WO2021189909A1 - 病灶检测分析方法、装置、电子设备及计算机存储介质 - Google Patents

病灶检测分析方法、装置、电子设备及计算机存储介质 Download PDF

Info

Publication number
WO2021189909A1
WO2021189909A1 PCT/CN2020/131989 CN2020131989W WO2021189909A1 WO 2021189909 A1 WO2021189909 A1 WO 2021189909A1 CN 2020131989 W CN2020131989 W CN 2020131989W WO 2021189909 A1 WO2021189909 A1 WO 2021189909A1
Authority
WO
WIPO (PCT)
Prior art keywords
slice
lesion
input
slices
standard data
Prior art date
Application number
PCT/CN2020/131989
Other languages
English (en)
French (fr)
Inventor
高良心
刘新卉
叶苓
李康
李楠楠
黄凌云
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 WO2021189909A1 publication Critical patent/WO2021189909A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0012Biomedical image inspection
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • G06N3/084Backpropagation, e.g. using gradient descent
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/11Region-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10072Tomographic images
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20081Training; Learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20084Artificial neural networks [ANN]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30096Tumor; Lesion
    • YGENERAL 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
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02ATECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Definitions

  • This application relates to the field of artificial intelligence technology, and in particular to a lesion detection and analysis method, device, electronic equipment, and computer-readable storage medium.
  • Lesion detection, analysis and segmentation is an important branch.
  • Computers are used to automatically identify the lesion area in the image, determine the severity of the patient, and facilitate follow-up doctors to make detailed diagnosis.
  • the existing lesion detection and analysis methods are based on deep learning algorithms.
  • the inventor realizes that the segmentation accuracy of lesions is low, and only the area size of the lesion is detected, and no further analysis of each lesion area is performed, so that It is difficult to detect the cure if the quantitative index of the volume of the lesion area has not changed significantly.
  • a lesion detection and analysis method provided in this application includes:
  • Select one of the slices in the standard data set calculate the slice step length according to the standard data set, and select a total of N slices before and after the slice in the standard data set according to the slice step length to obtain an input slice set ;
  • the present application also provides a lesion detection and analysis device, which includes:
  • a data processing module configured to obtain CT slice data, and perform a normalization operation on the CT slice data to obtain a standard data set
  • the input slice set acquisition module is used to select one of the slices in the standard data set, calculate the slice step length according to the standard data set, and select the total before and after the slice in the standard data set according to the slice step length N slices, get the input slice set;
  • the lesion area determination module is configured to input the input slice set into a pre-built lesion segmentation model, use the lesion segmentation model to perform hole convolution on the input slice set, extract feature data of the input slice set, and Determine the lesion area according to the characteristic data;
  • the density analysis module is used to perform density analysis on the lesion area, and feed back the result of the density analysis to the user.
  • an electronic device which includes:
  • At least one processor and,
  • a memory communicatively connected with the at least one processor; wherein,
  • the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor, so that the at least one processor can execute the following steps:
  • Select one of the slices in the standard data set calculate the slice step length according to the standard data set, and select a total of N slices before and after the slice in the standard data set according to the slice step length to obtain an input slice set ;
  • this application also provides a computer-readable storage medium, including a storage data area and a storage program area.
  • the storage data area stores created data
  • the storage program area stores a computer program; wherein the computer program is processed
  • Select one of the slices in the standard data set calculate the slice step length according to the standard data set, and select a total of N slices before and after the slice in the standard data set according to the slice step length to obtain an input slice set ;
  • FIG. 1 is a schematic flowchart of a lesion detection and analysis method provided by an embodiment of the application
  • Figure 2 is a schematic diagram of a detailed implementation process of one of the steps in Figure 1;
  • Fig. 3 is a schematic diagram of a detailed implementation process of another step in Fig. 1;
  • FIG. 4 is a schematic diagram of modules of a lesion detection and analysis device provided by an embodiment of the application.
  • FIG. 5 is a schematic diagram of the internal structure of an electronic device that implements a lesion detection and analysis method provided by an embodiment of the application;
  • the execution body of the lesion detection and analysis method provided in the embodiment of the present application includes but is not limited to at least one of the electronic devices that can be configured to execute the method provided in the embodiment of the present application, such as a server and a terminal.
  • the lesion detection and analysis 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, etc.
  • This application provides a method for detection and analysis of lesions.
  • FIG. 1 it is a schematic flowchart of a lesion detection and analysis method provided by an embodiment of this application.
  • the method can be executed by a device, and the device can be implemented by software and/or hardware.
  • the lesion detection and analysis method includes:
  • the CT slice data in the embodiments of the present application includes a plurality of sequentially arranged medical image pictures, such as a lung CT image of a patient with pneumonia, which can be obtained from a database of a medical device.
  • the CT slice data can be obtained from a node of a blockchain.
  • the security of the CT slice data can be improved.
  • performing a normalization operation on the CT slice data to obtain a standard data set includes:
  • MN (x, y, z) is normalized data
  • MO (x, y, z) is the three-dimensional matrix
  • the above-mentioned formula is used to normalize the three-dimensional matrix, which is to calculate each element in the three-dimensional matrix with the formula to obtain the calculation result; if the calculation If the result is greater than 1, the value of the element is assigned as 1; if the calculation result is less than 0, the value of the element is assigned as 0.
  • the normalization of the feature data can ensure that each Each feature is treated equally by the classifier.
  • the standard data set includes a plurality of slice images arranged in order, the order of the slices corresponds to the Z-axis value of each slice, the lower the Z-axis value is, the lower the Z-axis value Bigger.
  • N is 2 in the embodiment of the present application.
  • the S2 includes:
  • S z is the maximum coordinate of the Z axis of the slice in the standard data set
  • z top is the z-axis coordinate of the front slice of three adjacent slices
  • z mid is the z-axis coordinate of the middle slice of three adjacent slices
  • z bot is the z-axis coordinate of the back slice of three adjacent slices.
  • the traditional segmentation model will input a single slice or zoom first and then input the whole.
  • the front and back two slices are also input.
  • the positions of the front and back two slices are based on the layer thickness. Dynamically adjusted, by inputting three slices, the accuracy of lesion segmentation can be improved.
  • the lesion segmentation model is a convolutional neural network model, which can be used for classification, detection and segmentation.
  • the lesion segmentation model includes a down-sampling layer for extracting abstract features to expand the receptive field, an up-sampling layer for restoring detailed information, and a skip connection layer for supplementing detailed information with low-level features as high-level features.
  • the S3 includes:
  • the lesion segmentation model is used to determine the category of each pixel in the output image when outputting, to determine whether it is a lesion, and the output lesion segmentation result map can be used to detect the corresponding CT slice data The lesion area is identified, and the lesion area is segmented.
  • the lesion segmentation model in the embodiment of the application adopts the switchable atrous convolution (SAC, Switchable Atrous Convolution), which can convolve the input image with different void rates, and use the switch function to merge the convolution
  • SAC switchable Atrous Convolution
  • the lesion segmentation model can perform different convolutions on lesions of different sizes, identify more lesion areas, output a lesion segmentation map, and improve the accuracy of the lesion segmentation result.
  • this embodiment of the present application further includes training the lesion segmentation model, which is specifically as follows:
  • the loss function includes:
  • DiceLoss is the loss function value
  • Y_true represents the true lesion label
  • Y_pred represents the lesion label predicted by the convolutional neural network model
  • N is the total number of samples in the training set.
  • the embodiment of the present application uses the model obtained after training as the lesion segmentation model, and the pneumonia lesion area can be predicted through the lesion segmentation model.
  • the embodiment of the application also performs density analysis on the detected lesion area, and the nature of the corresponding lesion area can be determined through the density analysis.
  • the embodiment of the application uses the CT value in the CT image to perform density analysis on each lesion area.
  • the CT value is a measurement unit for determining the density of a certain local tissue or organ of the human body. It is usually called Hounsfield Unit (HU) for air as -1000 and dense bone as +1000.
  • HU Hounsfield Unit
  • CT value can be used to understand CT The component attributes of a certain part of the image.
  • the performing density analysis on the lesion area includes:
  • the analysis result of the lesion area is determined according to the histogram.
  • the analysis result in the embodiment of the present application may be the nature of the lesion area, including pure ground glass lesions, solid lesions, and half-ground glass half-solid lesions.
  • the determining the analysis result of the lesion area according to the histogram includes:
  • the analysis result of the lesion is determined according to the ratio threshold condition.
  • the embodiment of the present application determines the ratio threshold condition according to the following calculation formula:
  • F(hu) represents the total number of Heinz unit values that meet the requirements in the histogram, for example, F(hu>-50) is the total number of Heinz unit values greater than -50 in the histogram.
  • the determination of the analysis result of the lesion according to the ratio threshold condition in the embodiment of the present application includes:
  • pGGO is a pure ground glass lesion
  • solid is a solid lesion
  • mGGO is a semi-ground glass and semi-solid lesion.
  • the embodiment of the application determines the nature of the lesion area according to the lesion density, and adds qualitative analysis on the basis of the quantitative analysis of the lesion in the traditional lesion detection analysis model, so as to realize the multi-faceted analysis of the lesion and make the result of the lesion detection analysis. More precise and comprehensive.
  • the embodiment of the present application extracts a Heinz unit histogram and uses a threshold method to determine the nature of each lesion area, so that in the follow-up follow-up, even if the lesion area is the same, it can be evaluated by the nature of the lesion and the calculated quantitative value.
  • follow-up treatment plans are formulated based on the results, so as to help patients get better treatment.
  • the embodiment of the application obtains CT slice data and normalizes the CT slice data to obtain a standard data set, which can reduce the interference of extreme data and ensure the accuracy of the result; select a slice in the standard data set Data, calculate the slice step length according to the standard data set, and select a total of N slices before and after the slice in the standard data set according to the slice step length to obtain an input slice set, and input multiple slice data instead of a single slice
  • the slice data can be adapted to more CT data of different layer thicknesses and improve the accuracy of the model;
  • the lesion segmentation model also adopts a switchable cavity convolution mechanism, which can identify more lesion areas of different sizes and improve The accuracy of the result of lesion segmentation; meanwhile, density analysis of the lesion area is also performed, so that the analysis of the lesion area is more comprehensive. Therefore, the lesion detection and analysis method, device, and computer-readable storage medium proposed in the present application can improve the accuracy of lesion segmentation and perform multi-faceted analysis of the lesion.
  • FIG. 4 it is a schematic diagram of the modules of the lesion detection and analysis device of the present application.
  • the lesion detection and analysis apparatus 100 described in this application can be installed in an electronic device.
  • the lesion detection and analysis device may include a data processing module 101, an input slice set acquisition module 102, a lesion area determination module 103, and a density analysis module 104.
  • the module described in this application can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.
  • each module/unit is as follows:
  • the data processing module 101 is configured to obtain CT slice data, and perform a normalization operation on the CT slice data to obtain a standard data set;
  • the CT slice data in the embodiments of the present application includes a plurality of sequentially arranged medical image pictures, such as a lung CT image of a patient with pneumonia, which can be obtained from a database of a medical device.
  • the CT slice data can be obtained from a node of a blockchain.
  • the security of the CT slice data can be improved.
  • the data processing module when performing a normalization operation on the CT slice data to obtain a standard data set, the data processing module specifically performs the following operations:
  • MN (x, y, z) is normalized data
  • MO (x, y, z) is the three-dimensional matrix
  • the above-mentioned formula is used to normalize the three-dimensional matrix, which is to calculate each element in the three-dimensional matrix with the formula to obtain the calculation result; if the calculation If the result is greater than 1, the value of the element is assigned as 1; if the calculation result is less than 0, the value of the element is assigned as 0.
  • the normalization of the feature data can ensure that each Each feature is treated equally by the classifier.
  • the input slice set acquisition module 102 is configured to select a piece of slice data in the standard data set, calculate the slice step size according to the standard data set, and select the slice step size in the standard data set according to the slice step size. There are a total of N slices before and after slicing, and the input slice set is obtained.
  • the standard data set includes a plurality of slice images arranged in order, the order of the slices corresponds to the Z-axis value of each slice, the lower the Z-axis value is, the lower the Z-axis value Bigger.
  • N is 2 in the embodiment of the present application.
  • the input slice set acquisition module 102 is specifically configured to:
  • S z is the maximum coordinate of the Z axis in the standard data set
  • z top is the z-axis coordinate of the front slice of three adjacent slices
  • z mid is the z-axis coordinate of the middle slice of three adjacent slices
  • z bot is the z-axis coordinate of the back slice of three adjacent slices.
  • the traditional segmentation model will input a single slice or zoom first and then input the whole.
  • the front and back two slices are also input.
  • the positions of the front and back two slices are based on the layer thickness. Dynamically adjusted, by inputting three slices, the accuracy of lesion segmentation can be improved.
  • the lesion area determination module 103 is configured to input the input slice set into a pre-built lesion segmentation model, and use the lesion segmentation model to perform hole convolution on the input slice set to extract features of the input slice set Data, the lesion area is determined according to the characteristic data.
  • the lesion segmentation model is a convolutional neural network model, which can be used for classification, detection and segmentation.
  • the lesion segmentation model includes a down-sampling layer for extracting abstract features to expand the receptive field, an up-sampling layer for restoring detailed information, and a skip connection layer for supplementing detailed information with low-level features as high-level features.
  • the lesion area determining module 103 is specifically configured to:
  • the skip connection layer and the down-sampling layer in the lesion segmentation model are used to fuse the feature maps of the multiple scales, and the lesion segmentation result map is output, and the lesion area is determined according to the lesion segmentation result map.
  • the lesion segmentation model is used to determine the category of each pixel in the output image when outputting, to determine whether it is a lesion, and the output lesion segmentation result map can be used to detect the corresponding CT slice data The lesion area is identified, and the lesion area is segmented.
  • the lesion segmentation model in this embodiment of the application adopts a switchable atrous convolution (SAC, Switchable Atrous Convolution), which can convolve the input image with different void rates, and use the switch function to merge the convolution
  • SAC Switchable Atrous Convolution
  • the lesion segmentation model can perform different convolutions on lesions of different sizes, identify more lesion areas, output a lesion segmentation map, and improve the accuracy of the lesion segmentation result.
  • this embodiment of the present application further includes training the lesion segmentation model, which is specifically as follows:
  • the loss function includes:
  • DiceLoss is the loss function value
  • Y_true represents the true lesion label
  • Y_pred represents the lesion label predicted by the convolutional neural network model
  • N is the total number of samples in the training set.
  • the embodiment of the present application uses the model obtained after training as the lesion segmentation model, and the pneumonia lesion area can be predicted through the lesion segmentation model.
  • the density analysis module 104 is configured to perform density analysis on the lesion area, and feed back the result of the density analysis to the user.
  • the embodiment of the application also performs density analysis on the detected lesion area, and the nature of the corresponding lesion area can be determined through the density analysis.
  • the embodiment of the application uses the CT value in the CT image to perform density analysis on each lesion area.
  • the CT value is a measurement unit for determining the density of a certain local tissue or organ of the human body. It is usually called Hounsfield Unit (HU) for air as -1000 and dense bone as +1000.
  • HU Hounsfield Unit
  • CT value can be used to understand CT The component attributes of a certain part of the image.
  • the density analysis module when performing density analysis on the lesion area, specifically performs the following operations:
  • the analysis result of the lesion area is determined according to the histogram.
  • the analysis result in the embodiment of the present application may be the nature of the lesion area, including pure ground glass lesions, solid lesions, and half-ground glass half-solid lesions.
  • the determining the analysis result of the lesion area according to the histogram includes:
  • the analysis result of the lesion is determined according to the ratio threshold condition.
  • the embodiment of the present application determines the ratio threshold condition according to the following calculation formula:
  • F(hu) represents the total number of Heinz unit values that meet the requirements in the histogram, for example, F(hu>-50) is the total number of Heinz unit values greater than -50 in the histogram.
  • the determination of the analysis result of the lesion according to the ratio threshold condition in the embodiment of the present application includes:
  • pGGO is a pure ground glass lesion
  • solid is a solid lesion
  • mGGO is a semi-ground glass and semi-solid lesion.
  • the embodiment of the application determines the nature of the lesion area according to the lesion density, and adds qualitative analysis on the basis of the quantitative analysis of the lesion in the traditional lesion detection analysis model, so as to realize the multi-faceted analysis of the lesion and make the result of the lesion detection analysis. More precise and comprehensive.
  • the embodiment of the present application extracts a Heinz unit histogram and uses a threshold method to determine the nature of each lesion area, so that in the follow-up follow-up, even if the lesion area is the same, it can be evaluated by the nature of the lesion and the calculated quantitative value.
  • follow-up treatment plans are formulated based on the results, so as to help patients get better treatment.
  • FIG. 5 it is a schematic diagram of the structure of an electronic device that implements the method for detecting and analyzing a lesion in 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 running on the processor 10, such as a lesion detection and analysis program 12.
  • the memory 11 may be volatile or non-volatile.
  • the memory 11 includes at least one type of readable storage medium.
  • the readable storage medium includes flash memory, mobile hard disk, and multimedia card.
  • Card-type memory for example: SD or DX memory, etc.
  • magnetic memory magnetic disk, optical disk, etc.
  • the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, for example, 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 plug-in mobile hard disk, a smart media card (SMC), and a secure digital (Secure Digital) equipped on the electronic device 1.
  • SD Secure Digital
  • flash Card Flash Card
  • 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 be used not only to store application software and various data installed in the electronic device 1, such as the code of the lesion detection and analysis program 12, etc., but also to temporarily store data that has been output or will be output.
  • the processor 10 may be composed of integrated circuits in some embodiments, for example, may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits with the same function or different functions, including one or more Combinations of central processing unit (CPU), microprocessor, digital processing chip, graphics processor, and various control chips, etc.
  • the processor 10 is the control unit of the electronic device, which uses various interfaces and lines to connect the various components of the entire electronic device, and runs or executes programs or modules stored in the memory 11 (such as executing Lesion detection and analysis program, etc.), and call the data stored in the memory 11 to execute various functions of the electronic device 1 and process data.
  • the bus may be a peripheral component interconnect standard (PCI) bus or an extended industry standard architecture (EISA) bus, etc.
  • PCI peripheral component interconnect standard
  • 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 and 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 shown in the figure. Components, or a combination of certain components, or different component arrangements.
  • the electronic device 1 may also include a power source (such as a battery) for supplying power to various components.
  • the power source may be logically connected to the at least one processor 10 through a power management device, thereby controlling power
  • the device implements functions such as charge management, discharge management, and power consumption management.
  • the power supply may also include any components such as one or more DC or AC power supplies, recharging devices, power failure detection circuits, power converters or inverters, and power status indicators.
  • the electronic device 1 may also 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.
  • 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 also include a user interface.
  • the user interface may be a display (Display) and an input unit (such as a keyboard (Keyboard)).
  • 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, etc.
  • the display can also be appropriately called a display screen or a display unit, which is used to display the information processed in the electronic device 1 and to display a visualized user interface.
  • the lesion detection and analysis program 12 stored in the memory 11 in the electronic device 1 is a combination of multiple computer programs. When running in the processor 10, it can realize:
  • the integrated module/unit of the electronic device 1 can be stored in a computer-readable storage medium. It can be volatile or non-volatile.
  • the computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory) .
  • the computer usable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function, etc.; the storage data area may store a block chain node Use the created data, etc.
  • modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional modules in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional modules.
  • the blockchain referred to in this application is a new application mode of computer technology 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 for verification. The validity of the information (anti-counterfeiting) and the 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)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

一种病灶检测分析方法,涉及人工智能技术,包括:获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;选择所述标准数据集中的一张切片数据,根据所述标准数据集计算切片步长,并在所述标准数据集中选择所述切片数据前后共N张切片数据,得到输入切片集;将所述输入切片集输入至预先构建的病灶分割模型进行空洞卷积,提取特征数据,并根据所述特征数据确定病灶区域;对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。还提出了一种病灶检测分析装置、设备及计算机可读存储介质。此外,所述CT切片数据可存储于区块链节点中。该方法可以提高病灶分割的精度,并对病灶进行多方面的分析。

Description

病灶检测分析方法、装置、电子设备及计算机存储介质
本申请要求于2020年9月22日提交中国专利局、申请号为CN202011000056.9、名称为“病灶检测分析方法、装置、电子设备及计算机存储介质”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
本申请涉及人工智能技术领域,尤其涉及一种病灶检测分析方法、装置、电子设备及计算机可读存储介质。
背景技术
随着深度学习在图像领域取得的巨大发展,基于深度学习方法在医疗影像数据上也被广泛应用。病灶检测分析和分割是其中的一个重要分支,利用计算机自动化识别影像中的病灶区域,确定患者的严重程度,便于后续医生进行精细诊断
目前已有的病灶检测分析方法是基于深度学习算法的,发明人意识到,对病灶的分割精度较低,并且只对病灶进行区域大小的检测,并没有对各病灶区域进行进一步分析,使对于病灶区域体积的定量指标未有明显改变的治愈情况难以检测出来。
发明内容
本申请提供的一种病灶检测分析方法,包括:
获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
本申请还提供一种病灶检测分析装置,所述装置包括:
数据处理模块,用于获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
输入切片集获取模块,用于选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
病灶区域确定模块,用于将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
密度分析模块,用于对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
为了解决上述问题,本申请还提供一种电子设备,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计算机程序,所述计算机程序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CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
为了解决上述问题,本申请还提供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CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
附图说明
图1为本申请实施例提供的病灶检测分析方法的流程示意图;
图2为图1中其中一个步骤的详细实施流程示意图;
图3为图1中另外一个步骤的详细实施流程示意图;
图4为本申请实施例提供的病灶检测分析装置的模块示意图;
图5为本申请实施例提供的实现病灶检测分析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。
本申请实施例提供的病灶检测分析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,所述病灶检测分析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。
本申请提供一种病灶检测分析方法。参照图1所示,为本申请一实施例提供的病灶检测分析方法的流程示意图。该方法可以由一个装置执行,该装置可以由软件和/或硬件实现。
在本实施例中,病灶检测分析方法包括:
S1、获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集。
本申请实施例中所述CT切片数据包括多个按顺序排列的医学影像图片,如肺炎患者的肺部CT影像,可以从医疗设备的数据库中获取。
优选地,所述CT切片数据可以从一区块链的节点中获取,通过将CT切片数据存储于区块链中,可以提高CT切片数据的安全性。
详细地,所述将所述CT切片数据进行归一化操作,得到标准数据集,包括:
根据所述CT切片数据中各切片的像素值构建对应的三维矩阵;
利用下述公式对所述三维矩阵进行归一化处理,得到归一化后的标准数据集:
Figure PCTCN2020131989-appb-000001
其中,MN(x,y,z)为归一化的数据,MO(x,y,z)为所述三维矩阵。
具体地,本申请实施例中利用上述公式对所述三维矩阵进行归一化处理,是对所述三维矩阵中的每个元素,都用所述公式进行计算,得到计算结果;若所述计算结果大于1,则将所述元素的值赋值为1;若所述计算结果小于0,则将所述元素的值赋值为0。
优选地,当不同的特征成列在一起的时候,由于特征本身表达方式的原因会导致在绝对数值上的小数据被大数据覆盖的情况,因此对特征数据进行归一化处理,可以保证每个特征被分类器平等对待。
S2、选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集。
优选地,所述标准数据集包含多个按顺序排列的切片图像,所述切片的顺序与各切片的Z轴值对应,排在前的Z轴值越小,排在后的则Z轴值越大。
优选地,本申请实施例中N为2。
详细地,参照图2所示,所述S2包括:
S20、在所述标准数据集中选择一张切片数据,并采用下述公式计算切片步长stride:
Figure PCTCN2020131989-appb-000002
其中,S z为所述标准数据集中切片的Z轴最大坐标;
S21、根据所述切片步长计算与所述切片前后共N张切片在所述标准数据集中的位置,根据所述位置采用下述公式选择对应的N张切片:
z top=max(1,z mid-stride);
z bot=min(S z,z mid+stride);
其中,z top为相邻三张切片的前切片的z轴坐标、z mid为相邻三张切片的中间切片的z轴坐标,z bot为相邻三张切片的后切片的z轴坐标。
优选地,传统的分割模型会输入单张切片或者先缩放再整体输入,本申请实施例在输入单张切片的同时,还输入前后两张切片,其中,前后两张切片的位置是根据层厚动态调整的,通过输入三张切片,可以提高病灶分割的精度。
S3、将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,根据所述特征数据确定病灶区域。
优选地,所述病灶分割模型是一种卷积神经网络模型,可以用于分类、检测和分割。所述病灶分割模型包括用于抽取抽象特征以扩大感受野的下采样层、用于恢复细节信息的上采样层和用于使用低层特征为高层特征补充细节信息的跳跃连接层。
详细地,参照图3所示,所述S3包括:
S30、利用所述病灶分割模型中的上采样层对所述输入切片集进行多次不同空洞率的空洞卷积、池化操作,得到多个尺度的特征图;
S31、利用所述病灶分割模型中的跳跃连接层和下采样层将所述多个尺度的特征图进行融合,并输出病灶分割结果图,并根据所述病灶分割结果图确定病灶区域。
本申请实施例中利用所述病灶分割模型在输出时会对输出图片中各个像素点进行类别判断,判断其是否为病灶,通过输出的病灶分割结果图可以检测出对应所述CT切片数据中的病灶区域,并对所述病灶区域进行标识,分割出所述病灶区域。
较佳地,本申请实施例中所述病灶分割模型采用了可切换的空洞卷积(SAC,Switchable  Atrous Convolution),可以以不同的空洞率对输入图片进行卷积,并使用switch函数合并卷积后的结果,通过上述卷积操作,使得所述病灶分割模型能够对不同大小区域的病灶进行不同的卷积,识别出更多病灶区域,输出病灶分割图,提高病灶分割结果的准确度。
可选地,本申请实施例在将所述输入切片集输入至预先构建的病灶分割模型之前,还包括对所述病灶分模型进行训练,具体如下:
获取训练集,并将所述训练集输入至所述病灶分割模型中,得到预测结果;
利用预设的损失函数对所述预测结果进行损失计算,得到损失函数值;
根据所述损失函数值反向传播更新所述病灶分割模型的参数;
返回上述的损失计算步骤,直到达到预设迭代次数,得到训练完成的病灶分割模型。
进一步地,所述损失函数包括:
Figure PCTCN2020131989-appb-000003
其中,DiceLoss为损失函数值,Y_true表示真实的病灶标签,Y_pred则表示所述卷积神经网络模型预测出的病灶标签,N是所述训练集的样本总数。
优选地,本申请实施例将经过训练后得到的模型作为病灶分割模型,通过所述病灶分割模型,即可预测肺炎病灶区域。
S4、对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
本申请实施例对检测出来的病灶区域还进行密度分析,通过密度分析可以确定对应病灶区域的性质,本申请实施例利用CT影像中的CT值来对各病灶区域进行密度分析。其中,所述CT值是测定人体某一局部组织或器官密度大小的一种计量单位,通常称亨氏单位(hounsfield unit,HU)空气为-1000,致密骨为+1000,通过CT值可以了解CT影像中某个部位的成分属性。
详细地,所述对所述病灶区域进行密度分析,包括:
对每个所述病灶区域,从所述CT切片数据中提取与所述病灶区域对应位置的多个亨氏单位值;
对所述多个亨氏单位值进行统计,生成关于亨氏单位值数量的直方图;
根据所述直方图确定所述病灶区域的分析结果。
优选地,本申请实施例中所述分析结果可以是所述病灶区域的性质,包括纯磨玻璃病灶、实性病灶和半磨玻璃半实性病灶。
进一步地,所述根据所述直方图确定所述病灶区域的分析结果,包括:
基于所述直方图确定比例阈值条件;
根据所述比例阈值条件确定所述病灶的分析结果。
优选地,本申请实施例按照下述计算公式确定比例阈值条件:
Figure PCTCN2020131989-appb-000004
Figure PCTCN2020131989-appb-000005
其中,F(hu)表示所述直方图中符合要求的亨氏单位值的总数量,如F(hu>-50)为所述直方图中亨氏单位值大于-50的总数量。
进一步地,本申请实施例所述根据所述比例阈值条件确定所述病灶的分析结果,包括:
Figure PCTCN2020131989-appb-000006
其中,pGGO为纯磨玻璃病灶,solid为实性病灶,mGGO为半磨玻璃半实性病灶。
本申请实施例根据所述病灶密度确定所述病灶区域的性质,在传统的病灶检测分析模 型对病灶进行量化分析的基础上增加定性分析,以实现对病灶多方面的分析,使病灶检测分析结果更加精确、全面。
优选地,本申请实施例通过提取亨氏单位直方图,采用阈值方式确定各病灶区域的性质,让后续的随访中,即使病灶区域相同,也能够通过病灶的性质与计算得出的定量数值来评估后期疗效,并依据结果制定后续治疗方案,从而帮助患者得到更好的治疗。
本申请实施例通过获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集,可以减少极端数据的干扰,确保结果的准确性;选择所述标准数据集中的一张切片数据,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集,输入多张切片数据而不是单张切片数据,可以适配更多不同层厚的CT数据,提高模型的精确度;所述病灶分割模型中还采用了可切换的空洞卷积机制,可以识别出更多不同大小的病灶区域,提高病灶分割结果的准确度;同时还对所述病灶区域进行密度分析,对病灶区域的分析更加全面。因此本申请提出的病灶检测分析方法、装置及计算机可读存储介质,可以提高病灶分割的精度,并对病灶进行多方面的分析。
如图4所示,是本申请病灶检测分析装置的模块示意图。
本申请所述病灶检测分析装置100可以安装于电子设备中。根据实现的功能,所述病灶检测分析装置可以包括数据处理模块101、输入切片集获取模块102、病灶区域确定模块103和密度分析模块104。本申请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:
所述数据处理模块101,用于获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
本申请实施例中所述CT切片数据包括多个按顺序排列的医学影像图片,如肺炎患者的肺部CT影像,可以从医疗设备的数据库中获取。
优选地,所述CT切片数据可以从一区块链的节点中获取,通过将CT切片数据存储于区块链中,可以提高CT切片数据的安全性。
详细地,在将所述CT切片数据进行归一化操作,得到标准数据集时,所述数据处理模块具体执行下述操作:
根据所述CT切片数据中各切片的像素值构建对应的三维矩阵;
利用下述公式对所述三维矩阵进行归一化处理,得到归一化后的标准数据集:
Figure PCTCN2020131989-appb-000007
其中,MN(x,y,z)为归一化的数据,MO(x,y,z)为所述三维矩阵。
具体地,本申请实施例中利用上述公式对所述三维矩阵进行归一化处理,是对所述三维矩阵中的每个元素,都用所述公式进行计算,得到计算结果;若所述计算结果大于1,则将所述元素的值赋值为1;若所述计算结果小于0,则将所述元素的值赋值为0。
优选地,当不同的特征成列在一起的时候,由于特征本身表达方式的原因会导致在绝对数值上的小数据被大数据覆盖的情况,因此对特征数据进行归一化处理,可以保证每个特征被分类器平等对待。
所述输入切片集获取模块102,用于选择所述标准数据集中的一张切片数据,根据所 述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集。
优选地,所述标准数据集包含多个按顺序排列的切片图像,所述切片的顺序与各切片的Z轴值对应,排在前的Z轴值越小,排在后的则Z轴值越大。
优选地,本申请实施例中N为2。
详细地,所述输入切片集获取模块102具体用于:
在所述标准数据集中选择一张切片数据,并采用下述公式计算切片步长stride:
Figure PCTCN2020131989-appb-000008
其中,S z为所述标准数据集中的Z轴最大坐标;
根据所述切片步长计算与所述切片前后共N张切片在所述标准数据集中的位置,根据所述位置采用下述公式选择对应的N张切片:
z top=max(1,z mid-stride);
z bot=min(S z,z mid+stride);
其中,z top为相邻三张切片的前切片的z轴坐标、z mid为相邻三张切片的中间切片的z轴坐标,z bot为相邻三张切片的后切片的z轴坐标。
优选地,传统的分割模型会输入单张切片或者先缩放再整体输入,本申请实施例在输入单张切片的同时,还输入前后两张切片,其中,前后两张切片的位置是根据层厚动态调整的,通过输入三张切片,可以提高病灶分割的精度。
所述病灶区域确定模块103,用于将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,根据所述特征数据确定病灶区域。
优选地,所述病灶分割模型是一种卷积神经网络模型,可以用于分类、检测和分割。所述病灶分割模型包括用于抽取抽象特征以扩大感受野的下采样层、用于恢复细节信息的上采样层和用于使用低层特征为高层特征补充细节信息的跳跃连接层。
详细地,所述病灶区域确定模块103具体用于:
利用所述病灶分割模型中的上采样层对所述输入切片集进行多次不同空洞率的空洞卷积、池化操作,得到多个尺度的特征图;
利用所述病灶分割模型中的跳跃连接层和下采样层将所述多个尺度的特征图进行融合,并输出病灶分割结果图,并根据所述病灶分割结果图确定病灶区域。
本申请实施例中利用所述病灶分割模型在输出时会对输出图片中各个像素点进行类别判断,判断其是否为病灶,通过输出的病灶分割结果图可以检测出对应所述CT切片数据中的病灶区域,并对所述病灶区域进行标识,分割出所述病灶区域。
较佳地,本申请实施例中所述病灶分割模型采用了可切换的空洞卷积(SAC,Switchable Atrous Convolution),可以以不同的空洞率对输入图片进行卷积,并使用switch函数合并卷积后的结果,通过上述卷积操作,使得所述病灶分割模型能够对不同大小区域的病灶进行不同的卷积,识别出更多病灶区域,输出病灶分割图,提高病灶分割结果的准确度。
可选地,本申请实施例在将所述输入切片集输入至预先构建的病灶分割模型之前,还包括对所述病灶分模型进行训练,具体如下:
获取训练集,并将所述训练集输入至所述病灶分割模型中,得到预测结果;
利用预设的损失函数对所述预测结果进行损失计算,得到损失函数值;
根据所述损失函数值反向传播更新所述病灶分割模型的参数;
返回上述的损失计算步骤,直到达到预设迭代次数,得到训练完成的病灶分割模型。
进一步地,所述损失函数包括:
Figure PCTCN2020131989-appb-000009
其中,DiceLoss为损失函数值,Y_true表示真实的病灶标签,Y_pred则表示所述卷积神经网络模型预测出的病灶标签,N是所述训练集的样本总数。
优选地,本申请实施例将经过训练后得到的模型作为病灶分割模型,通过所述病灶分割模型,即可预测肺炎病灶区域。
所述密度分析模块104,用于对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
本申请实施例对检测出来的病灶区域还进行密度分析,通过密度分析可以确定对应病灶区域的性质,本申请实施例利用CT影像中的CT值来对各病灶区域进行密度分析。其中,所述CT值是测定人体某一局部组织或器官密度大小的一种计量单位,通常称亨氏单位(hounsfield unit,HU)空气为-1000,致密骨为+1000,通过CT值可以了解CT影像中某个部位的成分属性。
详细地,在对所述病灶区域进行密度分析时,所述密度分析模块具体执行下述操作:
对每个所述病灶区域,从所述CT切片数据中提取与所述病灶区域对应位置的多个亨氏单位值;
对所述多个亨氏单位值进行统计,生成关于亨氏单位值数量的直方图;
根据所述直方图确定所述病灶区域的分析结果。
优选地,本申请实施例中所述分析结果可以是所述病灶区域的性质,包括纯磨玻璃病灶、实性病灶和半磨玻璃半实性病灶。
进一步地,所述根据所述直方图确定所述病灶区域的分析结果,包括:
基于所述直方图确定比例阈值条件;
根据所述比例阈值条件确定所述病灶的分析结果。
优选地,本申请实施例按照下述计算公式确定比例阈值条件:
Figure PCTCN2020131989-appb-000010
Figure PCTCN2020131989-appb-000011
其中,F(hu)表示所述直方图中符合要求的亨氏单位值的总数量,如F(hu>-50)为所述直方图中亨氏单位值大于-50的总数量。
进一步地,本申请实施例所述根据所述比例阈值条件确定所述病灶的分析结果,包括:
Figure PCTCN2020131989-appb-000012
其中,pGGO为纯磨玻璃病灶,solid为实性病灶,mGGO为半磨玻璃半实性病灶。
本申请实施例根据所述病灶密度确定所述病灶区域的性质,在传统的病灶检测分析模型对病灶进行量化分析的基础上增加定性分析,以实现对病灶多方面的分析,使病灶检测分析结果更加精确、全面。
优选地,本申请实施例通过提取亨氏单位直方图,采用阈值方式确定各病灶区域的性质,让后续的随访中,即使病灶区域相同,也能够通过病灶的性质与计算得出的定量数值来评估后期疗效,并依据结果制定后续治疗方案,从而帮助患者得到更好的治疗。
如图5所示,是本申请实现病灶检测分析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,如病灶检测分析程序12。
其中,所述存储器11可以是易失性的,也可以是非易失性的,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,例如病灶检测分析程序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(例如执行病灶检测分析程序等),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存储的病灶检测分析程序12是多个计算机程序的组合,在所述处理器10中运行时,可以实现:
获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
选择所述标准数据集中的一张切片数据,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
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,所述计算机可读存储介质可以是易失性的,也可以是非易失性的,。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,所述计算机可用存储介质可主要包括存储程序区和存储数据区,其中,存储程序区可存储操作系统、至少一个功能所需的应用程序等;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。
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。
此外,显然“包括”一词不排除其他单元或步骤,单数不排除复数。系统权利要求中陈述的多个单元或装置也可以由一个单元或装置通过软件或者硬件来实现。第二等词语用来表示名称,而并不表示任何特定的顺序。
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. 一种病灶检测分析方法,其中,所述方法包括:
    获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
    选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
    将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
    对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
  2. 如权利要求1所述的病灶检测分析方法,其中,所述选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集,包括:
    在所述标准数据集中选择其中一张切片,并采用下述公式计算切片步长stride:
    Figure PCTCN2020131989-appb-100001
    其中,S z为所述标准数据集中的Z轴最大坐标;
    根据所述切片步长计算与所述切片前后共N张切片在所述标准数据集中的位置,根据所述位置采用下述公式选择对应的N张切片:
    z top=max(1,z mid-stride);
    z bot=min(S z,z mid+stride);
    其中,z top为相邻三张切片的前切片的z轴坐标、z mid为相邻三张切片的中间切片的z轴坐标,z bot为相邻三张切片的后切片的z轴坐标。
  3. 如权利要求1所述的病灶检测分析方法,其中,所述利用所述病灶分割模型不同的空洞参数对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域,包括:
    利用所述病灶分割模型中的上采样层对所述输入切片集进行多次不同空洞率的空洞卷积、池化操作,得到多个尺度的特征图;
    利用所述病灶分割模型中的跳跃连接层和下采样层将所述多个尺度的特征图进行融合,并输出病灶分割结果图,并根据所述病灶分割结果图确定病灶区域。
  4. 如权利要求3所述的病灶检测分析方法,其中,在将所述输入切片集输入至预先构建的病灶分割模型之前,该方法还包括:
    获取训练集,并将所述训练集输入至所述病灶分割模型中,得到预测结果;
    利用预设的损失函数对所述预测结果进行损失计算,得到损失函数值;
    根据所述损失函数值反向传播更新所述病灶分割模型的参数;
    返回上述的损失计算步骤,直到达到预设迭代次数,得到训练完成的病灶分割模型。
  5. 如权利要求1所述的病灶检测分析方法,其中,所述对所述病灶区域进行密度分析,包括:
    对每个所述病灶区域,从所述CT切片数据中提取与所述病灶区域对应位置的多个亨氏单位值;
    对所述多个亨氏单位值进行统计,生成关于亨氏单位值数量的直方图;
    根据所述直方图确定所述病灶区域的分析结果。
  6. 如权利要求5所述的病灶检测分析方法,其中,所述根据所述直方图确定所述病灶区域的分析结果,包括:
    基于所述直方图确定比例阈值条件;
    根据所述比例阈值条件确定所述病灶的分析结果。
  7. 如权利要求1所述的病灶检测分析方法,其中,所述将所述CT切片数据进行归一化操作,得到标准数据集,包括:
    根据所述CT切片数据中各切片的像素值构建对应的三维矩阵;
    利用下述公式对所述三维矩阵进行归一化处理,得到归一化后的标准数据集:
    Figure PCTCN2020131989-appb-100002
    其中,MN(x,y,z)为归一化的数据,MO(x,y,z)为所述三维矩阵。
  8. 一种病灶检测分析装置,其中,所述装置包括:
    数据处理模块,用于获取CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
    输入切片集获取模块,用于选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
    病灶区域确定模块,用于将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
    密度分析模块,用于对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
  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计算机程序,所述计算机程序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CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
    选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
    将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
    对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
  10. 如权利要求9所述的电子设备,其中,所述选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集,包括:
    在所述标准数据集中选择其中一张切片,并采用下述公式计算切片步长stride:
    Figure PCTCN2020131989-appb-100003
    其中,S z为所述标准数据集中的Z轴最大坐标;
    根据所述切片步长计算与所述切片前后共N张切片在所述标准数据集中的位置,根据所述位置采用下述公式选择对应的N张切片:
    z top=max(1,z mid-stride);
    z bot=min(S z,z mid+stride);
    其中,z top为相邻三张切片的前切片的z轴坐标、z mid为相邻三张切片的中间切片的z 轴坐标,z bot为相邻三张切片的后切片的z轴坐标。
  11. 如权利要求9所述的电子设备,其中,所述利用所述病灶分割模型不同的空洞参数对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域,包括:
    利用所述病灶分割模型中的上采样层对所述输入切片集进行多次不同空洞率的空洞卷积、池化操作,得到多个尺度的特征图;
    利用所述病灶分割模型中的跳跃连接层和下采样层将所述多个尺度的特征图进行融合,并输出病灶分割结果图,并根据所述病灶分割结果图确定病灶区域。
  12. 如权利要求11所述的电子设备,其中,在将所述输入切片集输入至预先构建的病灶分割模型之前,还实现以下步骤:
    获取训练集,并将所述训练集输入至所述病灶分割模型中,得到预测结果;
    利用预设的损失函数对所述预测结果进行损失计算,得到损失函数值;
    根据所述损失函数值反向传播更新所述病灶分割模型的参数;
    返回上述的损失计算步骤,直到达到预设迭代次数,得到训练完成的病灶分割模型。
  13. 如权利要求9所述的电子设备,其中,所述对所述病灶区域进行密度分析,包括:
    对每个所述病灶区域,从所述CT切片数据中提取与所述病灶区域对应位置的多个亨氏单位值;
    对所述多个亨氏单位值进行统计,生成关于亨氏单位值数量的直方图;
    根据所述直方图确定所述病灶区域的分析结果。
  14. 如权利要求13所述的电子设备,其中,所述根据所述直方图确定所述病灶区域的分析结果,包括:
    基于所述直方图确定比例阈值条件;
    根据所述比例阈值条件确定所述病灶的分析结果。
  15. 如权利要求9所述的电子设备,其中,所述将所述CT切片数据进行归一化操作,得到标准数据集,包括:
    根据所述CT切片数据中各切片的像素值构建对应的三维矩阵;
    利用下述公式对所述三维矩阵进行归一化处理,得到归一化后的标准数据集:
    Figure PCTCN2020131989-appb-100004
    其中,MN(x,y,z)为归一化的数据,MO(x,y,z)为所述三维矩阵。
  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CT切片数据,将所述CT切片数据进行归一化操作,得到标准数据集;
    选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长,并根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集;
    将所述输入切片集输入至预先构建的病灶分割模型,利用所述病灶分割模型对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域;
    对所述病灶区域进行密度分析,将所述密度分析的结果反馈给用户。
  17. 如权利要求16所述的计算机可读存储介质,其中,所述选择所述标准数据集中的其中一张切片,根据所述标准数据集计算切片步长根据所述切片步长在所述标准数据集中选择所述切片前后共N张切片,得到输入切片集,包括:
    在所述标准数据集中选择其中一张切片,并采用下述公式计算切片步长stride:
    Figure PCTCN2020131989-appb-100005
    其中,S z为所述标准数据集中的Z轴最大坐标;
    根据所述切片步长计算与所述切片前后共N张切片在所述标准数据集中的位置,根据所述位置采用下述公式选择对应的N张切片:
    z top=max(1,z mid-stride);
    z bot=min(S z,z mid+stride);
    其中,z top为相邻三张切片的前切片的z轴坐标、z mid为相邻三张切片的中间切片的z轴坐标,z bot为相邻三张切片的后切片的z轴坐标。
  18. 如权利要求16所述的计算机可读存储介质,其中,所述利用所述病灶分割模型不同的空洞参数对所述输入切片集进行空洞卷积,提取所述输入切片集的特征数据,并根据所述特征数据确定病灶区域,包括:
    利用所述病灶分割模型中的上采样层对所述输入切片集进行多次不同空洞率的空洞卷积、池化操作,得到多个尺度的特征图;
    利用所述病灶分割模型中的跳跃连接层和下采样层将所述多个尺度的特征图进行融合,并输出病灶分割结果图,并根据所述病灶分割结果图确定病灶区域。
  19. 如权利要求18所述的计算机可读存储介质,其中,在将所述输入切片集输入至预先构建的病灶分割模型之前,还实现以下步骤:
    获取训练集,并将所述训练集输入至所述病灶分割模型中,得到预测结果;
    利用预设的损失函数对所述预测结果进行损失计算,得到损失函数值;
    根据所述损失函数值反向传播更新所述病灶分割模型的参数;
    返回上述的损失计算步骤,直到达到预设迭代次数,得到训练完成的病灶分割模型。
  20. 如权利要求16所述的计算机可读存储介质,其中,所述对所述病灶区域进行密度分析,包括:
    对每个所述病灶区域,从所述CT切片数据中提取与所述病灶区域对应位置的多个亨氏单位值;
    对所述多个亨氏单位值进行统计,生成关于亨氏单位值数量的直方图;
    根据所述直方图确定所述病灶区域的分析结果。
PCT/CN2020/131989 2020-09-22 2020-11-27 病灶检测分析方法、装置、电子设备及计算机存储介质 WO2021189909A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011000056.9A CN111915609B (zh) 2020-09-22 2020-09-22 病灶检测分析方法、装置、电子设备及计算机存储介质
CN202011000056.9 2020-09-22

Publications (1)

Publication Number Publication Date
WO2021189909A1 true WO2021189909A1 (zh) 2021-09-30

Family

ID=73265369

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/131989 WO2021189909A1 (zh) 2020-09-22 2020-11-27 病灶检测分析方法、装置、电子设备及计算机存储介质

Country Status (2)

Country Link
CN (1) CN111915609B (zh)
WO (1) WO2021189909A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114066871A (zh) * 2021-11-19 2022-02-18 江苏科技大学 一种新冠肺炎病灶区域分割模型训练的方法
CN114549552A (zh) * 2022-02-15 2022-05-27 上海翰宇生物科技有限公司 基于空间邻域分析的肺部ct图像分割装置
CN115798725A (zh) * 2022-10-27 2023-03-14 佛山读图科技有限公司 用于核医学的带病灶人体仿真影像数据制作方法
CN117942161A (zh) * 2023-12-12 2024-04-30 中国人民解放军总医院第一医学中心 单根穿刺针功率确认方法、装置、电子设备及存储介质

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111915609B (zh) * 2020-09-22 2023-07-14 平安科技(深圳)有限公司 病灶检测分析方法、装置、电子设备及计算机存储介质
CN112634224B (zh) * 2020-12-17 2023-07-28 北京大学 基于目标影像的病灶检测方法和装置
CN112465819B (zh) * 2020-12-18 2024-06-18 平安科技(深圳)有限公司 图像异常区域检测方法、装置、电子设备及存储介质
CN112734710A (zh) * 2020-12-30 2021-04-30 上海睿刀医疗科技有限公司 一种基于历史病理信息的病灶识别模型构建装置及系统
CN115393323B (zh) * 2022-08-26 2023-05-30 数坤(上海)医疗科技有限公司 一种目标区域获得方法、装置、设备和存储介质
CN115147420B (zh) * 2022-09-05 2022-11-22 北方健康医疗大数据科技有限公司 切片间相关度检测模型训练方法、检测方法和电子设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109902717A (zh) * 2019-01-23 2019-06-18 平安科技(深圳)有限公司 病灶自动识别方法、装置及计算机可读存储介质
CN110111313A (zh) * 2019-04-22 2019-08-09 腾讯科技(深圳)有限公司 基于深度学习的医学图像检测方法及相关设备
CN110710986A (zh) * 2019-10-25 2020-01-21 华院数据技术(上海)有限公司 一种基于ct图像的脑部动静脉畸形检测方法及检测系统
CN111507965A (zh) * 2020-04-17 2020-08-07 中山仰视科技有限公司 新型冠状病毒肺炎病灶检测方法、系统、装置和存储介质
CN111915609A (zh) * 2020-09-22 2020-11-10 平安科技(深圳)有限公司 病灶检测分析方法、装置、电子设备及计算机存储介质

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10303986B2 (en) * 2009-04-07 2019-05-28 Kayvan Najarian Automated measurement of brain injury indices using brain CT images, injury data, and machine learning
CN108734708B (zh) * 2018-05-23 2021-02-19 平安科技(深圳)有限公司 胃癌识别方法、装置及存储介质
CN109829880A (zh) * 2018-12-07 2019-05-31 清影医疗科技(深圳)有限公司 一种基于深度学习的ct图像检测方法、装置和控制设备
CN109767458B (zh) * 2018-12-21 2023-01-20 西北大学 一种半自动分段的顺序优化配准方法
CN109978888B (zh) * 2019-02-18 2023-07-28 平安科技(深圳)有限公司 一种图像分割方法、装置及计算机可读存储介质
CN110111344B (zh) * 2019-05-13 2021-11-16 广州锟元方青医疗科技有限公司 病理切片图像分级方法、装置、计算机设备和存储介质
CN111595850B (zh) * 2020-04-27 2022-07-15 平安科技(深圳)有限公司 切片缺陷检测方法、电子装置及可读存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109902717A (zh) * 2019-01-23 2019-06-18 平安科技(深圳)有限公司 病灶自动识别方法、装置及计算机可读存储介质
CN110111313A (zh) * 2019-04-22 2019-08-09 腾讯科技(深圳)有限公司 基于深度学习的医学图像检测方法及相关设备
CN110710986A (zh) * 2019-10-25 2020-01-21 华院数据技术(上海)有限公司 一种基于ct图像的脑部动静脉畸形检测方法及检测系统
CN111507965A (zh) * 2020-04-17 2020-08-07 中山仰视科技有限公司 新型冠状病毒肺炎病灶检测方法、系统、装置和存储介质
CN111915609A (zh) * 2020-09-22 2020-11-10 平安科技(深圳)有限公司 病灶检测分析方法、装置、电子设备及计算机存储介质

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114066871A (zh) * 2021-11-19 2022-02-18 江苏科技大学 一种新冠肺炎病灶区域分割模型训练的方法
CN114549552A (zh) * 2022-02-15 2022-05-27 上海翰宇生物科技有限公司 基于空间邻域分析的肺部ct图像分割装置
CN115798725A (zh) * 2022-10-27 2023-03-14 佛山读图科技有限公司 用于核医学的带病灶人体仿真影像数据制作方法
CN115798725B (zh) * 2022-10-27 2024-03-26 佛山读图科技有限公司 用于核医学的带病灶人体仿真影像数据制作方法
CN117942161A (zh) * 2023-12-12 2024-04-30 中国人民解放军总医院第一医学中心 单根穿刺针功率确认方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN111915609A (zh) 2020-11-10
CN111915609B (zh) 2023-07-14

Similar Documents

Publication Publication Date Title
WO2021189909A1 (zh) 病灶检测分析方法、装置、电子设备及计算机存储介质
WO2021189912A1 (zh) 图像中目标物的检测方法、装置、电子设备及存储介质
WO2022213465A1 (zh) 基于神经网络的图像识别方法、装置、电子设备及介质
WO2021217851A1 (zh) 异常细胞自动标注方法、装置、电子设备及存储介质
WO2021189913A1 (zh) 图像中目标物的分割方法、装置、电子设备及存储介质
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
Tu et al. Automatic categorization and scoring of solid, part-solid and non-solid pulmonary nodules in CT images with convolutional neural network
JP7082993B2 (ja) 医療画像処理装置、方法及びプログラム、診断支援装置、方法及びプログラム並びに医療支援システム及び方法
WO2022126903A1 (zh) 图像异常区域检测方法、装置、电子设备及存储介质
US20210125724A1 (en) Medical image processing apparatus, medical image processing method, machine learning system, and program
CN111933274B (zh) 疾病分类诊断方法、装置、电子设备及存储介质
Warin et al. Maxillofacial fracture detection and classification in computed tomography images using convolutional neural network-based models
WO2022247007A1 (zh) 医学图像分级方法、装置、电子设备及可读存储介质
CN111862096A (zh) 图像分割方法、装置、电子设备及存储介质
US12106856B2 (en) Image processing apparatus, image processing method, and program for segmentation correction of medical image
WO2022227192A1 (zh) 图像分类方法、装置、电子设备及介质
WO2021189900A1 (zh) 医学图像分析方法、装置、电子设备及可读存储介质
WO2021189914A1 (zh) 电子设备、医学图像指标生成方法、装置及存储介质
CN113724185B (zh) 用于图像分类的模型处理方法、装置及存储介质
CN111681247A (zh) 肺叶肺段分割模型训练方法和装置
US20230005601A1 (en) Document creation support apparatus, method, and program
CN115760656A (zh) 一种医学影像处理方法及系统
JP2020032044A (ja) 類似度決定装置、方法およびプログラム
Tang et al. Lesion segmentation and RECIST diameter prediction via click-driven attention and dual-path connection
JP7504987B2 (ja) 情報処理装置、情報処理方法及び情報処理プログラム

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: 20926794

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: 20926794

Country of ref document: EP

Kind code of ref document: A1