线等。
所称处理器11可以是中央处理单元(Central Processing Unit,CPU),还可以是其他通用处理器、数字信号处理器(Digital Signal Processor,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现成可编程门阵列(Field-Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等,所述处理器11是所述终端设备的控制中心,利用各种接口和线路连接整个终端设备的各个部分。
所述存储器15可用于存储所述计算机程序和/或模块,所述处理器11通过运行或执行存储在所述存储器内的计算机程序和/或模块,以及调用存储在存储器内的数据,实现所述终端设备的各种功能。所述存储器15可主要包括存储程序区和存储数据区,其中,存储程序区可存储操作系统、至少一个功能所需的应用程序(比如声音播放功能、图像播放功能等)等;存储数据区可存储根据手机的使用所创建的数据(比如音频数据、电话本等)等。此外,存储器15可以包括高速随机存取存储器,还可以包括非易失性存储器,例如硬盘、内存、插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)、至少一个磁盘存储器件、闪存器件、或其他易失性固态存储器件。
其中,所述终端设备集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。基于这样的理解,本发明实现上述实施例方法中的全部或部分流程,也可以通过计算机程序来指令相关的硬件来完成,所述的计算机程序可存储于一计算机可读存储介质中,该计算机程序在被处理器执行时,可实现上述各个方法实施例的步骤。其中,所述计算机程序包括计算机程序代码,所述计算机程序代码可以为源代码形式、对象代码形式、可执行文件或某些中间形式等。所述计算机可读介质可以包括:能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、以及软件分发介质等。需要说明的是,所述计算机可读介质包含的内容可以根据司法管辖区内立法和专利实践的要求进行适当的增减。
以上所述是本发明的优选实施方式,应当指出,对于本技术领域的普通技术人员来说,在不脱离本发明原理的 前提下,还可以做出若干改进和润饰,这些改进和润饰也视为本发明的保护范围。

Claims (24)

  1. 一种医学图像的病理标注方法,其特征在于,适用于一移动终端,包括步骤:
    当接收到任一用户在所述移动终端输入的进入标注模式的请求时,判定所述用户是否具有病理标注资格;
    当判断所述用户具有病理标注资格中,随机调用待标注的医学图像并在所述移动终端的显示界面进行显示;
    接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果;其中,
    所述医学图像通过以下步骤生成:
    对病理切片的原始扫描图像分割成n个FOV图片后,计算所述n个FOV图片中每一所述FOV图片的病理指数;其中,10^10≥n≥10;所述FOV图片所占用的存储空间小于预设的阈值;
    所述n个小图片中每一所述FOV图片的病理指数,获取所述n个FOV图片中病理指数最高的前m个FOV图片作为所述医学图像;其中,50≥m≥5。
  2. 如权利要求1所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    当检测到任一所述医学图像具有两个用户的第一标注结果时,判断所述两个用户的第一标注结果是否一致;
    当判断所述两个用户的第一标注结果一致时,以所述两个用户的第一标注结果作为所述医学图像的参考标注结果,并分别为两个所述用户的正确标注数加1。
  3. 如权利要求2所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    当判断所述两个用户的第一标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求;
    当接收到所述其他用户或专家用户对所述医学图像的第二标注结果时,将该所述其他用户或专家用户对所述医学图像的第二标注结果作为所述医学图像的参考标注结果,并将所述第一标注结果与所述第二标注结果不一致的用户的错误标注数加1,将所述第一标注结果与所述第二标注结果相一致的用户的正确标注数加1。
  4. 如权利要求3所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    当检测到任一用户的错误标注数大于预设的第一阈值时,取消所述用户的病理标注资格。
  5. 如权利要求3所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    当接收到所述其他用户或专家用户对所述医学图像加入疑难病例库的请求时,将所述医学图像加入疑难病例库。
  6. 如权利要求1所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    在显示所述医学图像时,获取所述医学图像的AI标注结果;其中,所述医学图像的AI标注结果包括阳性结果和阴性结果;
    当所述医学图像的AI标注结果为阴性结果时,且任一用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果一致时,以所述医学图像的AI标注结果作为所述医学图像的参考标注结果;
    当所述医学图像的AI标注结果为阳性结果时,且任两个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果一致时,以所述医学图像的AI标注结果作为所述医学图像的参考标注结果;
    当所述医学图像的AI标注结果为阳性结果时,且任两个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果均不一致时,以所述任两个用户对所述医学图像的第一标注结果作为所述医学图像的参考标注结果。
  7. 如权利要求6所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    当所述医学图像的AI标注结果为阳性结果时,且任两个用户中其中一个用户对所述医学图像的第一标注结果与所述医学图像的AI标注结果不一致时,随机向任一其他用户或专家用户发送所述医学图像的标注请求或组织讨论请求;
    当接收到所述其他用户或专家用户对所述医学图像的第二标注结果时,将该所述其他用户或专家用户对所述医学图像的第二标注结果作为所述医学图像的参考标注结果;
    当接收到所述任两个用户和其他用户或专家用户经过讨论后对所述医学图像的第三标注结果时,将所述第三标注结果作为所述医学图像的参考标注结果。
  8. 如权利要求1所述的医学图像的病理标注方法,其特征在于,所述判定所述用户是否具有病理标注资格具体包括:
    当接收到所述用户输入的标注资格请求时,显示测试界面,随机调用a张用于测试所述用户的医学图像,获取所述用户对每一所述医学图像的标注结果;其中,所述a张用于测试所述用户的医学图像均具有正确标注结果;其中,a≥5;
    将所述用户对每一所述医学图像的标注结果与所述正确标注结果进行比对,获取所述用户对所述a张具有正确标记结果的医学图像的标注正确率;
    当判断所述标注正确率大于预设的第二阈值时,则确定所述用户具有标注资格。
  9. 如权利要求8所述的医学图像的病理标注方法,其特征在于,所述判定所述用户是否具有病理标注资格还包括:
    当判断所述标注正确率小于预设的第二阈值时,判断所述标注正确率是否大于预设的第三阈值;
    当判断所述标注正确率大于预设的第三阈值时,若接收到所述用户的训练请求,显示训练界面,调用x张用于训练 所述用户的医学图像;其中,所述x张用于训练所述用户的医学图像均具有正确标注结果;其中,x≥5。
  10. 如权利要求9所述的医学图像的病理标注方法,其特征在于,所述判定所述用户是否具有病理标注资格还包括:
    当判断所述标注正确率小于所述第三阈值时,显示学习界面,在所述学习界面上显示关于标注的系统知识。
  11. 如权利要求10所述的医学图像的病理标注方法,其特征在于,所述判定所述用户是否具有病理标注资格还包括:
    当判断所述标注正确率小于所述第三阈值时,在预设的时间段内拒绝所述用户的训练请求和标注资格请求;
    当判断所述标注正确率小于所述第三阈值时,在预设的时间段后,重新接受所述用户的训练请求和标注资格请求。
  12. 如权利要求1所述的医学图像的病理标注方法,其特征在于,所述接收所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果包括:
    在显示所述医学图像时,在所述显示界面上显示是否有病变细胞和病变细胞类型的第一选项按钮;
    当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果,并保存所述用户对所述医学图像的第一标注结果。
  13. 如权利要求12所述的医学图像的病理标注方法,其特征在于,所述当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果包括步骤:
    在显示所述医学图像时,在所述显示界面上显示其他用户或机器对所述医学图像的标注结果,并同时显示是否同意其他用户或机器对所述医学图像的标注结果的第二选项按钮;
    当接收到所述用户对任一所述第二选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果。
  14. 如权利要求13所述的医学图像的病理标注方法,其特征在于,所述第二选项按钮包括同意按钮和反对按钮;
    所述当接收到所述用户对任一所述第一选项按钮的点击操作时,获得所述用户对所述医学图像的第一标注结果具体为:
    当接收到所述用户对所述同意按钮的点击操作时,将所述其他用户或机器对所述医学图像的标注结果作为所述用户的第一标注结果;
    当接收到所述用户对所述反对按钮的点击操作时,将所述用户重新输入的标注结果作为所述用户的第一标注结果。
  15. 如权利要求14所述的医学图像的病理标注方法,其特征在于,所述方法还包括步骤:
    在显示所述第二选项按钮时,同时显示语音输入标记;其中,所述语音输入标记的显示状态包括正在检测语音输入状态和停止检测语音输入状态;
    所述当接收到所述用户对所述反对按钮的点击操作时,将所述用户重新输入的标注结果作为所述用户的第一标注结果具体为:
    当接收到所述用户对所述反对按钮的点击操作时,将所述语音输入标记的显示状态修改为正在检测语音输入状态,并通过语音设备获取所述用户输入的标注结果作为所述用户的第一标注结果。
  16. 如权利要求15所述的医学图像的病理标注方法,其特征在于,所述语音输入标记为语音输入进度条,当所述语音输入标记的显示状态为停止检测语音输入状态时,在所述显示界面显示处于满格状态的所述语音进度条;当所述语音输入标记的显示状态为正在检测语音输入状态,在所述显示界面在预设的时间段内逐渐缩短的所述语音进度条。
  17. 一种基于医学图像的报告签发方法,其特征在于,包括步骤:
    当接收到关于任一医学图像的报告签发指令时,获取所述医学图像的AI判读结果;其中,所述医学图像的AI判读结果包括阴性结果和阳性结果;其中,所述医学图像的AI判读结果通过一预设的分类器进行判读后获得,所述分类器通过如权利要求2至7任一项所述的所述医学图像的参考标注结果进行训练;
    当判断所述医学图像的AI判读结果为阴性结果时,向任一医生用户请求所述医学图像的人工判读结果;
    当判断所述医学图像的AI判读结果为阴性结果,且所述医生用户对所述医学图像的人工判读结果与所述AI判读结果一致时,签发关于所述医学图像的报告单,并在所述报告单中标明阴性结果;
    当判断所述医学图像的AI判读结果为阳性结果时,向任两个医生用户请求所述医学图像的人工判读结果;
    当判断所述医学图像的AI判读结果为阳性结果,且所述两个医生用户对所述医学图像的人工判读结果均与所述AI判读结果一致时,签发关于所述医学图像的报告单,并在所述报告单中标明阳性结果。
  18. 如权利要求17所述的基于医学图像的报告签发方法,其特征在于,所述方法还包括步骤:
    当判断所述医学图像的AI判读结果为阳性结果,且所述两个医生用户中存在至少一个医生用户对所述医学图像的人工判读结果均与所述AI判读结果不一致时,向其他医生用户或专家医生用户发送所述医学图像的判读请求或组织讨论请求。
  19. 如权利要求18所述的基于医学图像的报告签发方法,其特征在于,所述方法还包括步骤:
    当接收到所述其他医生用户或专家医生用户对所述医学图像的人工判读结果时,签发关于所述医学图像的报告 单,并在所述报告单中标明所述其他用户或专家用户对所述医学图像的人工判读结果。
  20. 如权利要求19所述的基于医学图像的报告签发方法,其特征在于,所述方法还包括步骤:
    当接收到所述两个医生用户和所述其他医生用户或专家医生用户经过讨论后发送的最终人工判读结果时,签发关于所述医学图像的报告单,并在所述报告单中标明所述最终判读结果。
  21. 一种医学图像的病理标注装置,包括处理器、存储器以及存储在所述存储器中且被配置为由所述处理器执行的计算机程序,所述处理器执行所述计算机程序时实现如权利要求1至16中任意一项所述的医学图像的病理标注方法。
  22. 一种基于医学图像的报告签发装置,包括处理器、存储器以及存储在所述存储器中且被配置为由所述处理器执行的计算机程序,所述处理器执行所述计算机程序时实现如权利要求17至20中任意一项所述的基于医学图像的报告签发方法。
  23. 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质包括存储的计算机程序,其中,在所述计算机程序运行时控制所述计算机可读存储介质所在设备执行如权利要求1至16中任意一项所述的医学图像的病理标注方法。
  24. 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质包括存储的计算机程序,其中,在所述计算机程序运行时控制所述计算机可读存储介质所在设备执行如权利要求17至20中任意一项所述的基于医学图像的报告签发方法。
PCT/CN2019/073339 2018-10-26 2019-01-28 医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置、计算机可读存储介质 WO2020082638A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/628,683 US11094411B2 (en) 2018-10-26 2019-01-28 Methods and devices for pathologically labeling medical images, methods and devices for issuing reports based on medical images, and computer-readable storage media

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
CN201811260183.5A CN109473160A (zh) 2018-10-26 2018-10-26 医学图像的标注资格确定方法及装置
CN201811260183.5 2018-10-26
CN201811261521.7A CN109509541B (zh) 2018-10-26 2018-10-26 基于医学图像的报告签发方法及装置
CN201811260195.8 2018-10-26
CN201811260198.1 2018-10-26
CN201811260195.8A CN109461147B (zh) 2018-10-26 2018-10-26 应用于移动终端的fov图片的病理标注方法及装置
CN201811261521.7 2018-10-26
CN201811260198.1A CN109446370A (zh) 2018-10-26 2018-10-26 医学图像的病理标注方法及装置、计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2020082638A1 true WO2020082638A1 (zh) 2020-04-30

Family

ID=70331695

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/073339 WO2020082638A1 (zh) 2018-10-26 2019-01-28 医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置、计算机可读存储介质

Country Status (2)

Country Link
US (1) US11094411B2 (zh)
WO (1) WO2020082638A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111784284A (zh) * 2020-06-15 2020-10-16 杭州思柏信息技术有限公司 一种宫颈图像多人协同标记云服务系统及云服务方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114783575B (zh) * 2022-04-20 2023-09-29 广州唯顶软件科技有限公司 一种医疗用图像处理系统及方法

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101727674A (zh) * 2008-10-30 2010-06-09 北大方正集团有限公司 一种文档中图片标注及再现标注的方法
CN104462738A (zh) * 2013-09-24 2015-03-25 西门子公司 一种标注医学图像的方法、装置和系统
US9165360B1 (en) * 2012-09-27 2015-10-20 Zepmed, Llc Methods, systems, and devices for automated analysis of medical scans
CN108463814A (zh) * 2015-12-17 2018-08-28 北京安宁福祉科技有限公司 一种医学图像标注方法及系统
CN109446370A (zh) * 2018-10-26 2019-03-08 广州金域医学检验中心有限公司 医学图像的病理标注方法及装置、计算机可读存储介质
CN109461147A (zh) * 2018-10-26 2019-03-12 广州金域医学检验中心有限公司 应用于移动终端的fov图片的病理标注方法及装置
CN109473160A (zh) * 2018-10-26 2019-03-15 广州金域医学检验中心有限公司 医学图像的标注资格确定方法及装置
CN109509541A (zh) * 2018-10-26 2019-03-22 广州金域医学检验中心有限公司 基于医学图像的报告签发方法及装置

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014137262A (ja) 2013-01-16 2014-07-28 Sony Corp 画像取得装置、画像取得方法、および情報処理プログラム
CN105975980B (zh) 2016-04-27 2019-04-05 百度在线网络技术(北京)有限公司 监控图像标注质量的方法和装置
CN106845090A (zh) 2016-12-30 2017-06-13 海纳医信(北京)软件科技有限责任公司 医学影像标记方法和装置
CN107516005A (zh) 2017-07-14 2017-12-26 上海交通大学 一种数字病理图像标注的方法和系统
US11393587B2 (en) * 2017-12-04 2022-07-19 International Business Machines Corporation Systems and user interfaces for enhancement of data utilized in machine-learning based medical image review
CN108510495A (zh) 2018-04-09 2018-09-07 沈阳东软医疗系统有限公司 一种基于人工智能的肺部影像数据处理方法、装置及系统

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101727674A (zh) * 2008-10-30 2010-06-09 北大方正集团有限公司 一种文档中图片标注及再现标注的方法
US9165360B1 (en) * 2012-09-27 2015-10-20 Zepmed, Llc Methods, systems, and devices for automated analysis of medical scans
CN104462738A (zh) * 2013-09-24 2015-03-25 西门子公司 一种标注医学图像的方法、装置和系统
CN108463814A (zh) * 2015-12-17 2018-08-28 北京安宁福祉科技有限公司 一种医学图像标注方法及系统
CN109446370A (zh) * 2018-10-26 2019-03-08 广州金域医学检验中心有限公司 医学图像的病理标注方法及装置、计算机可读存储介质
CN109461147A (zh) * 2018-10-26 2019-03-12 广州金域医学检验中心有限公司 应用于移动终端的fov图片的病理标注方法及装置
CN109473160A (zh) * 2018-10-26 2019-03-15 广州金域医学检验中心有限公司 医学图像的标注资格确定方法及装置
CN109509541A (zh) * 2018-10-26 2019-03-22 广州金域医学检验中心有限公司 基于医学图像的报告签发方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111784284A (zh) * 2020-06-15 2020-10-16 杭州思柏信息技术有限公司 一种宫颈图像多人协同标记云服务系统及云服务方法
CN111784284B (zh) * 2020-06-15 2023-09-22 杭州思柏信息技术有限公司 一种宫颈图像多人协同标记云服务系统及云服务方法

Also Published As

Publication number Publication date
US11094411B2 (en) 2021-08-17
US20210035677A1 (en) 2021-02-04

Similar Documents

Publication Publication Date Title
US20200167881A1 (en) Automated clinical indicator recognition with natural language processing
US20200126667A1 (en) Automated clinical indicator recognition with natural language processing
US11942218B2 (en) Health testing and diagnostics platform
US9330454B2 (en) Method and apparatus for image-centric standardized tool for quality assurance analysis in medical imaging
CN116344071A (zh) 用于整合临床护理的信息学平台
CN109446370A (zh) 医学图像的病理标注方法及装置、计算机可读存储介质
CN109545394A (zh) 一种问诊方法及装置
WO2020082638A1 (zh) 医学图像的病理标注方法及装置、基于医学图像的报告签发方法及装置、计算机可读存储介质
CN108763080A (zh) 测试数据上传方法、装置、计算机设备及存储介质
CN109509541B (zh) 基于医学图像的报告签发方法及装置
CN109461147B (zh) 应用于移动终端的fov图片的病理标注方法及装置
CN107563113A (zh) 临床用血申请闭环管理系统及方法
CN113160914A (zh) 在线问诊方法、装置、电子设备及存储介质
CN113409280B (zh) 医学影像的处理方法、标注方法和电子设备
US20040199407A1 (en) System for processing data related to a partial reimbursement claim
US20140195266A1 (en) Techniques to improve accuracy of a medical report relating to medical imaging study
US20140195267A1 (en) Techniques to improve accuracy of a medical report relating to a medical imaging study
Chipambwa et al. An Assessment of ereadiness for IoT adoption in Zimbabwean Hospitals: A Case of Chitungwiza Hospital
CN115547479B (zh) 一种基于5g消息的智慧医疗控制方法和系统
Dacheva et al. Implementation of the value-based Healthcare (VBHC) concept with a focus on outcome measurement
Tilahun et al. Intention to Use Mobile-Based Partograph and Its Predictors Among Obstetric Health Care Providers Working at Public Referral Hospitals in the Oromia Region of Ethiopia in 2022: Cross-Sectional Questionnaire Study
CN116825268A (zh) 医学检查报告处理方法、装置、介质及设备
WO2016130879A1 (en) System and method to objectively measure quality assurance in anatomic pathology
Roshan et al. Eye tracking validation: Improving radiologist reporting and interpretation
CN115700683A (zh) 医疗合理性确定方法、设备及程序产品

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: 19876009

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19876009

Country of ref document: EP

Kind code of ref document: A1