CN111028924A - Method and system for labeling medical image data in various forms - Google Patents
Method and system for labeling medical image data in various forms Download PDFInfo
- Publication number
- CN111028924A CN111028924A CN201911001301.5A CN201911001301A CN111028924A CN 111028924 A CN111028924 A CN 111028924A CN 201911001301 A CN201911001301 A CN 201911001301A CN 111028924 A CN111028924 A CN 111028924A
- Authority
- CN
- China
- Prior art keywords
- picture
- labeling
- information
- user
- marking
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000002372 labelling Methods 0.000 title claims abstract description 138
- 238000000034 method Methods 0.000 title claims abstract description 53
- 201000010099 disease Diseases 0.000 claims abstract description 75
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 75
- 238000010586 diagram Methods 0.000 claims description 45
- 238000013500 data storage Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 9
- 239000000284 extract Substances 0.000 description 11
- 208000002177 Cataract Diseases 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 8
- 206010012689 Diabetic retinopathy Diseases 0.000 description 6
- 208000006550 Mydriasis Diseases 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 6
- 230000002526 effect on cardiovascular system Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 206010039722 scoliosis Diseases 0.000 description 5
- 208000024172 Cardiovascular disease Diseases 0.000 description 4
- 208000010412 Glaucoma Diseases 0.000 description 4
- 208000026106 cerebrovascular disease Diseases 0.000 description 4
- 210000000695 crystalline len Anatomy 0.000 description 4
- 208000024519 eye neoplasm Diseases 0.000 description 4
- 230000001771 impaired effect Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 210000001747 pupil Anatomy 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000004438 eyesight Effects 0.000 description 3
- 206010023644 Lacrimation increased Diseases 0.000 description 2
- 206010027646 Miosis Diseases 0.000 description 2
- 206010028813 Nausea Diseases 0.000 description 2
- 206010034960 Photophobia Diseases 0.000 description 2
- 208000005392 Spasm Diseases 0.000 description 2
- 206010047700 Vomiting Diseases 0.000 description 2
- 230000002457 bidirectional effect Effects 0.000 description 2
- 230000036772 blood pressure Effects 0.000 description 2
- 210000005252 bulbus oculi Anatomy 0.000 description 2
- 230000003500 cycloplegic effect Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 210000000744 eyelid Anatomy 0.000 description 2
- 230000004410 intraocular pressure Effects 0.000 description 2
- 230000004317 lacrimation Effects 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 230000008693 nausea Effects 0.000 description 2
- 210000001328 optic nerve Anatomy 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013079 data visualisation Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 208000030533 eye disease Diseases 0.000 description 1
- 230000012447 hatching Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000004304 visual acuity Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Engineering & Computer Science (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention relates to a method and a system for labeling medical image data in various forms, which comprise the following steps: receiving a picture uploaded by a user, and checking whether the format is correct according to a preset picture format; if the picture is correct, uploading the picture, and if the picture is incorrect, reminding the user to upload again; extracting picture information to match with an information base, and judging whether the picture has label or attribute information; setting the authority of marking the picture; providing a labeling tool box for a user according to the disease types; receiving the marking information of the user on the picture, and uploading and storing the marking information; and extracting the labeling information of the users at the same level on the pictures, comparing the labeling information, judging whether the labeling information of the users at the same level is different, if not, generating a final labeling result, and if so, submitting the final labeling result to the next user for further labeling. The invention can label the image by using different labeling tools, meets the labeling requirements of medical images in various forms, and improves the accuracy of labeling the image by adopting a hierarchical labeling process.
Description
Technical Field
The invention relates to the field of medical artificial intelligence, in particular to a method and a system for labeling medical image data in various forms.
Background
With the maturity of artificial intelligence technology, the application scope is wider, and in the medical AI field, the rapid development is slower compared with other fields. The main reasons include small data size, poor labeling quality, and the need of labeling professional people across fields and disciplines. It is the basis of artificial intelligence algorithm to collect, sort, segment and label high quality medical image data and convert it into structured data for computer to recognize and further process. However, in the current stage, the medical image data annotation platform cannot meet the annotation requirements of medical images in various forms, on one hand, different types of diseases need different medical images for auxiliary diagnosis, the focused points of the different types of medical images are different, on the other hand, different medical images need different annotation tools, and the existing annotation platform is difficult to meet simultaneously, so that the annotation efficiency of doctors is greatly reduced, and the accuracy of image annotation is influenced.
At the present stage, the medical image labeling platform has the following problems:
(1) at present, image annotation depends on a large number of professional doctors, and the doctors need to expend energy for annotation after working.
(2) The marking platform is only used as a tool at the present stage, and a complete marking and inspection process is not simulated, a huge specialized team is required for marking and structuring the medical image, and the marking tool at the present stage does not play a role in team cooperation.
(3) The marking tools in the current stage have poor expansibility and high use cost, and medical images with multiple diseases and multiple problems need a large number of special marking tools, for example, scoliosis needs a customized tool to measure the cobber angle. It is costly to customize a large number of annotation tools using existing annotation platforms.
Disclosure of Invention
The invention aims to overcome at least one defect of the prior art, and provides a method and a system for labeling medical image data in various forms, which can use different labeling tools to label pictures, meet medical labeling requirements in various forms, and improve the labeling accuracy by adopting hierarchical labeling.
The technical scheme adopted by the invention is as follows:
in one aspect, a method for labeling medical image data in multiple forms is provided, including:
s1, receiving a picture uploaded by a user, and checking whether the picture format is correct or not according to a preset picture format;
s2, if the picture is correct, uploading the picture, and if the picture is incorrect, reminding the user to upload again;
s3, extracting picture information to match with an information base, and judging whether the picture has label or attribute information;
s4, setting the authority of marking the picture, including the number of the users and the grades of the users;
s5, providing a marking tool box for a user according to the disease types, wherein the marking tool boxes corresponding to different disease types are different, and the marking tool box is provided with one or more marking tools;
s6, receiving marking information of a user on the picture, and uploading and storing the marking information;
and S7, extracting and comparing the label information of all users at the same level on the picture, judging whether the label information of the users at the same level is different, if not, generating a final label result, and if so, submitting the final label result to the next-level user for further labeling.
Medical images with multiple diseases and multiple problems need a large number of special labeling tools for labeling, and the method provides a large number of labeling tool boxes for users to select, so that the medical images with multiple forms are labeled simultaneously, the labeling efficiency of the users is improved, and the labeling quality of pictures is ensured; the method also adopts a hierarchical labeling process, extracts and compares the labeling information of all users in the same level on the same picture, judges whether the labeling information of the users in the same level is different, generates a final labeling result if the labeling information is not different, delivers the labeling information to the users in the next level if the labeling information is different, delivers the same picture to a plurality of users for labeling, and improves the accuracy of picture labeling.
Further, in step S3, if the picture has the label or the attribute information, inquiring whether the user continues to edit, receiving the reply information of the user, prompting the user to label the picture if the user continues to edit, and uploading and storing the label or the attribute information if the user does not continue to edit; and if the picture is not marked or the attribute information is not provided, prompting the user to mark the picture.
Further, the method may further include displaying the disease category and prompting the user to select the disease category.
Further, the picture format preset in step S1 has different picture formats according to different disease categories. Different picture formats are preset according to different disease types, and pictures uploaded by a user correspond to the disease types to a certain extent.
Further, the step S5 includes prompting the user to select a picture attribute and a block diagram attribute, and each disease category has a corresponding picture attribute database and a corresponding block diagram attribute database. The image attributes required by different disease types are different, and the efficiency and the accuracy of selecting the image attributes by the user can be improved by providing a corresponding image attribute database according to the disease type selected by the user; the method provides a corresponding block diagram attribute database according to the disease types selected by the user, and improves the accuracy of selecting the block diagram attributes by the user.
Further, the method divides the users into six levels, and the labeling process of the same picture is as follows: the same picture is simultaneously delivered to a plurality of users at a first level for marking, the system judges whether the marking results are different, and if the marking results are not different, the marking results are final results; if the labeling results are different, the users in the second level are handed to label, the system judges whether the labeling results are different, if the labeling results are not different, the labeling results are the final results, and if the labeling results are different, the users in the third level are handed to label; and if the labeling results are different all the time, taking the labeling of the image by the user with the highest level as the final labeling result. At the same time, this labeling procedure was simulated using computer technology. The method divides users into six grades according to the professional level of the users, simultaneously delivers a picture to a plurality of low-level users for marking, if the marking results are not different, the picture is a final result, if the marking results are different, the picture is delivered to a doctor at a higher level for marking, the marking result of the doctor at the highest level is the final result, the marking process is used for ensuring the quality and the efficiency of the marking, and the computer technology is used for simulating the marking process to provide auditing and checking functions of data for inspectors.
In another aspect, a system for labeling medical image data with multiple modalities is provided, including:
a receiving module: the system comprises a server and a server, wherein the server is used for receiving pictures uploaded by a user and marking information of the pictures;
a format checking module: the picture processing device is used for checking whether the format of the picture is the same as a preset picture format or not;
an information matching module: the system is used for extracting the picture information, matching the picture information with the information base and judging whether the picture has label or attribute information; the system is also used for extracting and comparing the label information of all users at the same level on the picture, and judging whether the label information of the users at the same level is different;
a tool box module: the system comprises a marking tool box, a display device and a display device, wherein the marking tool box is used for providing a marking tool box for a user according to disease types, the marking tool boxes corresponding to different disease types are different, and the marking tool box is provided with one or more marking tools;
a prompt module: the format checking module is used for prompting the user to upload the picture again when the picture format is checked to be different from the preset picture format by the format checking module;
a data storage module: the method is used for storing the pictures uploaded by the user and the marking information of the user on the pictures.
The method comprises the steps that a user uploads a picture after logging in a system, a receiving module of the system receives the picture uploaded by the user, a format checking module checks whether the picture format is the same as a preset picture format, if the picture format is the same as the preset picture format, the picture is stored in a data storage module, and when the format checking module checks that the picture format is not the same as the preset picture format, a prompting module prompts the user to upload the picture again; the information matching module extracts the picture information, matches the picture information with the information base and judges whether label or attribute information exists, the tool box module provides a label tool box for a user, and after the user selects a label tool to label the picture, the data storage module stores the label information of the picture by the user; after the users at the same level are marked, the information matching module extracts the marking information of all the users at the same level on the picture from the data storage module and compares the marking information to judge whether the marking information of the users at the same level is different. Through the implementation mode, the system provides various marking toolboxes for the user to select, so that the medical images in various forms are marked simultaneously, the marking efficiency of the user is improved, and the marking quality of the pictures is ensured; the system also compares the labeling information of the same picture of the same level user, judges whether the labeling information of the same level user is different, generates a final labeling result if the labeling information of the same level user is not different, delivers the same picture to the next level user for labeling if the labeling information of the same level user is different, and delivers the same picture to a plurality of users for labeling, so that the accuracy of picture labeling is improved.
Further, the prompting module is also used for prompting the user to label the picture.
Further, the system also comprises a disease selection module for entering a corresponding block diagram mode according to the disease type selected by the user.
Further, the system also comprises a picture attribute database and a block diagram attribute database, wherein different disease types comprise different picture attribute databases and block diagram attribute databases, the picture attribute database comprises all picture attributes corresponding to the disease types, and the block diagram attribute database comprises all block diagram attributes corresponding to the disease types.
Further, the prompting module is also used for prompting the user to select the picture attribute and the block diagram attribute.
Furthermore, the system is structured in such a way that a bidirectional data calibration mode is adopted at the front end, nginx is used for load balancing and reverse proxy, and django is adopted at the background. The method for binding the data in the two directions is simpler in data operation, view, data and structure are separated, the running speed is higher, nginx is used for load balancing and reverse proxy is used for processing and sending requests more efficiently, the background adopts django, and the method has the advantages of perfect ORM relation mapping, strong route mapping function, perfect view template realization, a sound background management system and strong cache support.
Compared with the prior art, the invention has the beneficial effects that:
(1) generally, a huge specialized team is needed in the marking and structuring process of medical hatching, medical image marking team personnel are divided into six grades according to the professional level of the personnel from low to high by the system, and the picture marking process is as follows: and simultaneously submitting one picture to a plurality of low-level users for marking, if the marking results are not different, giving a final result, if the marking results are different, submitting the picture to a higher-level user for marking, and if the marking results are different, giving a marking result of a doctor at the highest level as the final result.
(2) The method and the device realize simultaneous labeling of medical images in various forms, improve the labeling efficiency of doctors and ensure the labeling quality of images.
(3) And (3) multi-center cooperative use and resource sharing, namely adopting web development based on a Django framework to deploy unified data processing in cloud service, and serving the marking and learning of medical staff of different levels through visual operation processes. According to the unified medical standard, all levels of medical staff can process data through the system, and the system comprises the functions and the technology that the visual database embedding tool based on the digital image processing technology divides and marks images; storing and managing data related to the medical image based on the structured database; based on the data visualization correlation technology, the visualization display, modification, query, classification and the like of all data are realized.
(4) The marking tool has strong expansibility: the marking tool library based on the 'base stone' can be used for developing personalized customization tools according to different requirements, is quick in development and low in cost, has the characteristics of high expansion and pluggable performance, and is easy to maintain.
Drawings
Fig. 1 is a flowchart of a method for labeling medical image data with multiple modalities according to embodiment 1 of the present invention.
Fig. 2 is a hierarchical labeling flowchart of a labeling team in the labeling method of medical image data with multiple modalities according to embodiment 1 of the present invention.
Fig. 3 is a block diagram of a system for labeling medical image data with various modalities according to embodiment 2 of the present invention.
Fig. 4 is an architecture diagram of a system for labeling medical image data with various modalities according to embodiment 2 of the present invention.
Detailed Description
The drawings are only for purposes of illustration and are not to be construed as limiting the invention. For a better understanding of the following embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product; it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
Example 1
As shown in fig. 1, the present embodiment provides a method for labeling medical image data with multiple modalities, including:
s1, receiving a picture uploaded by a user, and checking whether the picture format is correct or not according to a preset picture format;
s2, if the picture is correct, uploading the picture, and if the picture is incorrect, reminding the user to upload again;
s3, extracting picture information to match with an information base, and judging whether the picture has label or attribute information;
s4, setting the authority of marking the picture, including the number of the users and the grades of the users;
s5, providing a marking tool box for a user according to the disease types, wherein the marking tool boxes corresponding to different disease types are different, and the marking tool box is provided with one or more marking tools;
s6, receiving marking information of a user on the picture, and uploading and storing the marking information;
and S7, extracting and comparing the label information of all users at the same level on the picture, judging whether the label information of the users at the same level is different, if not, generating a final label result, and if so, submitting the final label result to the next-level user for further labeling.
When a user inputs a correct account number and a correct password to log in a system to upload a picture, the system receives the picture uploaded by the user and checks whether the picture format is correct according to a preset picture format, if the picture format is correct, the picture is displayed and uploaded for storage, and if the picture format is incorrect, a display error prompts the user to upload again; extracting the information of the picture with the correct format to match with an information base, and judging whether the picture has label or attribute information; then the system sets the authority for marking the pictures, including the number of the users and the levels of the users; when a user marks the picture, the system provides a marking tool box for the user according to the types of diseases, the marking tool boxes corresponding to different types of diseases are different, and the marking tool box is provided with one or more marking tools; receiving the marking information of the user on the picture after the user marking is finished, and uploading and storing the marking information; when the users in the same level finish the marking of the same picture, the system extracts the marking information of all the users in the same level on the picture and compares the marking information, judges whether the marking information of the users in the same level is different, generates a final marking result if the marking information of the users in the same level is not different, and delivers the marking information to the users in the next level for marking if the marking information of the users in the same level is different. By the implementation mode, a user can select a plurality of different marking tools to mark the picture after logging in the system, so that medical images in various forms can be marked at the same time, the marking efficiency of the user is improved, and the marking quality of the picture is ensured; the method also adopts a hierarchical labeling process, extracts and compares the labeling information of all users in the same level on the same picture, judges whether the labeling information of the users in the same level is different, generates a final labeling result if the labeling information is not different, delivers the labeling information to the users in the next level if the labeling information is different, delivers the same picture to a plurality of users for labeling, and improves the accuracy of picture labeling.
In this embodiment, in the step S3, if the picture has the label or the attribute information, the user is asked whether to continue editing, the reply information of the user is received, the user is prompted to label the picture if the editing is continued, and the label or the attribute information is uploaded and stored if the editing is not continued; and if the picture is not marked or the attribute information is not provided, prompting the user to mark the picture.
In this embodiment, the method further comprises displaying the disease category and prompting the user to select the disease category.
Specifically, when the method of the invention is applied to medical image data annotation of eye diseases, the types of the diseases can be cataract, scoliosis, diabetic retinopathy, cardiovascular and cerebrovascular diseases, eye tumors and glaucoma; the image attributes of the cataract include a photographing mode and a photographing mode.
In this embodiment, the picture format preset in step S1 has different picture formats according to different disease types. Different picture formats are preset according to different disease types, and pictures uploaded by a user correspond to the disease types to a certain extent.
In this embodiment, the step S5 is provided with a labeling tool box, one or more labeling tools are configured corresponding to different disease types, and the corresponding labeling tool is called from the labeling tool box according to the disease type corresponding to the picture for the user to use; in a specific implementation, S5 further includes prompting the user to select a picture attribute and a block diagram attribute, where each disease category has a corresponding picture attribute database and a corresponding block diagram attribute database. The image attributes required by different disease types are different, and a corresponding image attribute database is provided according to the disease type selected by the user so as to improve the efficiency and accuracy of selecting the image attributes by the user; the method provides a corresponding block diagram attribute database according to the disease types selected by the user, and improves the accuracy of selecting the block diagram attributes by the user.
Specifically, for example, the image attribute library of scoliosis is divided into: the image attribute library of the diabetic retinopathy is below 10 degrees, 11-20 degrees, 21-45 degrees and more than 45 degrees, and comprises the following steps: the image attribute library of the cardiovascular and cerebrovascular is as follows: blood pressure, whether suffering from: cardiovascular and cerebrovascular diseases and disease types, and the image attribute library of eye tumor is: the library of image attributes for vision, glaucoma is: photophobia, lacrimation, eyelid spasm, cupping, closed angle, vision, intraocular pressure, emesis, nausea, impaired visual field, and impaired optic nerve.
In a specific implementation process of this embodiment, the users are divided into six levels, specifically, as shown in fig. 2, the hierarchical flow chart of the labeling team is shown, the users are medical interns, hospitalized physicians, attending physicians, assistant chief physicians, and experts, and are divided into six levels according to their professional levels, the medical interns are in a first level, and the experts are in a sixth level; the labeling process of the picture is as follows: the same image is simultaneously sent to six medical trainees for marking, the system judges whether the marking results are different, and if the marking results are not different, the marking results are the final results; if the labeling results are different, the marking results are handed to five inpatients for labeling, and then the system judges whether the labeling results are different, if the labeling results are not different, the labeling results are the final results, and if the labeling results are different, the marking results are handed to four main physicians for labeling; if the labeling results are different all the time, the labeling carried out on the picture by the expert is used as the final labeling result, and meanwhile, the labeling process is simulated by using a computer technology. The method divides users into six grades according to the professional level of the users, simultaneously delivers a picture to a plurality of low-level users for marking, if the marking results are not different, the picture is a final result, if the marking results are different, the picture is delivered to a doctor at a higher level for marking, the marking result of the doctor at the highest level is the final result, the marking process is used for ensuring the quality and the efficiency of the marking, and the computer technology is used for simulating the marking process to provide auditing and checking functions of data for inspectors. In the specific implementation process, the difference value can be preset according to different labeling information corresponding to different diseases, so that the judgment can be carried out according to the set difference value, and in addition, the difference can also be judged by utilizing a mathematical model.
In the specific implementation process of the embodiment, the system transmits the same picture to a plurality of medical interns, when the medical interns input correct account numbers and passwords to log in the system, the system enters a main page, the main page can display the types of diseases for the medical interns to select, after the medical interns select the types of the diseases, the medical interns select to upload local pictures or load original pictures, the system displays a corresponding picture attribute database and prompts the medical interns to set the picture attributes of the images, the medical interns select appropriate labeling tools from labeling toolboxes provided by different slave systems according to the picture attributes to label the pictures, generate labeling results, select corresponding block diagram attributes of the added diseases from the block diagram attribute database according to the labeling, and upload and store the labeling results and the block diagram attributes; after the marking of the medical interns is finished, the system extracts the marked information to judge whether the marking results of the multiple medical interns on the same picture are different, if not, the marking result is the final marking result, and if so, the marking results are handed to multiple inpatients for marking. For example, the disease type selected by the medical intern is cataract, the picture is uploaded to the server after the corresponding picture is selected from the local picture, the picture is stored in the server after the system checks that the picture format is correct, and then the medical intern sets the picture attribute to be a photographing mode and a photographing mode, wherein the photographing mode is various: the mydriasis processing can be divided into mydriasis photographing and small pupil photographing according to the existence of the patient before photographing; the slit-light photographing and the diffuse light photographing can be divided according to the width of the slit lamp. The diameter of the pupil is 2-3mm in the natural state, and mydriasis treatment is to use cycloplegic to make the diameter of the pupil more than 5mm, which is beneficial to the overall observation of the periphery of the crystalline lens. The slit light photographing section is convenient for observing the structures of crystalline lens and eyeball and the depth of lesion, and the diffuse light photographing can comprehensively observe the structure of the ocular surface. And then selecting a corresponding labeling tool from a labeling tool box provided by the system to label the picture, wherein the system can display a corresponding block diagram attribute database of the cataract for the addition of a medical intern, the corresponding block diagram attributes comprise a patient number, a block diagram name, a block diagram type, a cataract type, a nuclear hardness, whether the visual axis region is turbid or not and whether the subcapsular is turbid or not, and the corresponding block diagram attributes need to be supplemented every time one label is added. The system generates an annotation result according to the annotation of the medical interne to the image, and stores the added block diagram attribute as additional information into the server. And finally, judging whether the images are different according to the labeling results of the plurality of trainees in the hospital by the system, if not, taking the labeling result as a final result, and if so, transferring the images to a plurality of inpatients for labeling.
Example 2
As shown in fig. 3, the present embodiment provides a system for labeling medical image data, including:
a receiving module: the system comprises a server and a server, wherein the server is used for receiving pictures uploaded by a user and marking information of the pictures;
a format checking module: the picture processing device is used for checking whether the format of the picture is the same as a preset picture format or not;
an information matching module: the system is used for extracting the picture information, matching the picture information with the information base and judging whether the picture has label or attribute information; the system is also used for extracting and comparing the label information of all users at the same level on the picture, and judging whether the label information of the users at the same level is different;
a tool box module: the system comprises a marking tool box, a display device and a display device, wherein the marking tool box is used for providing a marking tool box for a user according to disease types, the marking tool boxes corresponding to different disease types are different, and the marking tool box is provided with one or more marking tools;
a prompt module: the format checking module is used for prompting the user to upload the picture again when the picture format is checked to be different from the preset picture format by the format checking module;
a data storage module: the method is used for storing the pictures uploaded by the user and the marking information of the user on the pictures.
Specifically, a user uploads a picture after logging in a system, a receiving module of the system receives the picture uploaded by the user, a format checking module checks whether the picture format is the same as a preset picture format, if so, the picture is stored in a data storage module, and when the format checking module checks that the picture format is not the same as the preset picture format, a prompting module prompts the user to upload the picture again; the information matching module extracts the picture information, matches the picture information with the information base and judges whether label or attribute information exists, the tool box module provides a label tool box for a user, and after the user selects a label tool to label the picture, the data storage module stores the label information of the picture by the user; after the users at the same level are marked, the information matching module extracts the marking information of all the users at the same level on the picture from the data storage module and compares the marking information to judge whether the marking information of the users at the same level is different. Through the implementation mode, the system provides various marking toolboxes for the user to select, so that the medical images in various forms are marked simultaneously, the marking efficiency of the user is improved, and the marking quality of the pictures is ensured; the system also compares the labeling information of the same picture of the same level user, judges whether the labeling information of the same level user is different, generates a final labeling result if the labeling information of the same level user is not different, delivers the same picture to the next level user for labeling if the labeling information of the same level user is different, and delivers the same picture to a plurality of users for labeling, so that the accuracy of picture labeling is improved. In the specific implementation process, the difference value can be preset according to different labeling information corresponding to different diseases, so that the judgment can be carried out according to the set difference value, and in addition, the difference can also be judged by utilizing a mathematical model.
In this embodiment, the prompting module is further configured to prompt a user to label the picture.
In this embodiment, the system further includes a disease selection module, configured to enter a corresponding block diagram mode according to a disease category selected by a user. Specifically, when the system of the invention is applied to medical image data annotation of ophthalmic diseases, spine diseases and the like, the types of the diseases which can be selected in the disease selection module include cataract, scoliosis, diabetic retinopathy, cardiovascular and cerebrovascular diseases, eye tumors and glaucoma; the image attributes of the cataract include a photographing mode and a photographing mode.
In this embodiment, the system further includes a picture attribute database and a block diagram attribute database, where different disease types include different picture attribute databases and block diagram attribute databases, the picture attribute database includes all picture attributes corresponding to the disease types, and the block diagram attribute database stores all block diagram attributes corresponding to the disease types.
Specifically, for example, the image attribute database of scoliosis is divided into: the image attribute database of diabetic retinopathy is below 10 degrees, between 11 and 20 degrees, between 21 and 45 degrees and above 45 degrees: the domestic diabetic retinopathy stage (optional stage 1-5) and the international diabetic retinopathy stage (optional stage 1-6) are as follows: blood pressure, whether suffering from: cardiovascular and cerebrovascular diseases and disease types, and the image attribute database of the eye tumor is as follows: the visual acuity, glaucoma image attribute database is: photophobia, lacrimation, eyelid spasm, cupping, closed angle, vision, intraocular pressure, emesis, nausea, impaired visual field, and impaired optic nerve.
In this embodiment, the prompting module is further configured to prompt the user to select the picture attribute and the block diagram attribute.
In this embodiment, as shown in fig. 4, the architecture of the system is that the front end adopts a bidirectional data calibration mode, nginx is used for load balancing and reverse proxy, and the background adopts django. The method for binding the data in the two directions is simpler in data operation, view, data and structure are separated, the running speed is higher, nginx is used for load balancing and reverse proxy is used for processing and sending requests more efficiently, the background adopts django, and the method has the advantages of perfect ORM relation mapping, strong route mapping function, perfect view template realization, a sound background management system and strong cache support.
In this embodiment, the users are medical interns, hospitalized physicians, attending physicians, assistant chief physicians, and experts, and are classified into six levels according to their professional levels, the medical interns are in the first level, and the experts are in the sixth level. Specifically, the system transmits the same picture to a plurality of medical interns, when the medical interns input correct account numbers and passwords, the medical interns enter a main page of the system, a disease selection module enters a corresponding block diagram mode according to the types of diseases selected by the medical interns, a receiving module receives the picture after the medical interns upload the picture, a format checking module checks whether the format of the picture is the same as a preset picture format, and if the format of the picture is not the same as the preset picture format, a prompting module prompts the medical interns to upload the picture again; if the pictures are stored in the data storage module, the information matching module extracts the picture information, matches the picture information with the information base, judges whether the pictures have marks or attribute information, and prompts a user to mark the pictures by the prompting module. The toolbox module provides a labeling toolbox for the medical interne according to the disease type, the system provides a corresponding picture attribute database and a corresponding block diagram attribute database according to the disease type, and the prompt module prompts the medical interne to select the picture attribute and the block diagram attribute. After the medical intern finishes labeling, the data storage module stores the labeling information of the medical intern on the picture, and the information matching module extracts the labeling information of all users at the same level on the picture from the database module and compares the labeling information to judge whether the labeling information of the users at the same level is different. For example, the disease type selected by the medical intern is cataract, the picture is uploaded after the corresponding picture is selected from the local picture, the picture is stored in the data storage module after the system format check module checks that the picture format is correct, and then the medical intern sets the picture attribute to be a photographing mode and a photographing mode according to the prompt of the prompt module, wherein the photographing mode is various: the mydriasis processing can be divided into mydriasis photographing and small pupil photographing according to the existence of the patient before photographing; the slit-light photographing and the diffuse light photographing can be divided according to the width of the slit lamp. The diameter of the pupil is 2-3mm in the natural state, and mydriasis treatment is to use cycloplegic to make the diameter of the pupil more than 5mm, which is beneficial to the overall observation of the periphery of the crystalline lens. The slit light photographing section is convenient for observing the structures of crystalline lens and eyeball and the depth of lesion, and the diffuse light photographing can comprehensively observe the structure of the ocular surface. And then selecting a corresponding labeling tool from the tool box module to label the picture, wherein the system can display a corresponding block diagram attribute database of the cataract for the addition of a medical intern, the corresponding block diagram attribute database comprises a patient number, a block diagram name, a block diagram type, a cataract type, a nuclear hardness, whether the visual axis region is turbid or not and whether the subcapsular region is turbid or not, and the corresponding block diagram attribute needs to be supplemented every time one label is added. And the data storage module generates a labeling result according to the labeling of the medical interne on the image, and stores the added block diagram attribute as additional information into the server. The information matching module judges whether the difference exists according to the labeling results of multiple hospital trainees, if the difference does not exist, the labeling result is the final result, and if the difference exists, the task sending module sends the image to multiple inpatients for labeling. And finally, the system extracts the picture, the label, the block diagram attribute and the final result from a data storage module to generate a data file, wherein the data file comprises the coordinate of the picture label, the attribute corresponding to the coordinate and a diagnosis result, and the data file is used as a training sample to be transmitted to a neural network.
Claims (10)
1. A multi-form medical image data labeling method is characterized by comprising the following steps:
s1, receiving a picture uploaded by a user, checking whether the picture format is correct according to a preset picture format,
s2, if the picture is correct, uploading the picture, and if the picture is incorrect, reminding the user to upload again;
s3, extracting picture information to match with an information base, and judging whether the picture has label or attribute information;
s4, setting the authority of marking the picture, including the number of the users and the grades of the users;
s5, providing a marking tool box for a user according to the disease types, wherein the marking tool boxes corresponding to different disease types are different, and the marking tool box is provided with one or more marking tools;
s6, receiving marking information of a user on the picture, and uploading and storing the marking information;
and S7, extracting and comparing the label information of all users at the same level on the picture, judging whether the label information of the users at the same level is different, if not, generating a final label result, and if so, submitting the final label result to the next-level user for further labeling.
2. The method for labeling medical image data with multiple modalities according to claim 1, wherein in step S3, if the picture has the label or attribute information, the user is asked whether to continue editing, the reply information of the user is received, the user is prompted to label the picture if the editing continues, and the label or attribute information is uploaded and stored if the editing does not continue; and if the picture is not marked or the attribute information is not provided, prompting the user to mark the picture.
3. The method of claim 1, further comprising displaying a disease category and prompting a user to select a disease category.
4. The method for labeling medical images with multiple modalities according to claim 3, wherein the preset picture format in step S1 has different picture formats according to different disease types.
5. The method for labeling multi-modality medical influence data according to claim 1, wherein the step S5 further comprises prompting the user to select a picture attribute and a frame attribute, and each disease category has a corresponding picture attribute database and a corresponding frame attribute database.
6. A multi-form medical image data labeling system is characterized by comprising:
a receiving module: the system comprises a server and a server, wherein the server is used for receiving pictures uploaded by a user and marking information of the pictures;
a format checking module: the picture processing device is used for checking whether the format of the picture is the same as a preset picture format or not;
an information matching module: the system is used for extracting the picture information, matching the picture information with the information base and judging whether the picture has label or attribute information; the system is also used for extracting and comparing the label information of all users at the same level on the picture, and judging whether the label information of the users at the same level is different;
a tool box module: the system comprises a marking tool box, a display device and a display device, wherein the marking tool box is used for providing a marking tool box for a user according to disease types, the marking tool boxes corresponding to different disease types are different, and the marking tool box is provided with one or more marking tools;
a prompt module: the format checking module is used for prompting the user to upload the picture again when the picture format is checked to be different from the preset picture format by the format checking module;
a data storage module: the method is used for storing the pictures uploaded by the user and the marking information of the user on the pictures.
7. The system for labeling medical image data with multiple modalities of claim 6, wherein the prompting module is further configured to prompt a user to label the picture.
8. The system for labeling medical image data as claimed in claim 6, further comprising a disease selection module for entering a corresponding frame mode according to a disease category selected by a user.
9. The system for labeling medical image data with multiple modalities according to claim 6, further comprising a picture attribute database and a frame diagram attribute database, wherein different disease types include different picture attribute databases and frame diagram attribute databases, the picture attribute database contains all picture attributes corresponding to the disease types, and the frame diagram attribute database stores all frame diagram attributes corresponding to the disease types.
10. The system for labeling medical image data according to claim 9, wherein the prompting module is further configured to prompt a user to select the picture attribute and the block attribute.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911001301.5A CN111028924A (en) | 2019-10-21 | 2019-10-21 | Method and system for labeling medical image data in various forms |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911001301.5A CN111028924A (en) | 2019-10-21 | 2019-10-21 | Method and system for labeling medical image data in various forms |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111028924A true CN111028924A (en) | 2020-04-17 |
Family
ID=70201272
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911001301.5A Pending CN111028924A (en) | 2019-10-21 | 2019-10-21 | Method and system for labeling medical image data in various forms |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111028924A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111830039A (en) * | 2020-07-22 | 2020-10-27 | 南京认知物联网研究院有限公司 | Intelligent product quality detection method and device |
CN114171166A (en) * | 2021-01-20 | 2022-03-11 | 赛维森(广州)医疗科技服务有限公司 | Management system of model of visual digital pathological artificial intelligence |
WO2022068228A1 (en) * | 2020-09-29 | 2022-04-07 | 平安国际智慧城市科技股份有限公司 | Lesion mark verification method and apparatus, and computer device and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104462738A (en) * | 2013-09-24 | 2015-03-25 | 西门子公司 | Method, device and system for labeling medical images |
CN108682453A (en) * | 2018-05-16 | 2018-10-19 | 四川大学 | A kind of Lung neoplasm labeling system |
CN109446370A (en) * | 2018-10-26 | 2019-03-08 | 广州金域医学检验中心有限公司 | Pathology mask method and device, the computer readable storage medium of medical image |
CN109686423A (en) * | 2018-11-06 | 2019-04-26 | 众安信息技术服务有限公司 | A kind of medical imaging mask method and system |
-
2019
- 2019-10-21 CN CN201911001301.5A patent/CN111028924A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104462738A (en) * | 2013-09-24 | 2015-03-25 | 西门子公司 | Method, device and system for labeling medical images |
CN108682453A (en) * | 2018-05-16 | 2018-10-19 | 四川大学 | A kind of Lung neoplasm labeling system |
CN109446370A (en) * | 2018-10-26 | 2019-03-08 | 广州金域医学检验中心有限公司 | Pathology mask method and device, the computer readable storage medium of medical image |
CN109686423A (en) * | 2018-11-06 | 2019-04-26 | 众安信息技术服务有限公司 | A kind of medical imaging mask method and system |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111830039A (en) * | 2020-07-22 | 2020-10-27 | 南京认知物联网研究院有限公司 | Intelligent product quality detection method and device |
WO2022068228A1 (en) * | 2020-09-29 | 2022-04-07 | 平安国际智慧城市科技股份有限公司 | Lesion mark verification method and apparatus, and computer device and storage medium |
CN114171166A (en) * | 2021-01-20 | 2022-03-11 | 赛维森(广州)医疗科技服务有限公司 | Management system of model of visual digital pathological artificial intelligence |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111028924A (en) | Method and system for labeling medical image data in various forms | |
EP3513761B1 (en) | 3d platform for aesthetic simulation | |
US8311848B2 (en) | Electronic medical record creation and retrieval system | |
WO2022166399A1 (en) | Fundus oculi disease auxiliary diagnosis method and apparatus based on bimodal deep learning | |
JP2009530711A (en) | System and method for psychological assessment | |
WO2022247007A1 (en) | Medical image grading method and apparatus, electronic device, and readable storage medium | |
CN116072255B (en) | Artificial intelligence-based cervical vertebra degenerative disease CT (computed tomography) examination structured report system | |
JPWO2019220833A1 (en) | Diagnostic support system, diagnostic support device and diagnostic support method | |
CN116259407A (en) | Disease diagnosis method, device, equipment and medium based on multi-mode data | |
JP2024515448A (en) | Systems and methods for dynamic identification of surgical trays and items contained thereon - Patents.com | |
KR102224596B1 (en) | A system and method for automatically generating facial correction designs and application protocols for handling identifiable facial deviations | |
CN117442154A (en) | Visual detection system based on children's attention | |
CN113918609A (en) | Test paper creating method and device, computer equipment and storage medium | |
Cohn et al. | Retinal topography maps in R: new tools for the analysis and visualization of spatial retinal data | |
KR20220000851A (en) | Dermatologic treatment recommendation system using deep learning model and method thereof | |
WO2024023584A2 (en) | Systems and methods using multidimensional language and vision models and maps to categorize, describe, coordinate, and track anatomy and health data | |
CN107341189A (en) | A kind of indirect labor carries out the method and system of examination, classification and storage to image | |
CN116824209A (en) | Bone window prediction method and system | |
CN110059186A (en) | A kind of method of medical terminology check and correction task distribution | |
CN114613477A (en) | Psychological monitoring method, device, equipment and storage medium based on deep learning | |
WO2022055394A1 (en) | Digital computer-implemented platform for creating medical applications using artificial intelligence and operating method thereof | |
CN114693587A (en) | Quality control method and quality control system for data annotation of fundus images | |
JP2018142249A (en) | Answer sheet grading method | |
CN112509688B (en) | Automatic analysis system, method, equipment and medium for pressure sore picture | |
CN118364340B (en) | Student course short board positioning method, system and storage medium based on deep learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200417 |