WO2022077914A1 - 医学图片优化方法、装置、设备及计算机可读存储介质 - Google Patents

医学图片优化方法、装置、设备及计算机可读存储介质 Download PDF

Info

Publication number
WO2022077914A1
WO2022077914A1 PCT/CN2021/096529 CN2021096529W WO2022077914A1 WO 2022077914 A1 WO2022077914 A1 WO 2022077914A1 CN 2021096529 W CN2021096529 W CN 2021096529W WO 2022077914 A1 WO2022077914 A1 WO 2022077914A1
Authority
WO
WIPO (PCT)
Prior art keywords
picture
medical
script
sample
duplicate
Prior art date
Application number
PCT/CN2021/096529
Other languages
English (en)
French (fr)
Inventor
王俊
高鹏
谢国彤
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2022077914A1 publication Critical patent/WO2022077914A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/29—Graphical models, e.g. Bayesian networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/60—Rotation of whole images or parts thereof
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Definitions

  • the present application relates to the technical field of image processing, and in particular, to a medical image optimization method, apparatus, device, and computer-readable storage medium.
  • the commonly used image target detection method is based on the deep learning model, but before using the deep learning model for detection, the model needs to be trained. Since a large number of high-quality sample pictures are required for the training of the deep learning model, how to optimize the high-quality samples from the massive sample pictures Pictures are extremely important.
  • the inventor realizes that the current medical picture optimization mainly uses a classification algorithm, and the classification algorithm is used to judge whether each picture is a high-quality sample picture, so as to eliminate the pictures with poor quality and achieve the purpose of optimization.
  • the classification algorithm cannot automatically expand the image data of the sample images, and can only optimize according to the existing sample images, so the sample image data cannot be fully utilized, resulting in a serious waste of image features, and the calculation method is slightly simpler, and the optimization accuracy is low. The phenomenon.
  • a medical image optimization method comprising:
  • transformation perturbation script set performing transformation perturbation on the medical sample picture set to obtain a plurality of medical replica picture sets
  • the sample pictures of the medical sample picture set are screened, and the duplicate pictures in a plurality of the medical duplicate picture sets are screened to obtain an optimized picture set.
  • a medical picture optimization device includes:
  • a transformation perturbation script generation module used for acquiring a medical sample picture set, and generating a transformation perturbation script set according to the medical sample picture set;
  • a duplicate picture generation module configured to transform and perturb the medical sample picture set by using the transform perturbation script set to obtain multiple medical duplicate picture sets
  • the image recognition module is configured to use a pre-built target detection model to identify the medical sample image set and a plurality of the medical duplicate image sets, respectively, to obtain a sample detection frame set and a duplicate detection frame set;
  • the picture optimization module is used for screening the sample pictures of the medical sample picture set according to the sample detection frame set and the duplicate detection frame set, and screening a plurality of duplicate pictures in the medical duplicate picture set to obtain optimized pictures set.
  • An electronic device comprising:
  • a processor that executes the instructions stored in the memory to achieve the following steps:
  • transformation perturbation script set performing transformation perturbation on the medical sample picture set to obtain a plurality of medical replica picture sets
  • the sample pictures of the medical sample picture set are screened, and the duplicate pictures in the medical duplicate picture set are screened to obtain an optimized picture set.
  • a computer-readable storage medium comprising a storage data area and a storage program area, the storage data area stores data created, and the storage program area stores a computer program; wherein, the computer program is executed by a processor The following steps are implemented:
  • transformation perturbation script set performing transformation perturbation on the medical sample picture set to obtain a plurality of medical replica picture sets
  • the sample pictures of the medical sample picture set are screened, and the duplicate pictures in a plurality of the medical duplicate picture sets are screened to obtain an optimized picture set.
  • the present application can solve the phenomenon of image feature waste and low optimization accuracy.
  • FIG. 1 is a schematic flowchart of a medical image optimization method provided by an embodiment of the present application.
  • FIG. 2 is a schematic flowchart of one of the steps of a medical image optimization method provided by an embodiment of the present application
  • FIG. 3 is a schematic flowchart of another step of the method for optimizing a medical image provided by an embodiment of the present application
  • FIG. 4 is a schematic block diagram of a medical image optimization apparatus provided by an embodiment of the present application.
  • FIG. 5 is a schematic diagram of an internal structure of an electronic device for implementing a method for optimizing a medical picture provided by an embodiment of the present application;
  • the embodiment of the present application provides a medical picture optimization method.
  • the execution body of the medical picture optimization method includes, but is not limited to, at least one of electronic devices that can be configured to execute the method provided by the embodiments of the present application, such as a server and a terminal.
  • the medical picture optimization method can be executed by software or hardware installed in a terminal device or a server device, and the software can be a blockchain platform.
  • the server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
  • the present application provides a medical image optimization method.
  • FIG. 1 a schematic flowchart of a method for optimizing a medical image provided by an embodiment of the present application is shown.
  • the medical picture optimization method includes:
  • the medical sample picture set includes images of people, animals, and objects in the medical field.
  • the medical image set can be obtained by using the crawler technology in the currently known programming languages such as java and Python, for example, the medical image set includes human organ images, lesion images, and the like.
  • each medical sample picture can be obtained, and each medical sample picture is first uploaded to the redis cache, and then each medical sample picture in the redis cache is uploaded to the database to obtain the medical sample picture. Sample image set.
  • the transformation perturbation script is a technical means of performing transformation and perturbation operations on the medical sample picture set, so as to realize the expansion of the medical sample picture set, which can quickly enrich the diversity and generalization of the medical sample picture set. advantages of chemistry.
  • the transformation disturbance script set includes a picture size change script set and a picture color change script set.
  • the generating a set of transformation perturbation scripts according to the medical sample picture set includes:
  • the template of the color dithering script can be edited in advance, and the exposure, saturation and hue are filled into the template as parameter values to obtain the color dithering script.
  • templates for generating contrast transformation scripts can be pre-edited.
  • picture size change script set includes: rotation transformation script, flip transformation script, zoom transformation script, translation transformation script, scale transformation script, region cropping script and random masking script.
  • each transformation perturbation script in the transformation perturbation script set is a script program automatically constructed by programming languages such as java and Python. Perform transformation operations such as contrast transformation and color dithering.
  • the S2 includes:
  • part of the content in the sample image is moved along the x or y axis to obtain a copy image.
  • using the picture color change script set to perform color dithering and contrast transformation on the medical sample picture set to obtain a multi-color picture replica set includes: using the contrast transformation program to change the medical sample picture The saturation value and brightness value of each sample picture in the set; perform exponential operation on the changed saturation value and brightness value to obtain the illumination value, and use the illumination value to change the contrast of the sample picture to obtain a multi-contrast picture A replica set; using the color dithering script, randomly transform the exposure and color tone of each picture in the multi-contrast picture replica set to obtain the multi-color picture replica set.
  • x represents the saturation value or brightness value of each sample picture value
  • the s value is set between 0.25 and 4, so that the changed saturation and brightness values are calculated.
  • the S3 can be replaced by: using a pre-built target probability prediction model to identify the medical sample picture set and a plurality of the medical duplicate picture sets respectively, and obtain the sample picture category probability set and Replica image category probability set.
  • the target probability prediction model can predict the category probability of a sample picture or a duplicate picture.
  • a sample picture and a plurality of duplicate pictures are input into the target probability prediction model, and the output category probability is 0.9 for the sample picture.
  • the corresponding duplicate pictures are 0.95, 0.89, 0.92, ..., 0.97. It can be seen that the difference between the category probability of the sample picture and the duplicate picture is small, which means that the consistency is high.
  • the output category probability is 0.9 for the sample picture and the corresponding duplicate picture.
  • the pictures are 0.5, 0.3, 0.7, .
  • the target detection model or the target probability prediction model can be constructed based on a convolutional application network.
  • the S4 includes: calculating the degree of coincidence of the sample detection frame set and the duplicate detection frame set; calculating the similarity between the medical sample picture set and a plurality of the medical duplicate picture sets, based on the similarity The degree and the degree of coincidence are calculated to obtain a picture score set, and the picture score set is used to screen the sample pictures of the medical sample picture set and the duplicate pictures in the multiple medical duplicate picture sets to obtain an optimized picture set.
  • the following formula is used to calculate the degree of coincidence of the sample detection frame set and the duplicate detection frame set:
  • IOU represents the degree of coincidence between the sample image and the copy image
  • A represents the detection frame of the sample sample
  • B represents the detection frame of any copy image generated by A through transformation perturbation
  • ⁇ represents the intersection in the mathematical formula
  • ⁇ represents the mathematical formula union of .
  • the calculation method of the similarity is:
  • J(A, B) represents the similarity between any sample image and the duplicate image
  • A represents the detection frame of the sample image
  • B represents the detection frame of the duplicate image generated by the sample image A through transformation perturbation
  • ⁇ represents the intersection in the mathematical formula
  • the picture score set is calculated by the following method:
  • S i represents the picture score set
  • ci represents the category consistent score set calculated by using the coincidence degree
  • b i represents the coincidence degree
  • mi represents the similarity degree set.
  • a preset threshold is used to compare the scores in the picture score set. If the score is greater than the preset threshold, it is a required sample, and if the score is less than or equal to the preset threshold , no sample is needed, so that the filtered optimized picture set is obtained.
  • FIG. 4 it is a schematic diagram of a module of the medical image optimization apparatus of the present application.
  • the medical picture optimization apparatus 100 described in this application can be installed in an electronic device. According to the realized functions, the medical picture optimization apparatus 100 may include a transformation disturbance script generation module 101 , a duplicate picture generation module 102 , a picture identification module 103 and a picture optimization module 104 .
  • the modules described in the embodiments of the present application may also be referred to as units, which refer to a series of computer program segments that can be executed by the electronic device processor and can perform fixed functions, and are stored in the memory of the electronic device.
  • each module/unit is as follows:
  • the transformation perturbation script generation module 101 is configured to obtain a medical sample picture set, and generate a transformation perturbation script set according to the medical sample picture set;
  • the medical sample picture set includes images of various types such as people, animals, objects, medicine, etc.
  • the medical sample picture set of the medical category can be obtained from the medical image database of the hospital.
  • the medical image set can also be obtained by using the crawler technology in the currently known programming languages such as java and Python, for example, the medical image set includes human organ images, lesion images, and the like.
  • transformation perturbation script generation module 101 can obtain each sample picture, and upload each sample picture to the redis cache first, and then upload each sample picture in the redis cache to the database to obtain the above. Medical sample picture set.
  • the transformation perturbation script is a technical means of performing transformation and perturbation operations on the medical sample picture set, so as to realize the expansion of the medical sample picture set, which can quickly enrich the diversity and generalization of the medical sample picture set. advantages of chemistry.
  • the transformation disturbance script set includes a picture size change script set and a picture color change script set.
  • the transformation perturbation script generation module 101 generates a transformation perturbation script set according to the medical sample picture set, including: generating a color jitter script by using the exposure, saturation and hue of the medical sample picture set; judging the Whether each picture in the medical sample picture set is an HSV color space picture, if each picture is an HSV color space picture, a contrast transformation script is generated; the color dithering script and the contrast transformation script are collected to obtain the picture color Change script set.
  • the transformation disturbance script generation module 101 can pre-edit the template of the color dithering script, and fill in the exposure, saturation and hue as parameter values into the template to obtain the color dithering script .
  • the transformation perturbation script generation module 101 may pre-edit a template for generating a contrast transformation script.
  • picture size change script set includes: rotation transformation script, flip transformation script, zoom transformation script, translation transformation script, scale transformation script, region cropping script and random masking script.
  • each transformation perturbation script in the transformation perturbation script set is a script program automatically constructed by programming languages such as java and Python. Perform transformation operations such as contrast transformation and color dithering.
  • the duplicate picture generation module 102 is configured to use the transform perturbation script set to perform transform perturbation on the medical sample picture set to obtain multiple medical duplicate picture sets.
  • the duplicate picture generation module 102 uses the following operations to transform and perturb the medical sample picture set to obtain a plurality of medical duplicate picture sets: using the picture size change script set, respectively for the medical sample picture set Perform rotation, flip, zoom, translation, scale transformation and area cropping to obtain a multi-size picture copy set; use the picture color change script set to perform color dithering and contrast transformation on the medical sample picture set to obtain a multi-color picture copy collection; summarizing the multi-size image copy set and the multi-color image copy set to obtain a plurality of the medical copy image sets.
  • One of the embodiments of the present application may move some content in the sample picture along the x or y axis direction according to the translation transformation script, so as to obtain the duplicate picture.
  • using the picture color change script set to perform color dithering and contrast transformation on the medical sample picture set to obtain a multi-color picture replica set includes: using the contrast transformation program to change the medical sample picture The saturation value and brightness value of each sample picture in the set; perform exponential operation on the changed saturation value and brightness value to obtain the illumination value, and use the illumination value to change the contrast of the sample picture to obtain a multi-contrast picture A replica set; using the color dithering script, randomly transform the exposure and color tone of each picture in the multi-contrast picture replica set to obtain the multi-color picture replica set.
  • a pre-built exponential function s x is used, and the value of s is set between 0.25 and 4, where x represents the saturation value or brightness value of each sample image.
  • the picture recognition module 103 is configured to use a pre-built target detection model to identify the medical sample picture set and a plurality of the medical duplicate picture sets, respectively, to obtain a sample detection frame set and a duplicate detection frame set.
  • the pre-built target detection model is used to identify the medical sample picture set and a plurality of the medical duplicate picture sets, respectively, to obtain a sample detection frame set and a duplicate detection frame set, which can be replaced
  • the steps are: using a pre-built target probability prediction model to identify the medical sample picture set and a plurality of the medical duplicate picture sets, respectively, to obtain a sample picture category probability set and a duplicate picture category probability set.
  • the target probability prediction model can predict the category probability of a sample picture or a duplicate picture.
  • a sample picture and a plurality of duplicate pictures are input into the target probability prediction model, and the output category probability is 0.9 for the sample picture.
  • the corresponding duplicate pictures are 0.95, 0.89, 0.92, ..., 0.97. It can be seen that the difference between the category probability of the sample picture and the duplicate picture is small, which means that the consistency is high.
  • the output category probability is 0.9 for the sample picture and the corresponding duplicate picture.
  • the pictures are 0.5, 0.3, 0.7, .
  • the target detection model or the target probability prediction model can be constructed based on a convolutional application network.
  • the picture optimization module 104 is configured to screen the sample pictures of the medical sample picture set according to the sample detection frame set and the duplicate detection frame set, and to filter duplicate pictures in the medical duplicate picture set to obtain Optimize the picture set.
  • the sample pictures of the medical sample picture set are screened, and the duplicate pictures in the medical duplicate picture set are screened to obtain an optimized picture set , including: calculating the degree of coincidence of the sample detection frame set and the duplicate detection frame set; calculating the similarity of the medical sample picture set and a plurality of the medical duplicate picture sets; based on the similarity and the coincidence, A picture scoring set is obtained by calculation, and the picture scoring set is used to screen the sample pictures of the medical sample picture set and the duplicate pictures in the plurality of medical duplicate picture sets to obtain an optimized picture set.
  • the following formula is used to calculate the degree of coincidence of the sample detection frame set and the duplicate detection frame set:
  • IOU represents the degree of coincidence between the sample image and the copy image
  • A represents the detection frame of the sample sample
  • B represents the detection frame of any copy image generated by A through transformation perturbation
  • ⁇ represents the intersection in the mathematical formula
  • ⁇ represents the mathematical formula union of .
  • the calculation method of the similarity is:
  • J(A, B) represents the similarity between any sample image and the duplicate image
  • A represents the detection frame of the sample image
  • B represents the detection frame of the duplicate image generated by the sample image A through transformation perturbation
  • ⁇ represents the intersection in the mathematical formula
  • the picture score set is calculated by the following method:
  • S i represents the picture score set
  • ci represents the category consistent score set calculated by using the coincidence degree
  • b i represents the coincidence degree
  • mi represents the similarity degree set.
  • a preset threshold is used to compare the scores in the picture score set. If the score is greater than the preset threshold, it is a required sample, and if the score is less than or equal to the preset threshold , no sample is needed, so that the filtered optimized picture set is obtained.
  • a medical sample picture set is used to generate a transformation perturbation script set, and through the transformation perturbation script set, the medical sample picture set is transformed and perturbed to obtain multiple medical duplicate picture sets.
  • the transformation perturbation script set includes multiple transformation perturbations Script
  • each transformation perturbation script can transform and perturb the sample pictures, so that the number of medical copy picture sets generated is huge. Therefore, in the embodiment of the present application, the sample pictures can be used to achieve the purpose of automatically expanding the pictures, and the phenomenon of waste of the sample pictures will not be caused.
  • the embodiment of the present application is based on a pre-built target detection model, which not only identifies the accuracy of medical sample picture sets, but also identifies the accuracy of medical duplicate picture sets.
  • the image recognition process can further improve the recognition ability of the target detection model and improve the accuracy. Therefore, the medical image optimization method, device and computer-readable storage medium proposed in this application can solve the phenomenon of image feature waste and low optimization accuracy.
  • FIG. 5 it is a schematic structural diagram of an electronic device for implementing the medical picture optimization method of the present application.
  • the electronic device 1 may include a processor 10, a memory 11 and a bus, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a medical image optimization program 12.
  • the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, magnetic disk, CD etc.
  • the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as a mobile hard disk of the electronic device 1 .
  • the memory 11 may also be an external storage device of the electronic device 1, such as a pluggable mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital) equipped on the electronic device 1. , SD) card, flash memory card (Flash Card), etc.
  • the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device.
  • the memory 11 can not only be used to store application software installed in the electronic device 1 and various types of data, such as the codes of the medical image optimization program 12, etc., but also can be used to temporarily store data that has been output or will be output.
  • the processor 10 may be composed of integrated circuits, for example, may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits packaged with the same function or different functions, including one or more integrated circuits.
  • Central Processing Unit CPU
  • microprocessor digital processing chip
  • graphics processor and combination of various control chips, etc.
  • the processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect the various components of the entire electronic device, by running or executing the program or module (for example, executing the program) stored in the memory 11. Medical picture optimization program, etc.), and call the data stored in the memory 11 to perform various functions of the electronic device 1 and process data.
  • the bus may be a peripheral component interconnect (PCI for short) bus or an extended industry standard architecture (Extended industry standard architecture, EISA for short) bus or the like.
  • PCI peripheral component interconnect
  • EISA Extended industry standard architecture
  • the bus can be divided into address bus, data bus, control bus and so on.
  • the bus is configured to implement connection communication between the memory 11 and at least one processor 10 and the like.
  • FIG. 5 only shows an electronic device with components. Those skilled in the art can understand that the structure shown in FIG. 5 does not constitute a limitation on the electronic device 1, and may include fewer or more components than those shown in the drawings. components, or a combination of certain components, or a different arrangement of components.
  • the electronic device 1 may also include a power supply (such as a battery) for powering the various components, preferably, the power supply may be logically connected to the at least one processor 10 through a power management device, so that the power management
  • the device implements functions such as charge management, discharge management, and power consumption management.
  • the power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and any other components.
  • the electronic device 1 may further include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.
  • the electronic device 1 may also include a network interface, optionally, the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used in the electronic device 1 Establish a communication connection with other electronic devices.
  • a network interface optionally, the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used in the electronic device 1 Establish a communication connection with other electronic devices.
  • the electronic device 1 may further include a user interface, and the user interface may be a display (Display), an input unit (eg, a keyboard (Keyboard)), optionally, the user interface may also be a standard wired interface or a wireless interface.
  • the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, and the like.
  • the display may also be appropriately called a display screen or a display unit, which is used for displaying information processed in the electronic device 1 and for displaying a visualized user interface.
  • the medical picture optimization program 12 stored in the memory 11 in the electronic device 1 is a combination of multiple instructions, and when running in the processor 10, it can realize:
  • transformation perturbation script set performing transformation perturbation on the medical sample picture set to obtain a plurality of medical replica picture sets
  • the sample pictures of the medical sample picture set are screened, and the duplicate pictures in a plurality of the medical duplicate picture sets are screened to obtain an optimized picture set medical sample picture set medical Copy picture set.
  • the modules/units integrated in the electronic device 1 may be stored in a computer-readable storage medium.
  • the computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, removable hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory) .
  • the computer-readable storage medium may mainly include a program storage area and a storage data area, and the computer-readable storage medium may be volatile or non-volatile, wherein the program storage area can store An operating system, an application program required for at least one function, etc.; the storage data area can store data created according to the use of blockchain nodes, etc., and the application program is executed by the processor to implement the following steps:
  • transformation perturbation script set performing transformation perturbation on the medical sample picture set to obtain a plurality of medical replica picture sets
  • the sample pictures of the medical sample picture set are screened, and the duplicate pictures in a plurality of the medical duplicate picture sets are screened to obtain an optimized picture set.
  • modules described as separate components may or may not be physically separated, and components shown as modules may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment.
  • each functional module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit.
  • the above-mentioned integrated units can be implemented in the form of hardware, or can be implemented in the form of hardware plus software function modules.
  • the blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

一种医学图片优化方法、装置、电子设备及计算机可读存储介质，所述方法包括：获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集（S1），利用所述变换扰动脚本集对所述医学样本图片集进行变换扰动，得到多个医学副本图片集（S2），利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集医学样本图片集医学副本图片集（S3）；根据所述样本检测框集及所述副本检测框集，优化得到优化图片集。所述方法可以解决医学图片特征浪费、优化准确率较低的现象。

Description

医学图片优化方法、装置、设备及计算机可读存储介质
本申请要求于2020年10月15日提交中国专利局、申请号为CN202011103243.X，发明名称为“医学图片优化方法、装置、设备及计算机可读存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及图片处理技术领域，尤其涉及一种医学图片优化方法、装置、设备及计算机可读存储介质。
背景技术
目前常用的图片目标检测手段是基于深度学习模型，但利用深度学习模型进行检测之前需要训练模型，由于深度学习模型训练时需要大量的高质量样本图片，如何从海量样本图片中优化得到高质量样本图片是极其重要的。
发明人意识到当前医学图片优化主要使用分类算法，通过分类算法判断每张图片是否是高质量的样本图片，从而剔除质量较差的图片，达到优化目的。但是分类算法无法自动扩充样本图片的图片数据，只能根据已有样本图片进行优化，因此无法充分利用样本图片数据，造成图片特征浪费严重，且计算方法稍显简单，进而出现优化准确率较低的现象。
发明内容
一种医学图片优化方法，包括：
获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本识别准确率集及副本识别准确率集；
根据所述样本识别准确率集及副本识别准确率集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
一种医学图片优化装置，所述装置包括：
变换扰动脚本生成模块，用于获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
副本图片生成模块，用于利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
图片识别模块，用于利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集；
图片优化模块，用于根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
一种电子设备，所述电子设备包括：
存储器，存储至少一个指令；及
处理器，执行所述存储器中存储的指令以实现如下步骤：
获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本识别准确率集及副本识别准确率集；
根据所述样本识别准确率集及副本识别准确率集，筛选所述医学样本图片集的样本图 片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
一种计算机可读存储介质，包括存储数据区和存储程序区，存储数据区存储创建的数据，存储程序区存储有计算机程序；其中，所述计算机程序被处理器执行时实现如下步骤：
获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本识别准确率集及副本识别准确率集；
根据所述样本识别准确率集及副本识别准确率集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
本申请可以解决图片特征浪费、优化准确率较低的现象。
附图说明
图1为本申请一实施例提供的医学图片优化方法的流程示意图；
图2为本申请一实施例提供的医学图片优化方法的其中一个步骤的流程示意图；
图3为本申请一实施例提供的医学图片优化方法的另一个步骤的流程示意图；
图4为本申请一实施例提供的医学图片优化装置的模块示意图；
图5为本申请一实施例提供的实现医学图片优化方法的电子设备的内部结构示意图；
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
具体实施方式
应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请实施例提供一种医学图片优化方法。所述医学图片优化方法的执行主体包括但不限于服务端、终端等能够被配置为执行本申请实施例提供的该方法的电子设备中的至少一种。换言之，所述医学图片优化方法可以由安装在终端设备或服务端设备的软件或硬件来执行，所述软件可以是区块链平台。所述服务端包括但不限于：单台服务器、服务器集群、云端服务器或云端服务器集群等。
本申请提供一种医学图片优化方法。参照图1所示，为本申请一实施例提供的医学图片优化方法的流程示意图。在本实施例中，所述医学图片优化方法包括：
S1、获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集。
本申请较佳实施例中，所述医学样本图片集包括人、动物、物体在医学领域中的图像，如医学领域中，可从医院的医学图像数据库中得到医学类的医学样本图片集，也可利用当今已知的java、Python等编程语言中的爬虫技术来获取该医学图像集，如医学图像集中包含人的器官图像、病灶图像等。
进一步的，本申请实施例可获取每张医学样本图片，并将每张医学样本图片先上传到redis缓存中，然后将redis缓存中的每张医学样本图片均上传至数据库中，得到所述医学样本图片集。
较佳的，所述变换扰动脚本是对所述医学样本图片集进行变换扰动操作，以来实现所述医学样本图片集扩增的技术手段，具有快速丰富所述医学样本图片集的多样性及泛化能力的优点。优选地，本申请实施例中，所述变换扰动脚本集包括图片尺寸变化脚本集及图片颜色变化脚本集。
详细地，参阅图2所示，所述根据所述医学样本图片集生成变换扰动脚本集，包括：
S11、利用所述医学样本图片集的曝光度、饱和度及色调，生成颜色抖动脚本；
本申请较佳实施例中，可预先编辑好颜色抖动脚本的模板，将曝光度、饱和度及色调 作为参数值填入所述模板内，得到所述颜色抖动脚本。
S12、判断所述医学样本图片集内每张图片是否为HSV颜色空间图片，若每张图片均为HSV颜色空间图片，则生成对比度变换脚本；
同样地，可预先编辑好生成对比度变换脚本的模板。
S13、汇集所述颜色抖动脚本及所述对比度变换脚本，得到所述图片颜色变化脚本集。
进一步地，所述图片尺寸变化脚本集包括：旋转变换脚本、翻转变脚本、缩放变换脚本、平移变换脚本、尺度变换脚本、区域裁剪脚本及随机掩盖脚本。
本申请实施例中，所述变换扰动脚本集中每种变换扰动脚本均是通过java、Python等编程语言自动构建的脚本程序，将医学样本图片集作为该脚本程序输入参数并运行脚本程序，可自动进行对比度变换、颜色抖动等变换操作。
S2、利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集。
详细地，参阅图3所示，所述S2包括：
S21、利用所述图片尺寸变化脚本集，分别对所述医学样本图片集进行旋转、翻转、缩放、平移、尺度变换及区域裁剪，得到多尺寸图片副本集；
如按照平移变换脚本，将样本图片内部分内容沿x或y轴方向进行移动，从而得到副本图片。
S22、利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集；
详细地，所述利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集，包括：利用所述对比度变换程序，改变所述医学样本图片集内每个样本图片的饱和度值和亮度值；将改变后的所述饱和度值和亮度值进行指数运算得到光照值，利用所述光照值改变所述样本图片的对比度，得到多对比度图片副本集；利用所述颜色抖动脚本，随机变换所述多对比度图片副本集内每张图片的曝光度及色调，得到所述多颜色图片副本集。
本申请较佳实施例中，如使用预先构建的指数函数，改变所述医学样本图片集内每个样本图片的饱和度值和亮度值，x表示的是每个样本图片的饱和度值或亮度值，s数值设置在0.25到4之间，从而计算得到改变后的饱和度值和亮度值。
S23、汇总所述多尺寸图片副本集及所述多颜色图片副本集，得到多个所述医学副本图片集。
S3、利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集。医学样本图片集医学副本图片集。
本申请较佳实施例中，所述S3可以被替换为：利用预构建的目标概率预测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本图片类别概率集及副本图片类别概率集。
详细地，所述目标概率预测模型可以预测出样本图片或副本图片的类别概率，如向所述目标概率预测模型中输入某个样本图片及多个副本图片，输出的类别概率分别为样本图片0.9，对应的副本图片为0.95、0.89、0.92、…、0.97，可见样本图片与副本图片的类别概率差值较小，从而表示一致性较高，输出的类别概率分别为样本图片0.9，对应的副本图片为0.5、0.3、0.7、…、0.8，可见样本图片与副本图片的类别概率差值较大，从而表示一致性较低。
本申请较佳实施例中，所述目标检测模型或所述目标概率预测模型都可用以卷积申请网络为基础构建得到。
S4、根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
详细地，所述S4包括：计算所述样本检测框集及所述副本检测框集的重合度；计算所述医学样本图片集及多个所述医学副本图片集的相似度，基于所述相似度和重合度，计算得到图片评分集，利用所述图片评分集，筛选所述医学样本图片集的样本图片，及多个所述医学副本图片集内的副本图片，得到优化图片集。
本申请较佳实施例中，利用如下公式计算所述样本检测框集及所述副本检测框集的重合度：
Figure PCTCN2021096529-appb-000001
其中，IOU表示样本图片与副本图片的重合度，A表示样本样本的检测框，B表示A通过变换扰动生成的任意一个副本图片的检测框，∩表示数学公式中的交集，∪表示数学公式中的并集。
所述相似度的计算方法为：
Figure PCTCN2021096529-appb-000002
其中，J(A,B)表示任意样本图片与副本图片的相似度，A表示样本图片的检测框，B表示样本图片A通过变换扰动生成的副本图片的检测框，∩表示数学公式中的交集。
本申请较佳实施例中，利用如下方法计算所述图片评分集：
Figure PCTCN2021096529-appb-000003
其中，S i表示所述图片评分集，c i表示利用所述重合度计算出的类别一致评分集集，b i表示所述重合度，m i表示所述相似度集。
较佳地，本申请实施例利用预设的阈值与所述图片评分集中的评分进行比较，若所述评分大于预设的阈值，则为所需样本，若所述评分小于等于预设的阈值，则为无需样本，从而得到筛选后的所述优化图片集。
如图4所示，是本申请医学图片优化装置的模块示意图。
本申请所述医学图片优化装置100可以安装于电子设备中。根据实现的功能，所述医学图片优化装置100可以包括变换扰动脚本生成模块101、副本图片生成模块102、图片识别模块103及图片优化模块104。本申请实施例所述模块也可以称之为单元，是指一种能够被电子设备处理器所执行，并且能够完成固定功能的一系列计算机程序段，其存储在电子设备的存储器中。
在本实施例中，关于各模块/单元的功能如下：
所述变换扰动脚本生成模块101，用于获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
本申请较佳实施例中，所述医学样本图片集包括人、动物、物体、医学等各个类型的图像，如医学领域中，可从医院的医学图像数据库中得到医学类的医学样本图片集，也可利用当今已知的java、Python等编程语言中的爬虫技术来获取该医学图像集，如医学图像集中包含人的器官图像、病灶图像等。
进一步的，所述变换扰动脚本生成模块101可获取每张样本图片，并将每张样本图片先上传到redis缓存中，然后将redis缓存中的每张样本图片均上传至数据库中，得到所述医学样本图片集。
较佳的，所述变换扰动脚本是对所述医学样本图片集进行变换扰动操作，以来实现所述医学样本图片集扩增的技术手段，具有快速丰富所述医学样本图片集的多样性及泛化能力的优点。优选地，本申请实施例中，所述变换扰动脚本集包括图片尺寸变化脚本集及图片颜色变化脚本集。
详细地，所述变换扰动脚本生成模块101根据所述医学样本图片集生成变换扰动脚本集，包括：利用所述医学样本图片集的曝光度、饱和度及色调，生成颜色抖动脚本；判断所述医学样本图片集内每张图片是否为HSV颜色空间图片，若每张图片均为HSV颜色空间图片，则生成对比度变换脚本；汇集所述颜色抖动脚本及所述对比度变换脚本，得到所述图片颜色变化脚本集。
本申请较佳实施例中，所述变换扰动脚本生成模块101可预先编辑好颜色抖动脚本的模板，将曝光度、饱和度及色调作为参数值填入所述模板内，得到所述颜色抖动脚本。
同样地，所述变换扰动脚本生成模块101可预先编辑好生成对比度变换脚本的模板。
进一步地，所述图片尺寸变化脚本集包括：旋转变换脚本、翻转变脚本、缩放变换脚本、平移变换脚本、尺度变换脚本、区域裁剪脚本及随机掩盖脚本。
本申请实施例中，所述变换扰动脚本集中每种变换扰动脚本均是通过java、Python等编程语言自动构建的脚本程序，将医学样本图片集作为该脚本程序输入参数并运行脚本程序，可自动进行对比度变换、颜色抖动等变换操作。
所述副本图片生成模块102用于利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集。
详细地，所述副本图片生成模块102采用下述操作对所述医学样本图片集进行变换扰动，得到多个医学副本图片集：利用所述图片尺寸变化脚本集，分别对所述医学样本图片集进行旋转、翻转、缩放、平移、尺度变换及区域裁剪，得到多尺寸图片副本集；利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集；汇总所述多尺寸图片副本集及所述多颜色图片副本集，得到多个所述医学副本图片集。
本申请其中一个实施例可以按照平移变换脚本，将样本图片内部分内容沿x或y轴方向进行移动，从而得到副本图片。
详细地，所述利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集，包括：利用所述对比度变换程序，改变所述医学样本图片集内每个样本图片的饱和度值和亮度值；将改变后的所述饱和度值和亮度值进行指数运算得到光照值，利用所述光照值改变所述样本图片的对比度，得到多对比度图片副本集；利用所述颜色抖动脚本，随机变换所述多对比度图片副本集内每张图片的曝光度及色调，得到所述多颜色图片副本集。
本申请较佳实施例中，如使用预先构建的指数函数s x，将s数值设置在0.25到4之间，x表示的是每个样本图片的饱和度值或亮度值。
所述图片识别模块103用于利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集。
本申请较佳实施例中，所述利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集，可以被替换为：利用预构建的目标概率预测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本图片类别概率集及副本图片类别概率集。
详细地，所述目标概率预测模型可以预测出样本图片或副本图片的类别概率，如向所述目标概率预测模型中输入某个样本图片及多个副本图片，输出的类别概率分别为样本图片0.9，对应的副本图片为0.95、0.89、0.92、…、0.97，可见样本图片与副本图片的类别概率差值较小，从而表示一致性较高，输出的类别概率分别为样本图片0.9，对应的副本 图片为0.5、0.3、0.7、…、0.8，可见样本图片与副本图片的类别概率差值较大，从而表示一致性较低。
本申请较佳实施例中，所述目标检测模型或所述目标概率预测模型都可用以卷积申请网络为基础构建得到。
所述图片优化模块104用于根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
详细地，所述根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集，包括：计算所述样本检测框集及所述副本检测框集的重合度；计算所述医学样本图片集及多个所述医学副本图片集的相似度；基于所述相似度和重合度，计算得到图片评分集，利用所述图片评分集，筛选所述医学样本图片集的样本图片，及多个所述医学副本图片集内的副本图片，得到优化图片集。
本申请较佳实施例中，利用如下公式计算所述样本检测框集及所述副本检测框集的重合度：
Figure PCTCN2021096529-appb-000004
其中，IOU表示样本图片与副本图片的重合度，A表示样本样本的检测框，B表示A通过变换扰动生成的任意一个副本图片的检测框，∩表示数学公式中的交集，∪表示数学公式中的并集。
所述相似度的计算方法为：
Figure PCTCN2021096529-appb-000005
其中，J(A,B)表示任意样本图片与副本图片的相似度，A表示样本图片的检测框，B表示样本图片A通过变换扰动生成的副本图片的检测框，∩表示数学公式中的交集。
本申请较佳实施例中，利用如下方法计算所述图片评分集：
Figure PCTCN2021096529-appb-000006
其中，S i表示所述图片评分集，c i表示利用所述重合度计算出的类别一致评分集集，b i表示所述重合度，m i表示所述相似度集。
较佳地，本申请实施例利用预设的阈值与所述图片评分集中的评分进行比较，若所述评分大于预设的阈值，则为所需样本，若所述评分小于等于预设的阈值，则为无需样本，从而得到筛选后的所述优化图片集。
本申请实施例利用医学样本图片集生成变换扰动脚本集，通过变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集，由于变换扰动脚本集包括多个变换扰动脚本，每个变换扰动脚本均可对样本图片变换扰动，从而生成的医学副本图片集数量庞大，故本申请实施例可利用样本图片完成自动扩充图片的目的，不会造成样本图片浪费的现象，其次本申请实施例基于预构建的目标检测模型，不仅识别医学样本图片集的准确率，同时识别医学副本图片集的准确率，相比于仅单一识别样本图片来说，本申请添加了对副本图片的识别过程，可进一步提高目标检测模型的识别能力，提高准确率。因此本申请提出的医学图片优化方法、装置及计算机可读存储介质，可以解决图片特征浪费、优化准确率较低的现象。
如图5所示，是本申请实现医学图片优化方法的电子设备的结构示意图。
所述电子设备1可以包括处理器10、存储器11和总线，还可以包括存储在所述存储器11中并可在所述处理器10上运行的计算机程序，如医学图片优化程序12。
其中，所述存储器11至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、移动硬盘、多媒体卡、卡型存储器(例如：SD或DX存储器等)、磁性存储器、磁盘、光盘等。所述存储器11在一些实施例中可以是电子设备1的内部存储单元，例如该电子设备1的移动硬盘。所述存储器11在另一些实施例中也可以是电子设备1的外部存储设备，例如电子设备1上配备的插接式移动硬盘、智能存储卡(Smart Media Card，SMC)、安全数字(Secure Digital，SD)卡、闪存卡(Flash Card)等。进一步地，所述存储器11还可以既包括电子设备1的内部存储单元也包括外部存储设备。所述存储器11不仅可以用于存储安装于电子设备1的应用软件及各类数据，例如医学图片优化程序12的代码等，还可以用于暂时地存储已经输出或者将要输出的数据。
所述处理器10在一些实施例中可以由集成电路组成，例如可以由单个封装的集成电路所组成，也可以是由多个相同功能或不同功能封装的集成电路所组成，包括一个或者多个中央处理器(Central Processing unit，CPU)、微处理器、数字处理芯片、图形处理器及各种控制芯片的组合等。所述处理器10是所述电子设备的控制核心(Control Unit)，利用各种接口和线路连接整个电子设备的各个部件，通过运行或执行存储在所述存储器11内的程序或者模块(例如执行医学图片优化程序等)，以及调用存储在所述存储器11内的数据，以执行电子设备1的各种功能和处理数据。
所述总线可以是外设部件互连标准(peripheral component interconnect，简称PCI)总线或扩展工业标准结构(extended industry standard architecture，简称EISA)总线等。该总线可以分为地址总线、数据总线、控制总线等。所述总线被设置为实现所述存储器11以及至少一个处理器10等之间的连接通信。
图5仅示出了具有部件的电子设备，本领域技术人员可以理解的是，图5示出的结构并不构成对所述电子设备1的限定，可以包括比图示更少或者更多的部件，或者组合某些部件，或者不同的部件布置。
例如，尽管未示出，所述电子设备1还可以包括给各个部件供电的电源(比如电池)，优选地，电源可以通过电源管理装置与所述至少一个处理器10逻辑相连，从而通过电源管理装置实现充电管理、放电管理、以及功耗管理等功能。电源还可以包括一个或一个以上的直流或交流电源、再充电装置、电源故障检测电路、电源转换器或者逆变器、电源状态指示器等任意组件。所述电子设备1还可以包括多种传感器、蓝牙模块、Wi-Fi模块等，在此不再赘述。
进一步地，所述电子设备1还可以包括网络接口，可选地，所述网络接口可以包括有线接口和/或无线接口(如WI-FI接口、蓝牙接口等)，通常用于在该电子设备1与其他电子设备之间建立通信连接。
可选地，该电子设备1还可以包括用户接口，用户接口可以是显示器(Display)、输入单元(比如键盘(Keyboard))，可选地，用户接口还可以是标准的有线接口、无线接口。可选地，在一些实施例中，显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode，有机发光二极管)触摸器等。其中，显示器也可以适当的称为显示屏或显示单元，用于显示在电子设备1中处理的信息以及用于显示可视化的用户界面。
应该了解，所述实施例仅为说明之用，在专利申请范围上并不受此结构的限制。
所述电子设备1中的所述存储器11存储的医学图片优化程序12是多个指令的组合，在所述处理器10中运行时，可以实现：
获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集医学样本图片集医学副本图片集；
根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集医学样本图片集医学副本图片集。
进一步地，所述电子设备1集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。所述计算机可读介质可以包括：能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM，Read-Only Memory)。
进一步地，所述计算机可读存储介质可主要包括存储程序区和存储数据区，所述计算机可读存储介质可以是易失性的，也可以是非易失性的，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序等；存储数据区可存储根据区块链节点的使用所创建的数据等，所述应用程序被处理器执行时实现如下步骤：
获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本识别准确率集及副本识别准确率集；
根据所述样本识别准确率集及副本识别准确率集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
在本申请所提供的几个实施例中，应该理解到，所揭露的设备，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述模块的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。
所述作为分离部件说明的模块可以是或者也可以不是物理上分开的，作为模块显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。
另外，在本申请各个实施例中的各功能模块可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能模块的形式实现。
对于本领域技术人员而言，显然本申请不限于上述示范性实施例的细节，而且在不背离本申请的精神或基本特征的情况下，能够以其他的具体形式实现本申请。
因此，无论从哪一点来看，均应将实施例看作是示范性的，而且是非限制性的，本申请的范围由所附权利要求而不是上述说明限定，因此旨在将落在权利要求的等同要件的含义和范围内的所有变化涵括在本申请内。不应将权利要求中的任何附关联图表记视为限制所涉及的权利要求。
本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
最后应说明的是，以上实施例仅用以说明本申请的技术方案而非限制，尽管参照较佳实施例对本申请进行了详细说明，本领域的普通技术人员应当理解，可以对本申请的技术方案进行修改或等同替换，而不脱离本申请技术方案的精神和范围。

Claims (20)

  1. 一种医学图片优化方法，其中，所述方法包括：
    获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
    利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
    利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集医学样本图片集医学副本图片集；
    根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集医学样本图片集医学副本图片集。
  2. 如权利要求1所述的医学图片优化方法，其中，所述变换扰动脚本集包括图片尺寸变化脚本集及图片颜色变化脚本集。
  3. 如权利要求2所述的医学图片优化方法，其中，所述根据所述医学样本图片集生成变换扰动脚本集，包括：
    利用所述医学样本图片集内每张图片的曝光度、饱和度及色调，生成颜色抖动脚本；
    判断所述医学样本图片集内每张图片是否为HSV颜色空间图片，若每张图片均为HSV颜色空间图片，则生成对比度变换脚本；
    汇集所述颜色抖动脚本及所述对比度变换脚本，得到所述图片颜色变化脚本集。
  4. 如权利要求2所述的医学图片优化方法，其中，所述尺寸变化脚本集包括：旋转变换脚本、翻转变换脚本、缩放变换脚本、平移变换脚本、尺度变换脚本、区域裁剪脚本及/或随机掩盖脚本。
  5. 如权利要求2所述的医学图片优化方法，其中，所述利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集，包括：
    利用所述图片尺寸变化脚本集，分别对所述医学样本图片集进行旋转、翻转、缩放、平移、尺度变换及区域裁剪，得到多尺寸图片副本集；
    利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集；
    汇总所述多尺寸图片副本集及所述多颜色图片副本集，得到多个所述医学副本图片集。
  6. 如权利要求3所述的医学图片优化方法，其中，所述利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集，包括：
    利用所述对比度变换脚本，改变所述医学样本图片集内每个样本图片的饱和度值和亮度值；
    将改变后的所述饱和度值和亮度值进行指数运算得到光照值，利用所述光照值改变所述样本图片的对比度，得到多对比度图片副本集；
    利用所述颜色抖动脚本，随机变换所述多对比度图片副本集内每张图片的曝光度及色调，得到所述多颜色图片副本集。
  7. 如权利要求1至6中任意一项所述的医学图片优化方法，其中，所述根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集，包括：
    计算所述样本检测框集及所述副本检测框集的重合度；
    计算所述医学样本图片集及多个所述医学副本图片集的相似度；
    基于所述相似度和重合度，计算得到图片评分集，利用所述图片评分集，筛选所述医学样本图片集的样本图片，及多个所述医学副本图片集内的副本图片，得到优化图片集。
  8. 一种医学图片优化装置，其中，所述装置包括：
    变换扰动脚本生成模块，用于获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
    副本图片生成模块，用于利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
    图片识别模块，用于利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集；
    图片优化模块，用于根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集。
  9. 一种电子设备，其中，所述电子设备包括：
    至少一个处理器；以及
    与所述至少一个处理器通信连接的存储器；
    其中，所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如下步骤：
    获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
    利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
    利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集医学样本图片集医学副本图片集；
    根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集医学样本图片集医学副本图片集。
  10. 如权利要求9所述的电子设备，其中，所述变换扰动脚本集包括图片尺寸变化脚本集及图片颜色变化脚本集。
  11. 如权利要求10所述的电子设备，其中，所述根据所述医学样本图片集生成变换扰动脚本集，包括：
    利用所述医学样本图片集内每张图片的曝光度、饱和度及色调，生成颜色抖动脚本；
    判断所述医学样本图片集内每张图片是否为HSV颜色空间图片，若每张图片均为HSV颜色空间图片，则生成对比度变换脚本；
    汇集所述颜色抖动脚本及所述对比度变换脚本，得到所述图片颜色变化脚本集。
  12. 如权利要求10所述的电子设备，其中，所述尺寸变化脚本集包括：旋转变换脚本、翻转变换脚本、缩放变换脚本、平移变换脚本、尺度变换脚本、区域裁剪脚本及/或随机掩盖脚本。
  13. 如权利要求10所述的电子设备，其中，所述利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集，包括：
    利用所述图片尺寸变化脚本集，分别对所述医学样本图片集进行旋转、翻转、缩放、平移、尺度变换及区域裁剪，得到多尺寸图片副本集；
    利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集；
    汇总所述多尺寸图片副本集及所述多颜色图片副本集，得到多个所述医学副本图片集。
  14. 如权利要求11所述的电子设备，其中，所述利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集，包括：
    利用所述对比度变换脚本，改变所述医学样本图片集内每个样本图片的饱和度值和亮度值；
    将改变后的所述饱和度值和亮度值进行指数运算得到光照值，利用所述光照值改变所述样本图片的对比度，得到多对比度图片副本集；
    利用所述颜色抖动脚本，随机变换所述多对比度图片副本集内每张图片的曝光度及色调，得到所述多颜色图片副本集。
  15. 如权利要求9至14中任意一项所述的电子设备，其中，所述根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集，包括：
    计算所述样本检测框集及所述副本检测框集的重合度；
    计算所述医学样本图片集及多个所述医学副本图片集的相似度；
    基于所述相似度和重合度，计算得到图片评分集，利用所述图片评分集，筛选所述医学样本图片集的样本图片，及多个所述医学副本图片集内的副本图片，得到优化图片集。
  16. 一种计算机可读存储介质，包括存储数据区和存储程序区，存储数据区存储创建的数据，存储程序区存储有计算机程序；其中，所述计算机程序被处理器执行时实现如下步骤：
    获取医学样本图片集，根据所述医学样本图片集生成变换扰动脚本集；
    利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集；
    利用预构建的目标检测模型，分别识别所述医学样本图片集及多个所述医学副本图片集，得到样本检测框集及副本检测框集医学样本图片集医学副本图片集；
    根据所述样本检测框集及所述副本检测框集，筛选所述医学样本图片集的样本图片，及筛选多个所述医学副本图片集内的副本图片，得到优化图片集医学样本图片集医学副本图片集。
  17. 如权利要求16所述的计算机可读存储介质，其中，所述变换扰动脚本集包括图片尺寸变化脚本集及图片颜色变化脚本集。
  18. 如权利要求17所述的计算机可读存储介质，其中，所述根据所述医学样本图片集生成变换扰动脚本集，包括：
    利用所述医学样本图片集内每张图片的曝光度、饱和度及色调，生成颜色抖动脚本；
    判断所述医学样本图片集内每张图片是否为HSV颜色空间图片，若每张图片均为HSV颜色空间图片，则生成对比度变换脚本；
    汇集所述颜色抖动脚本及所述对比度变换脚本，得到所述图片颜色变化脚本集。
  19. 如权利要求17所述的计算机可读存储介质，其中，所述尺寸变化脚本集包括：旋转变换脚本、翻转变换脚本、缩放变换脚本、平移变换脚本、尺度变换脚本、区域裁剪脚本及/或随机掩盖脚本。
  20. 如权利要求17所述的计算机可读存储介质，其中，所述利用所述变换扰动脚本集，对所述医学样本图片集进行变换扰动，得到多个医学副本图片集，包括：
    利用所述图片尺寸变化脚本集，分别对所述医学样本图片集进行旋转、翻转、缩放、平移、尺度变换及区域裁剪，得到多尺寸图片副本集；
    利用所述图片颜色变化脚本集，对所述医学样本图片集进行颜色抖动及对比度变换，得到多颜色图片副本集；
    汇总所述多尺寸图片副本集及所述多颜色图片副本集，得到多个所述医学副本图片集。
PCT/CN2021/096529 2020-10-15 2021-05-27 医学图片优化方法、装置、设备及计算机可读存储介质 WO2022077914A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011103243.X 2020-10-15
CN202011103243.XA CN112233194B (zh) 2020-10-15 2020-10-15 医学图片优化方法、装置、设备及计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2022077914A1 true WO2022077914A1 (zh) 2022-04-21

Family

ID=74113638

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/096529 WO2022077914A1 (zh) 2020-10-15 2021-05-27 医学图片优化方法、装置、设备及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN112233194B (zh)
WO (1) WO2022077914A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112233194B (zh) * 2020-10-15 2023-06-02 平安科技（深圳）有限公司 医学图片优化方法、装置、设备及计算机可读存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109801275A (zh) * 2019-01-11 2019-05-24 北京邮电大学 基于图像识别的马铃薯病害检测方法和系统
CN110298829A (zh) * 2019-06-19 2019-10-01 上海国民集团健康科技有限公司 一种舌诊方法、装置、系统、计算机设备和存储介质
WO2019233355A1 (en) * 2018-06-08 2019-12-12 Guangdong Oppo Mobile Telecommunications Corp., Ltd. Method and apparatus for training, classification model, mobile terminal, and readable storage medium
CN111275129A (zh) * 2020-02-17 2020-06-12 平安科技（深圳）有限公司 一种图像数据的增广策略选取方法及系统
CN111476760A (zh) * 2020-03-17 2020-07-31 平安科技（深圳）有限公司 医学图像的生成方法、装置、电子设备及介质
CN112233194A (zh) * 2020-10-15 2021-01-15 平安科技（深圳）有限公司 医学图片优化方法、装置、设备及计算机可读存储介质

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109427060A (zh) * 2018-10-30 2019-03-05 腾讯科技（深圳）有限公司 一种影像识别的方法、装置、终端设备和医疗系统
CN109829920B (zh) * 2019-02-25 2021-06-15 上海商汤智能科技有限公司 图像处理方法及装置、电子设备和存储介质
CN109978886B (zh) * 2019-04-01 2021-11-09 北京市商汤科技开发有限公司 图像处理方法及装置、电子设备和存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019233355A1 (en) * 2018-06-08 2019-12-12 Guangdong Oppo Mobile Telecommunications Corp., Ltd. Method and apparatus for training, classification model, mobile terminal, and readable storage medium
CN109801275A (zh) * 2019-01-11 2019-05-24 北京邮电大学 基于图像识别的马铃薯病害检测方法和系统
CN110298829A (zh) * 2019-06-19 2019-10-01 上海国民集团健康科技有限公司 一种舌诊方法、装置、系统、计算机设备和存储介质
CN111275129A (zh) * 2020-02-17 2020-06-12 平安科技（深圳）有限公司 一种图像数据的增广策略选取方法及系统
CN111476760A (zh) * 2020-03-17 2020-07-31 平安科技（深圳）有限公司 医学图像的生成方法、装置、电子设备及介质
CN112233194A (zh) * 2020-10-15 2021-01-15 平安科技（深圳）有限公司 医学图片优化方法、装置、设备及计算机可读存储介质

Also Published As

Publication number Publication date
CN112233194A (zh) 2021-01-15
CN112233194B (zh) 2023-06-02

Similar Documents

Publication Publication Date Title
WO2021189906A1 (zh) 基于联邦学习的目标检测方法、装置、设备及存储介质
WO2021151338A1 (zh) 医学影像图片分析方法、装置、电子设备及可读存储介质
WO2021208703A1 (zh) 问题解析方法、装置、电子设备及存储介质
WO2022116424A1 (zh) 交通流预测模型训练方法、装置、电子设备及存储介质
WO2021189909A1 (zh) 病灶检测分析方法、装置、电子设备及计算机存储介质
WO2022222943A1 (zh) 科室推荐方法、装置、电子设备及存储介质
WO2022179123A1 (zh) 数据更新及展示方法、装置、电子设备及存储介质
WO2022048209A1 (zh) 车牌识别方法、装置、电子设备及存储介质
WO2022105179A1 (zh) 生物特征图像识别方法、装置、电子设备及可读存储介质
WO2021189829A1 (zh) 数据查询方法、装置、电子设备及存储介质
WO2021189827A1 (zh) 识别模糊图像的方法、装置、设备及计算机可读存储介质
WO2022048210A1 (zh) 命名实体识别方法、装置、电子设备及可读存储介质
WO2021189911A1 (zh) 基于视频流的目标物位置检测方法、装置、设备及介质
WO2021135474A1 (zh) 多数据来源的数据融合方法、装置、电子设备及存储介质
WO2022088632A1 (zh) 用户数据监控分析方法、装置、设备及介质
WO2022048204A1 (zh) 图像生成方法、装置、电子设备及计算机可读存储介质
WO2023050651A1 (zh) 图像语义分割方法、装置、设备及存储介质
WO2022194062A1 (zh) 疾病标签检测方法、装置、电子设备及存储介质
WO2021238563A1 (zh) 基于配置算法的企业运行数据分析方法、装置、电子设备及介质
WO2022016977A1 (zh) 线上实时数据交互方法、装置、电子设备及存储介质
WO2021184576A1 (zh) 医学图像的生成方法、装置、电子设备及介质
WO2022227192A1 (zh) 图像分类方法、装置、电子设备及介质
WO2022077914A1 (zh) 医学图片优化方法、装置、设备及计算机可读存储介质
WO2022179122A1 (zh) 基于大数据的数据存储方法、装置、电子设备及存储介质
WO2022048362A1 (zh) 数据存储方法、装置、电子设备及存储介质

Legal Events

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

Ref document number: 21878968

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21878968

Country of ref document: EP

Kind code of ref document: A1