CN110853737B - Updating method and system of medical image recognition model - Google Patents
Updating method and system of medical image recognition model Download PDFInfo
- Publication number
- CN110853737B CN110853737B CN201910976085.XA CN201910976085A CN110853737B CN 110853737 B CN110853737 B CN 110853737B CN 201910976085 A CN201910976085 A CN 201910976085A CN 110853737 B CN110853737 B CN 110853737B
- Authority
- CN
- China
- Prior art keywords
- user
- medical image
- analysis result
- image analysis
- server
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000010191 image analysis Methods 0.000 claims abstract description 131
- 238000012986 modification Methods 0.000 claims abstract description 99
- 230000004048 modification Effects 0.000 claims abstract description 99
- 238000012549 training Methods 0.000 claims description 24
- 238000012790 confirmation Methods 0.000 claims description 20
- 238000002059 diagnostic imaging Methods 0.000 claims description 18
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000010801 machine learning Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 3
- 239000000758 substrate Substances 0.000 claims 3
- 238000004458 analytical method Methods 0.000 abstract 1
- 238000002591 computed tomography Methods 0.000 description 6
- 238000003384 imaging method Methods 0.000 description 4
- 238000005481 NMR spectroscopy Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction 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
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)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The embodiment of the invention discloses a method and a device for updating a medical image identification model, wherein the method comprises the following steps: the film reading device uses a first medical image identification model to identify the target medical image to obtain a first image analysis result, receives an instruction, and modifies the first image analysis result according to a modification instruction if the received instruction is a modification instruction to generate a second image analysis result; the film reading device sends a second influence analysis result to a server, the server updates the image analysis result modification confidence of the first medical image recognition model, judges whether the image analysis result modification confidence reaches a first preset value, and if so, obtains a second medical image recognition model based on data stored in a second database and sends the second medical image recognition model to all film reading devices. By the method, the problems caused by frequent updating of the model are avoided.
Description
Technical Field
The embodiment of the invention relates to the field of artificial intelligence, in particular to a method and a system for updating a medical image recognition model.
Background
With the rapid development of computer technology, artificial intelligence starts to fall to the ground deeply in the medical field, and under the support of big data and model algorithms, a machine film reading technology becomes a research hot spot in the intelligent medical field in recent years. The machine film reading technology is that a film reading robot is used for automatically identifying and processing medical images such as MRI images, CT images, ultrasonic images and the like to form an image report so as to assist doctors in diagnosing diseases. Compared with manual film reading, the film reading efficiency is greatly improved by adopting a machine to read the film.
In the prior art, when the recognition result in the image report given by the film reading robot is different from the recognition result finally confirmed by the doctor, the recognition model of the film reading robot is updated, but frequent updating of the recognition model can cause instability of the model, and the recognition result given by the doctor has a certain error rate, so that the error updating of the recognition model of the film reading robot can be caused.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a method and a system for updating a medical image recognition model.
The embodiment of the invention provides a method for updating a medical image recognition model, which comprises the following steps:
step 101, at least one film reading device acquires a target medical image from at least one medical imaging device;
102, the film reading device uses a first medical image recognition model to recognize the target medical image, and a first image analysis result corresponding to the target medical image is obtained;
step 103, the film reading device receives a user instruction and judges whether the received instruction is a confirmation instruction or a modification instruction; if the instruction is a confirmation instruction, executing step 104; if it is a modification instruction, steps 105-107 are performed;
104, the film reading device sends the target medical image and the first image analysis result to a server, and the server correspondingly stores the target medical image and the first image analysis result in a first database as one data record;
step 105, the film reading device modifies the first image analysis result according to the modification instruction to generate a second image analysis result; the film reading device sends the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to a server, the server correspondingly stores the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction in a second database as one data record, and the server updates the image analysis result modification confidence of the first medical image recognition model, specifically: the server determines the user confidence coefficient according to the user identification, updates the image analysis result modification confidence coefficient based on the user confidence coefficient, and the updated image analysis result modification confidence coefficient is the image analysis result modification confidence coefficient before updating plus the user confidence coefficient corresponding to the user giving the modification instruction; the value of the user confidence is between 0 and 1;
step 106, the server judges whether the image analysis result modification confidence of the first medical image recognition model reaches a first preset value, if so, the first medical image recognition model is updated based on the data stored in a second database, and a second medical image recognition model is obtained; wherein the first preset value is a positive integer greater than 1.
Step 107, the server sends the second medical image recognition model to all the film reading devices, the film reading devices replace the local first medical image recognition model by the second medical image recognition model, and after the replacement is successful, an update success message is sent to the server.
The embodiment of the invention provides a system for updating a medical image recognition model, which comprises at least one medical imaging device, at least one film reading device and at least one server;
the medical imaging device is used for generating a target medical image;
the film reading device is used for acquiring a target medical image from medical imaging equipment; identifying the target medical image by using a first medical image identification model to obtain a first image analysis result corresponding to the target medical image; receiving a user instruction and judging whether the received instruction is a confirmation instruction or a modification instruction; if the target medical image is the confirmation instruction, the target medical image and the first image analysis result are sent to a server; if the target medical image is the modification instruction, modifying the first image analysis result according to the modification instruction, generating a second image analysis result, and sending the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to a server; receiving a second medical image recognition model from a server, replacing a local first medical image recognition model by adopting the second medical image recognition model, and sending an update success message to the server after successful replacement;
the server is used for correspondingly storing the target medical image received from the film reading device and the first image analysis result in a first database as one data record; correspondingly storing the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction as a data record in a second database, and updating the image analysis result modification confidence of the first medical image identification model, wherein the method specifically comprises the following steps: determining user confidence coefficient according to the user identification, updating the image analysis result modification confidence coefficient based on the user confidence coefficient, wherein the updated image analysis result modification confidence coefficient is the image analysis result modification confidence coefficient before updating plus the user confidence coefficient corresponding to the user giving the modification instruction; judging whether the image analysis result modification confidence of the first medical image recognition model reaches a first preset value or not, if so, updating the first medical image recognition model based on data stored in a second database to obtain a second medical image recognition model, and sending the second medical image recognition model to all the film reading devices;
wherein, the value of the user confidence is between 0 and 1; the first preset value is a positive integer greater than 1.
By the method, the model is updated when the first updating parameters of the medical image recognition model are met, and the problem caused by frequent updating of the model is avoided. In addition, before updating the model, the invention also counts the distribution of the data sources such as hospitals or users in the second database, so as to avoid the error update of the system model caused by the deviation or error of individual hospitals or individual users in the film reading.
Drawings
FIG. 1 is a method for updating a medical image recognition model in an embodiment of the present invention.
FIG. 2 is an update system of a medical image recognition model in an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings.
The device involved in the updating method of the medical image identification model comprises at least one medical imaging device, at least one film reading device and at least one server.
The medical imaging device may be a different type of device, such as a computed tomography device, a positron emission computed tomography device, a digital direct imaging device, a nuclear magnetic resonance device, a computer X-ray scanning device, etc., each type of medical imaging device may include one or more medical imaging devices, and the medical imaging devices may be located at different locations, such as different hospitals.
The film reading device can also be of different types, such as a film reading device capable of reading medical images obtained by a computed tomography device, a film reading device capable of reading medical images obtained by a positron emission computed tomography device, a film reading device capable of reading medical images obtained by a digital direct imaging device, and the like, and in another embodiment, the film reading device can read medical images obtained by different medical imaging devices, such as a film reading device capable of reading medical images obtained by a digital direct imaging device and a film reading device capable of reading medical images obtained by a nuclear magnetic resonance device. The at least one film reading device may be located in different locations, such as different hospitals.
The at least one film reading device is connected with the at least one server through a wireless network or a wired network, and data is sent or received through the connection.
The method for updating the medical image recognition model of the present invention comprises the steps of, referring to fig. 1,
step 101, at least one film reading device acquires a target medical image from at least one medical imaging device;
102, the film reading device uses a first medical image recognition model to recognize the target medical image, and a first image analysis result corresponding to the target medical image is obtained;
step 103, the film reading device receives a user instruction and judges whether the received instruction is a confirmation instruction or a modification instruction; if the instruction is a confirmation instruction, executing step 104; if it is a modification instruction, steps 105-107 are performed;
104, the film reading device sends the target medical image and the first image analysis result to a server, and the server correspondingly stores the target medical image and the first image analysis result in a first database as one data record;
step 105, the film reading device modifies the first image analysis result according to the modification instruction to generate a second image analysis result; the film reading device sends the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to a server, the server correspondingly stores the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction in a second database as one data record, and the server updates the image analysis result modification confidence of the first medical image recognition model, specifically: the server determines the user confidence coefficient according to the user identification, updates the image analysis result modification confidence coefficient based on the user confidence coefficient, and the updated image analysis result modification confidence coefficient is the image analysis result modification confidence coefficient before updating plus the user confidence coefficient corresponding to the user giving the modification instruction; the value of the user confidence is between 0 and 1;
step 106, the server judges whether the image analysis result modification confidence of the first medical image recognition model reaches a first preset value, if so, the first medical image recognition model is updated based on the data stored in a second database, and a second medical image recognition model is obtained; wherein the first preset value is a positive integer greater than 1.
Step 107, the server sends the second medical image recognition model to all the film reading devices, the film reading devices replace the local first medical image recognition model by the second medical image recognition model, and after the replacement is successful, an update success message is sent to the server.
Preferably, the target medical image is a medical image to be read, which is a medical image obtained by scanning or photographing a part to be detected of a patient.
Preferably, the first image analysis result includes a state of the part to be detected, a focus type, a focus position, a focus size, and the like.
Preferably, before step 101, the server acquires a plurality of medical images and image analysis results corresponding to the medical images to form a training set, generates a first medical image recognition model through machine learning training or neural network training based on the training set, and sends the generated first medical image recognition model to all the film reading devices connected with the first medical image recognition model.
Preferably, before step 101, the server acquires a plurality of medical images of a certain type and image analysis results corresponding to the medical images to form a training set, generates a first medical image recognition model corresponding to the medical images of the type through machine learning training or neural network training based on the training set, and sends the generated first medical image recognition model to all the film reading devices capable of reading the medical images of the type.
Preferably, after step 102 and before step 103, the film reading device prompts the user that the first image analysis result has been generated, and displays the target medical image and the first image analysis result in response to the operation of the user, while simultaneously displaying a confirmation control and a modification control on an interface displaying the target medical image and the first image analysis result. If the user selects the confirmation control, the film reading device receives the confirmation instruction, and if the user selects the modification control, the film reading device receives the modification instruction.
Preferably, after step 102 and before step 103, the film reading device prompts the user that the first image analysis result has been generated, and displays the target medical image and the first image analysis result in response to a first voice command of the user, after that, the film reading device acquires a second voice command of the user, if the second voice command includes a positive keyword, the film reading device receives a confirmation command, and if the second voice command includes a negative keyword, the film reading device receives a modification command.
Preferably, the user is a doctor or a reader.
Preferably, the user confidence is determined according to the level of the user giving the modification instruction, and the user confidence is different according to the level of the user giving the modification instruction. For example, the expert's user confidence level is 0.99, the primary physician's user confidence level is 0.98, the secondary primary physician's user confidence level is 0.95, the attending physician's user confidence level is 0.92, and the practicing physician's user confidence level is 0.8. The server determines the user confidence according to the user identification, specifically: the corresponding relation between the user identification and the user level is stored in the server, and the server determines the user level according to the corresponding relation and the received user identification, so as to determine the user confidence level.
Preferably, the user confidence is determined based on the medical representation of the user giving the modification instruction. The medical image comprises the following parameters: the hospital level, the user level, the working years, the number of the read sheets, the patient evaluation and the like. The user confidence is a weighted sum of the parameters included in the medical representation. The server determines the user confidence according to the user identification, specifically: the corresponding relation between the user identification and the user medical portrait is stored in the server, and the server determines the user medical portrait according to the corresponding relation and the received user identification, and further determines the user confidence coefficient.
Preferably, when the received user identifier is not found in the corresponding relationship, determining that the user confidence coefficient is the lowest value.
Preferably, when the received user identifier is not found in the correspondence, the target medical image, the first image analysis result and the data record corresponding to the second image analysis result, which are simultaneously transmitted to the server with the user identifier, are deleted from the second database.
Preferably, the server updates the first medical image identification model based on data records stored in the first database and the second database.
Preferably, the server updates the first medical image identification model based on a preset number of data records in the first database and data records stored in the second database.
Preferably, the film reading device sends the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to the server, and simultaneously sends the hospital to which the film reading device belongs to the server, and the server correspondingly stores the data in the second database as a data record.
In step 106, the server determines whether the image analysis result modification confidence coefficient reaches a first preset value, if so, further counts the distribution of the identification of the hospital to which the film reading device belongs and the user giving the modification instruction in each piece of data record stored in the second database, if the counted distribution result meets the first condition and the second condition at the same time, updates the first medical image recognition model based on the data stored in the second database to obtain a second medical image recognition model, and if the image analysis result modification confidence coefficient does not reach the first preset value or the counted distribution result does not meet at least one of the first condition and the second condition, does not update the first medical image recognition model. The first condition is that the statistical duty ratio of a hospital to which any piece reading device belongs in the distribution results obtained through statistics does not exceed a first preset value, and the second condition is that the statistical duty ratio of the identification of any user in the distribution results obtained through statistics does not exceed a second preset value.
Preferably, if the statistical proportion of the hospitals to which at least one film reading device belongs in the distribution results obtained through statistics exceeds a first preset value, that is, the distribution results obtained through statistics do not meet a first condition, deleting the data record corresponding to the hospitals to which the film reading device belongs from the second database, and updating the image analysis result based on the data record in the second database to modify the confidence coefficient.
Preferably, if the statistical proportion of at least one user identifier in the distribution result obtained by statistics exceeds a second preset value, that is, the distribution result obtained by statistics does not meet a second condition, deleting the data record corresponding to the user identifier from the second database, and updating the image analysis result based on the data record in the second database to modify the confidence coefficient.
Preferably, the reader can identify different types of target medical images. The types include: a target medical image generated by a computer tomography device, a target medical image generated by a positron emission type computer tomography device, a target medical image generated by a digital direct imaging device, a target medical image generated by a nuclear magnetic resonance device, a target medical image generated by a computer X-ray scanning device, and the like.
In step 102, the film reader determines the type of the target medical image, and uses a first medical image recognition model corresponding to the type to recognize the target medical image, in step 104, the server stores the target medical image and the first image analysis result as a data record correspondence in a first database corresponding to the type, and in step 105, the server stores the target medical image, the first image analysis result, the second image analysis result, and the identification of the user giving the modification instruction as a data record correspondence in a second database corresponding to the type. In step 107, the server transmits the second medical image recognition model to all of the film reading devices capable of reading the target medical images of the type described above.
Preferably, the method further includes step 108, the server determines whether an update success message is received from all the film reading devices that have sent the second medical image identification model, if there is at least one film reading device from which the update success message is not received, the server sends the second medical image identification model again, and continues to determine whether an update success message is received from the server, if the update success message is not received yet, the server notifies a manager associated with the film reading device, so that the manager manually updates or replaces the medical image identification model in the film reading device.
By the method, the user confidence is introduced, the image analysis result modification confidence of the medical image recognition model is updated based on the user confidence, the model is updated only when the image analysis result modification confidence meets the conditions, and the problem caused by frequent updating of the model is avoided.
In the invention, before updating the model, the distribution of data sources such as hospitals or users in the second database is counted, when the duty ratio of one hospital or user exceeds a corresponding preset value, the first medical image identification model is not updated, and when the duty ratio does not exceed the preset value, the first medical image identification model is updated, at the moment, even if the data corresponding to the hospital or user has deviation, the data are discarded as bad data in the model updating process, so that the error or the error of individual hospitals or individual users in reading the film can be avoided to cause the system model to be updated in error. In addition, by the method, when the duty ratio of a certain hospital or a user exceeds a corresponding preset value, the data corresponding to the hospital or the user is deleted, and the image analysis result modification confidence of the medical image identification model is updated, so that the data cannot influence the updating time and result of the model.
The embodiment of the invention provides a system for updating a medical image recognition model, which is shown in fig. 2, and comprises at least one medical imaging device, at least one film reading device and at least one server;
the medical imaging device is used for generating a target medical image;
the film reading device is used for acquiring a target medical image from medical imaging equipment; identifying the target medical image by using a first medical image identification model to obtain a first image analysis result corresponding to the target medical image; receiving a user instruction and judging whether the received instruction is a confirmation instruction or a modification instruction; if the target medical image is the confirmation instruction, the target medical image and the first image analysis result are sent to a server; if the target medical image is the modification instruction, modifying the first image analysis result according to the modification instruction, generating a second image analysis result, and sending the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to a server; receiving a second medical image recognition model from a server, replacing a local first medical image recognition model by adopting the second medical image recognition model, and sending an update success message to the server after successful replacement;
the server is used for correspondingly storing the target medical image received from the film reading device and the first image analysis result in a first database as one data record; correspondingly storing the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction as a data record in a second database, and updating the image analysis result modification confidence of the first medical image identification model, wherein the method specifically comprises the following steps: determining user confidence coefficient according to the user identification, updating the image analysis result modification confidence coefficient based on the user confidence coefficient, wherein the updated image analysis result modification confidence coefficient is the image analysis result modification confidence coefficient before updating plus the user confidence coefficient corresponding to the user giving the modification instruction; judging whether the image analysis result modification confidence of the first medical image recognition model reaches a first preset value or not, if so, updating the first medical image recognition model based on data stored in a second database to obtain a second medical image recognition model, and sending the second medical image recognition model to all the film reading devices;
wherein, the value of the user confidence is between 0 and 1; the first preset value is a positive integer greater than 1.
Preferably, the server is further configured to obtain a plurality of medical images and image analysis results corresponding to the medical images, form a training set, generate a first medical image recognition model based on the training set through machine learning training or neural network training, and send the generated first medical image recognition model to all the film reading devices.
In another embodiment, the server is further configured to obtain a plurality of medical images of a certain type and image analysis results corresponding to the medical images, form a training set, generate a first medical image recognition model corresponding to the medical images of the type through machine learning training or neural network training based on the training set, and send the generated first medical image recognition model to all the film reading devices capable of reading the medical images of the type.
The film reading device is also used for prompting a user that the first image analysis result is generated, displaying the target medical image and the first image analysis result in response to the operation of the user, and simultaneously displaying a confirmation control and a modification control on an interface for displaying the target medical image and the first image analysis result. If the user selects the confirmation control, the film reading device receives the confirmation instruction, and if the user selects the modification control, the film reading device receives the modification instruction.
Preferably, the film reading device is further configured to prompt the user that the first image analysis result has been generated, and respond to the first voice command of the user to display the target medical image and the first image analysis result, and obtain a second voice command of the user, where if the second voice command includes a positive keyword, the film reading device receives a confirmation command, and if the second voice command includes a negative keyword, the film reading device receives a modification command.
Preferably, the server is further configured to store a correspondence between the user identifier and the user level, and determine the user level according to the correspondence and the received user identifier, thereby determining the user confidence level.
Preferably, the server is further configured to store a correspondence between the user identifier and the user medical portrait, and determine the user medical portrait according to the correspondence and the received user identifier, thereby determining the user confidence level.
Preferably, the server is further configured to determine that the user confidence coefficient is the lowest value when the received identifier of the user is not found in the corresponding relationship.
Preferably, the server is further configured to delete, from the second database, a data record corresponding to the target medical image, the first image analysis result, and the second image analysis result, which are simultaneously transmitted to the server with the identifier of the user, when the identifier of the received user is not found in the correspondence.
Preferably, the film reading device is further configured to send the target medical image, the first image analysis result, the second image analysis result, and the identifier of the user who gives the modification instruction to a server, and simultaneously send the hospital to which the film reading device belongs to the server, where the server correspondingly stores the data as a data record in a second database.
The server is further configured to determine whether the image analysis result modification confidence coefficient reaches a first preset value, if so, further count distribution of the identifiers of the hospitals to which the film reading device belongs and the users who give the modification instructions in each piece of data record stored in the second database, if the counted distribution result meets both the first condition and the second condition, update the first medical image recognition model based on the data stored in the second database to obtain a second medical image recognition model, and if the image analysis result modification confidence coefficient does not reach the first preset value or the counted distribution result does not meet at least one of the first condition and the second condition, not update the first medical image recognition model. The first condition is that the statistical duty ratio of a hospital to which any piece reading device belongs in the distribution results obtained through statistics does not exceed a first preset value, and the second condition is that the statistical duty ratio of the identification of any user in the distribution results obtained through statistics does not exceed a second preset value.
And the server is further configured to delete a data record corresponding to the hospital to which the film reading device belongs from the second database if the statistical duty ratio of the hospital to which the film reading device belongs in the distribution result obtained by statistics exceeds a first preset value, update the image analysis result modification confidence level based on the data record in the second database, and delete the data record corresponding to the identification of the user from the second database if the statistical duty ratio of the identification of the at least one user in the distribution result obtained by statistics exceeds a second preset value, and update the image analysis result modification confidence level based on the data record in the second database.
The film reading device is further used for determining the type of the target medical image, and the first medical image identification model corresponding to the type is used for identifying the target medical image. The server is further configured to store the target medical image and the first image analysis result as one data record in a first database corresponding to the type, store the target medical image, the first image analysis result and the second image analysis result as one data record in a second database corresponding to the type, and send the second medical image identification model to all the film reading devices capable of reading the target medical image of the type.
The film reading device of the present invention may include a processor and a memory storing computer instructions executable by the processor, which when executed by the processor, perform the method steps performed by the film reading device.
The server of the present invention may comprise a processor and a memory storing computer instructions executable by the processor, which when executed by the processor, perform the method steps performed by the server.
An embodiment of the invention provides a computer-readable storage medium storing computer instructions for implementing a method as described above.
Any combination of one or more computer readable media may be employed. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. The computer readable storage medium may include: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), a flash memory, an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
Computer program code for carrying out operations of the present invention may be written in one or more programming languages, or combinations thereof
The above description is only an example for the convenience of understanding the present invention, and is not intended to limit the scope of the present invention. In the specific implementation, the person skilled in the art may change, increase, decrease the components of the apparatus according to the actual situation, and may change, increase, decrease or change the order of the steps of the method according to the actual situation on the basis of not affecting the functions implemented by the method.
While embodiments of the invention have been shown and described, it will be understood by those skilled in the art that: various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents, and modifications which are not to be construed as being within the scope of the invention.
Claims (10)
1. A method for updating a medical image recognition model, the method comprising the steps of:
step 101, at least one film reading device acquires a target medical image from at least one medical imaging device;
102, the film reading device uses a first medical image recognition model to recognize the target medical image, and a first image analysis result corresponding to the target medical image is obtained;
step 103, the film reading device receives a user instruction and judges whether the received instruction is a confirmation instruction or a modification instruction; if the instruction is a confirmation instruction, executing step 104; if it is a modification instruction, steps 105-107 are performed;
104, the film reading device sends the target medical image and the first image analysis result to a server, and the server correspondingly stores the target medical image and the first image analysis result in a first database as one data record;
step 105, the film reading device modifies the first image analysis result according to the modification instruction to generate a second image analysis result; the film reading device sends the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to a server, the server correspondingly stores the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction in a second database as one data record, and the server updates the image analysis result modification confidence of the first medical image recognition model, specifically: the server determines the user confidence coefficient according to the user identification, updates the image analysis result modification confidence coefficient based on the user confidence coefficient, and the updated image analysis result modification confidence coefficient is the image analysis result modification confidence coefficient before updating plus the user confidence coefficient corresponding to the user giving the modification instruction; the value of the user confidence is between 0 and 1;
step 106, the server judges whether the image analysis result modification confidence of the first medical image recognition model reaches a first preset value, if so, the first medical image recognition model is updated based on the data stored in a second database, and a second medical image recognition model is obtained; wherein the first preset value is a positive integer greater than 1;
step 107, the server sends the second medical image recognition model to all the film reading devices, the film reading devices replace the local first medical image recognition model by the second medical image recognition model, and after the replacement is successful, an update success message is sent to the server.
2. The method of claim 1, wherein the step of determining the position of the substrate comprises,
before step 101, a server acquires a plurality of medical images and corresponding image analysis results thereof to form a training set, generates a first medical image recognition model through machine learning training or neural network training based on the training set, and sends the generated first medical image recognition model to a film reading device.
3. The method of claim 1, wherein the step of determining the position of the substrate comprises,
the server stores the corresponding relation between the user identification and the user level, and the server determines the user level according to the corresponding relation and the user identification received from the film reading device, so as to determine the user confidence level; or,
the server stores the correspondence between the user's identification and the user's medical representation, and the server determines the user's medical representation, and thus the user's confidence level, based on the correspondence and the user's identification received from the reader.
4. The method of claim 1, wherein the step of determining the position of the substrate comprises,
the medical portrait of the user giving the modification instruction comprises parameters of the hospital level, the user level, the working years, the number of the read pictures and the patient evaluation; the user confidence is a weighted sum of the parameters included in the medical representation.
5. The method of claim 3, wherein the step of,
when the identification of the user received from the film reading device is not found in the corresponding relation, determining that the user confidence coefficient is the lowest value; or,
and deleting the data record corresponding to the user identification from the second database when the user identification received from the film reading device is not found in the corresponding relation.
6. A system for updating a medical image recognition model, the system comprising at least one medical imaging device, at least one reading apparatus, and at least one server;
the medical imaging device is used for generating a target medical image;
the film reading device is used for acquiring a target medical image from medical imaging equipment; identifying the target medical image by using a first medical image identification model to obtain a first image analysis result corresponding to the target medical image; receiving a user instruction and judging whether the received instruction is a confirmation instruction or a modification instruction; if the target medical image is the confirmation instruction, the target medical image and the first image analysis result are sent to a server; if the target medical image is the modification instruction, modifying the first image analysis result according to the modification instruction, generating a second image analysis result, and sending the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction to a server; receiving a second medical image recognition model from a server, replacing a local first medical image recognition model by adopting the second medical image recognition model, and sending an update success message to the server after successful replacement;
the server is used for correspondingly storing the target medical image received from the film reading device and the first image analysis result in a first database as one data record; correspondingly storing the target medical image, the first image analysis result, the second image analysis result and the identification of the user giving the modification instruction as a data record in a second database, and updating the image analysis result modification confidence of the first medical image identification model, wherein the method specifically comprises the following steps: determining user confidence coefficient according to the user identification, updating the image analysis result modification confidence coefficient based on the user confidence coefficient, wherein the updated image analysis result modification confidence coefficient is the image analysis result modification confidence coefficient before updating plus the user confidence coefficient corresponding to the user giving the modification instruction; judging whether the image analysis result modification confidence of the first medical image recognition model reaches a first preset value or not, if so, updating the first medical image recognition model based on data stored in a second database to obtain a second medical image recognition model, and sending the second medical image recognition model to all the film reading devices;
wherein, the value of the user confidence is between 0 and 1; the first preset value is a positive integer greater than 1.
7. The system of claim 6, wherein the system further comprises a controller configured to control the controller,
the server is further used for acquiring a plurality of medical images and corresponding image analysis results thereof to form a training set, generating a first medical image recognition model through machine learning training or neural network training based on the training set, and sending the generated first medical image recognition model to all the film reading devices.
8. The system of claim 6, wherein the system further comprises a controller configured to control the controller,
the server is also used for storing the corresponding relation between the user identification and the user level, determining the user level according to the corresponding relation and the user identification received from the film reading device, and further determining the user confidence coefficient; or,
the server is also used for storing the corresponding relation between the user identification and the user medical portrait, determining the user medical portrait according to the corresponding relation and the user identification received from the film reading device, and further determining the user confidence coefficient.
9. The system of claim 8, wherein the system further comprises a controller configured to control the controller,
the medical portrait of the user giving the modification instruction comprises parameters of the hospital level, the user level, the working years, the number of the read pictures and the patient evaluation; the user confidence is a weighted sum of the parameters included in the medical representation.
10. The system of claim 6, wherein the system further comprises a controller configured to control the controller,
the server is further configured to: when the identification of the user received from the film reading device is not found in the corresponding relation, determining that the user confidence coefficient is the lowest value; or deleting the data record corresponding to the user identification from the second database when the user identification received from the film reading device is not found in the corresponding relation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910976085.XA CN110853737B (en) | 2019-10-15 | 2019-10-15 | Updating method and system of medical image recognition model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910976085.XA CN110853737B (en) | 2019-10-15 | 2019-10-15 | Updating method and system of medical image recognition model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110853737A CN110853737A (en) | 2020-02-28 |
| CN110853737B true CN110853737B (en) | 2023-05-23 |
Family
ID=69597567
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910976085.XA Expired - Fee Related CN110853737B (en) | 2019-10-15 | 2019-10-15 | Updating method and system of medical image recognition model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110853737B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4109463A1 (en) * | 2021-06-24 | 2022-12-28 | Siemens Healthcare GmbH | Providing a second result dataset |
| CN114783575B (en) * | 2022-04-20 | 2023-09-29 | 广州唯顶软件科技有限公司 | Medical image processing system and method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108447557A (en) * | 2018-02-08 | 2018-08-24 | 国政通科技股份有限公司 | A kind of medical analysis method and device based on deep learning |
| EP3432313A1 (en) * | 2017-07-18 | 2019-01-23 | Koninklijke Philips N.V. | Training an image analysis system |
| CN109492675A (en) * | 2018-10-22 | 2019-03-19 | 深圳前海达闼云端智能科技有限公司 | Recognition methods, device, storage medium and the electronic equipment of medical image |
| CN109830284A (en) * | 2017-11-23 | 2019-05-31 | 天启慧眼(北京)信息技术有限公司 | The analysis method and device of medical image |
| CN109919928A (en) * | 2019-03-06 | 2019-06-21 | 腾讯科技（深圳）有限公司 | Detection method, device and the storage medium of medical image |
| CN109935305A (en) * | 2017-12-18 | 2019-06-25 | 天启慧眼(北京)信息技术有限公司 | Detect the methods, devices and systems of image |
| CN109934220A (en) * | 2019-02-22 | 2019-06-25 | 上海联影智能医疗科技有限公司 | A kind of methods of exhibiting, device and the terminal of image point of interest |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8223143B2 (en) * | 2006-10-27 | 2012-07-17 | Carl Zeiss Meditec, Inc. | User interface for efficiently displaying relevant OCT imaging data |
| US9177102B2 (en) * | 2011-04-28 | 2015-11-03 | Bioptigen, Inc. | Database and imaging processing system and methods for analyzing images acquired using an image acquisition system |
| DE102016219488A1 (en) * | 2016-10-07 | 2018-04-12 | Siemens Healthcare Gmbh | Method for providing a confidence information |
-
2019
- 2019-10-15 CN CN201910976085.XA patent/CN110853737B/en not_active Expired - Fee Related
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3432313A1 (en) * | 2017-07-18 | 2019-01-23 | Koninklijke Philips N.V. | Training an image analysis system |
| CN109830284A (en) * | 2017-11-23 | 2019-05-31 | 天启慧眼(北京)信息技术有限公司 | The analysis method and device of medical image |
| CN109935305A (en) * | 2017-12-18 | 2019-06-25 | 天启慧眼(北京)信息技术有限公司 | Detect the methods, devices and systems of image |
| CN108447557A (en) * | 2018-02-08 | 2018-08-24 | 国政通科技股份有限公司 | A kind of medical analysis method and device based on deep learning |
| CN109492675A (en) * | 2018-10-22 | 2019-03-19 | 深圳前海达闼云端智能科技有限公司 | Recognition methods, device, storage medium and the electronic equipment of medical image |
| CN109934220A (en) * | 2019-02-22 | 2019-06-25 | 上海联影智能医疗科技有限公司 | A kind of methods of exhibiting, device and the terminal of image point of interest |
| CN109919928A (en) * | 2019-03-06 | 2019-06-21 | 腾讯科技（深圳）有限公司 | Detection method, device and the storage medium of medical image |
Non-Patent Citations (1)
| Title |
|---|
| 魏小娜 等.医学影像人工智能辅助诊断的样本增广方法.《计算机应用》.2019,第39卷(第9期),第2558-2567页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110853737A (en) | 2020-02-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11227384B2 (en) | Methods and systems for determining a diagnostically unacceptable medical image | |
| US20190088359A1 (en) | System and Method for Automated Analysis in Medical Imaging Applications | |
| CN111696083B (en) | Image processing method and device, electronic equipment and storage medium | |
| US10339655B2 (en) | Automated image evaluation in x-ray imaging | |
| CN109492675B (en) | Medical image recognition method and device, storage medium and electronic equipment | |
| JP6768620B2 (en) | Learning support device, operation method of learning support device, learning support program, learning support system, terminal device and program | |
| CN110853737B (en) | Updating method and system of medical image recognition model | |
| JP6727176B2 (en) | Learning support device, method of operating learning support device, learning support program, learning support system, and terminal device | |
| CN112101162B (en) | Image recognition model generation method and device, storage medium and electronic equipment | |
| JP2008059071A (en) | Medical image processor | |
| CN111260647B (en) | CT scanning auxiliary method based on image detection, computer readable storage medium and CT scanning device | |
| CN114796891A (en) | Radiotherapy system | |
| US12118724B2 (en) | Interactive coronary labeling using interventional x-ray images and deep learning | |
| JP2020027507A (en) | Medical information processing device, medical information processing method, and program | |
| CN111986182A (en) | Auxiliary diagnosis method, system, electronic device and storage medium | |
| US9031284B2 (en) | Implant identification system and method | |
| JP2008161532A (en) | Medical image display device and program | |
| KR101611024B1 (en) | Method and system for managing tooth information service | |
| JP3594250B2 (en) | PACS | |
| US20230281817A1 (en) | Medical imaging apparatus for obtaining medical image of equine and operating method thereof | |
| KR102472600B1 (en) | Apparatus and method for recognizing implant fixtures | |
| CN112530580A (en) | Medical image picture processing method and computer readable storage medium | |
| EP4354452A1 (en) | Medical image search and retrieval | |
| CN116402584B (en) | Event generation method based on multiple data sources, storage medium and electronic equipment | |
| US20230107439A1 (en) | Medical image processing apparatus, medical image processing method, and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20230523 |