WO2021120961A1 - Procédé et appareil d'évaluation de carte de structure de dépendance cérébrale - Google Patents

Procédé et appareil d'évaluation de carte de structure de dépendance cérébrale Download PDF

Info

Publication number
WO2021120961A1
WO2021120961A1 PCT/CN2020/129512 CN2020129512W WO2021120961A1 WO 2021120961 A1 WO2021120961 A1 WO 2021120961A1 CN 2020129512 W CN2020129512 W CN 2020129512W WO 2021120961 A1 WO2021120961 A1 WO 2021120961A1
Authority
WO
WIPO (PCT)
Prior art keywords
brain
sample
training
fmri
addiction
Prior art date
Application number
PCT/CN2020/129512
Other languages
English (en)
Chinese (zh)
Inventor
王书强
余雯
吴国宝
胡圣烨
刘欣安
张炽堂
胡勇
Original Assignee
中国科学院深圳先进技术研究院
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 中国科学院深圳先进技术研究院 filed Critical 中国科学院深圳先进技术研究院
Publication of WO2021120961A1 publication Critical patent/WO2021120961A1/fr

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves 
    • A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves  involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/40—Detecting, measuring or recording for evaluating the nervous system
    • A61B5/4058—Detecting, measuring or recording for evaluating the nervous system for evaluating the central nervous system
    • A61B5/4064—Evaluating the brain
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/48—Other medical applications

Definitions

  • This application belongs to the field of artificial intelligence technology, and in particular relates to a method and device for evaluating brain addiction structure atlas.
  • fMRI Functional Magnetic Resonance Imaging
  • the embodiments of the present application provide a method and device for evaluating brain addiction structure atlas, which can solve the problem of poor recognition effect of the trained brain addiction trait recognition model due to the inability to obtain enough fMRI sample image training models.
  • the embodiments of the present application provide a method for evaluating brain addiction structure atlas, including:
  • the brain structure map is a brain fMRI image, including multiple brain region images.
  • the brain addiction character structure map evaluation model of the preset addictive substance the brain addiction trait types of different regions in the brain fMRI image are determined.
  • the brain addiction character structure map evaluation model of the preset addiction substance is based on the prediction It is assumed that the real fMRI images of the brain obtained under the influence of different concentrations of the addictive substance, and the multiple synthetic fMRI images of the brain under the influence of each concentration obtained by the sample generator corresponding to each concentration are obtained by training a preset classification model.
  • the multiple synthetic fMRI images of the brain under the influence of each concentration obtained by the sample generator corresponding to each concentration include: inputting a random noise vector into the sample generator corresponding to each concentration, wherein the sample generation
  • the detector is based on the real fMRI image of the brain under the corresponding concentration, and is obtained based on the training of the generative confrontation network.
  • the sample generator generates multiple synthetic fMRI images of the brain under the influence of the corresponding concentration according to the random noise vector.
  • the sample generator corresponding to each concentration can be obtained by training in the following steps: using a preset sample generator to generate training fMRI images according to random noise vectors.
  • the training fMRI image the real fMRI image of the brain with the concentration to be trained, and the sample discriminator
  • the training sample generator obtains the sample generator at the corresponding concentration, where the sample discriminator is used to determine the discrimination of the training fMRI image according to the real fMRI image
  • the discrimination result includes: true or false.
  • the training sample generator to obtain the sample generator at the corresponding concentration includes:
  • the training fMRI image and the real fMRI image are input to the sample discriminator, and the discrimination result of the training fMRI image is obtained. If the discrimination result of the training fMRI image is true, the sample discriminator is trained so that the sample discriminator determines that the discrimination result of the training fMRI image is false, and the trained sample discriminator is obtained. If the discrimination result of the training fMRI image is false, the sample generator is trained so that the sample discriminator determines that the discrimination result of the training fMRI image is true, and the sample generator is trained. Iteratively train the sample discriminator and the sample generator until the loss function of the trained sample generator is less than the preset threshold to obtain the sample generator at the corresponding concentration.
  • training the sample discriminator so that the sample discriminator determines that the discrimination result of the training fMRI image is false including:
  • the sample discriminator is trained until the sample discriminator determines that the discrimination result of the training fMRI image is false, and the trained sample discriminator is obtained.
  • training the sample generator to enable the sample discriminator to determine that the discrimination result of the training fMRI image is true includes:
  • the sample generator is trained until the sample discriminator determines that the discrimination result of the training fMRI image is true, and the trained sample generator is obtained.
  • the embodiment of the present application provides a brain addiction trait structure map evaluation device including:
  • the acquisition module is used to acquire a brain structure map to be classified.
  • the brain structure map is a brain fMRI image and includes multiple brain region images.
  • the determination module is used to determine the brain addiction trait types of different regions in the brain fMRI image according to the evaluation model of the brain addiction trait structure map of the preset addictive, wherein the brain addiction trait structure map of the preset addiction
  • the evaluation model is based on the real fMRI images of the brain obtained under the influence of different concentrations of preset addictions, and multiple synthetic fMRI images of the brain under the influence of each concentration obtained through the sample generator corresponding to each concentration. Training presets
  • the classification model is obtained.
  • the device further includes a synthesis module for inputting a random noise vector into a sample generator corresponding to each concentration, wherein the sample generator is based on the real fMRI image of the brain at the corresponding concentration, and is based on generating a confrontation network training owned.
  • the sample generator generates multiple synthetic fMRI images of the brain under the influence of the corresponding concentration according to the random noise vector.
  • the device further includes a training module for generating training fMRI images based on random noise vectors using a preset sample generator.
  • a training module for generating training fMRI images based on random noise vectors using a preset sample generator.
  • the training fMRI image the real fMRI image of the brain with the concentration to be trained, and the sample discriminator, the training sample generator obtains the sample generator at the corresponding concentration, where the sample discriminator is used to determine the discrimination of the training fMRI image according to the real fMRI image
  • the discrimination result includes: true or false.
  • the training module is specifically used to input the training fMRI image and the real fMRI image into the sample discriminator to obtain the discrimination result of the training fMRI image. If the discrimination result of the training fMRI image is true, the sample discriminator is trained so that the sample discriminator determines that the discrimination result of the training fMRI image is false, and the trained sample discriminator is obtained. If the discrimination result of the training fMRI image is false, train the sample generator so that the sample discriminator determines that the discrimination result of the training fMRI image is true, and the trained sample generator. Iteratively train the sample discriminator and the sample generator until the loss function of the trained sample generator is less than the preset threshold to obtain the sample generator at the corresponding concentration.
  • the training module is specifically used to obtain the loss function of the sample discriminator according to the discrimination result.
  • the sample discriminator is trained until the sample discriminator determines that the discrimination result of the training fMRI image is false, and the trained sample discriminator is obtained.
  • the training module is specifically used to obtain the loss function of the sample generator according to the discrimination result.
  • the sample generator is trained until the sample discriminator determines that the discrimination result of the training fMRI image is true, and the trained sample generator is obtained.
  • embodiments of the present application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and running on the processor.
  • the processor executes the computer program to implement the method provided in the first aspect. .
  • an embodiment of the present application provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method provided in the first aspect is implemented.
  • the embodiments of the present application provide a computer program product, which when the computer program product runs on a terminal device, causes the terminal device to execute the method provided in the above-mentioned first aspect.
  • Acquire fMRI images of the brain to be classified, and the fMRI images of the brain include images of multiple brain regions.
  • the pre-determined addiction’s brain addiction trait structure map evaluation model is based on the preset Real fMRI images of the brain obtained under the influence of different concentrations of addictive substances, and multiple synthetic fMRI images of the brain under the influence of each concentration obtained by the sample generator corresponding to each concentration are obtained by training a preset classification model.
  • the brain addiction trait structure map evaluation model trained with the expanded training samples can more accurately evaluate the brain fMRI images to be classified, and improve the classification effect of the brain addiction trait structure map evaluation model.
  • Fig. 1 is a schematic diagram of an application scenario of a brain addiction structure atlas evaluation method provided by an embodiment of the present application
  • FIG. 2 is a schematic flowchart of a method for evaluating a brain addiction structure atlas provided by an embodiment of the present application
  • FIG. 3 is a schematic flowchart of a method for evaluating a brain addiction structure atlas provided by another embodiment of the present application.
  • FIG. 4 is a schematic flowchart of a method for evaluating a brain addiction structure atlas provided by another embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of generating a confrontation network in a method for evaluating a brain addiction structure map provided by another embodiment of the present application.
  • Fig. 6 is a schematic flow chart of a method for evaluating a brain addiction structure atlas provided by another embodiment of the present application.
  • FIG. 7 is a schematic flowchart of a method for evaluating a brain addiction structure atlas provided by another embodiment of the present application.
  • FIG. 8 is a schematic flowchart of a method for evaluating a brain addiction structure atlas provided by another embodiment of the present application.
  • FIG. 9 is a schematic structural diagram of a brain addiction trait structure map evaluation device provided by an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of a brain addiction trait structure map evaluation device provided by another embodiment of the present application.
  • FIG. 11 is a schematic structural diagram of a brain addiction trait structure map evaluation device provided by another embodiment of the present application.
  • Fig. 12 is a schematic structural diagram of a terminal device provided by an embodiment of the present application.
  • Fig. 1 shows a schematic diagram of an application scenario of a brain addiction structure atlas evaluation method provided by an embodiment of the present application.
  • an fMRI image acquisition device 11 and a terminal device 12 are included.
  • the fMRI image acquisition device 11 may be a magnetic resonance imaging system (MRI).
  • MRI generally consists of three parts: The composition includes a magnet system, a radio frequency system, and a computer imaging system.
  • the magnet system is used to provide a constant magnetic field, place the part to be imaged (such as the head) in the magnetic field, and then transmit the radio frequency field through pulses through the radio frequency system to act on
  • the part to be imaged then receives the magnetic resonance information number generated by the part to be imaged, converts it into a digital signal, and obtains an fMRI image through a computer imaging system.
  • the terminal device 12 may be a device with computing capabilities, for example, a desktop computer, a notebook computer, a tablet computer, a smart phone, a server, a cloud server, etc., which are not limited here.
  • the terminal device 12 can communicate with the computer imaging system in the fMRI image acquisition device 11 through the network to acquire the brain fMRI image to be classified.
  • the terminal device 12 is used to implement the method for classifying nicotine addiction traits provided in this application.
  • FIG. 2 shows a schematic flow chart of a method for evaluating a brain addiction structure atlas. As an example and not a limitation, the method can be applied to the above-mentioned terminal device 12.
  • the brain structure map is a brain fMRI image, and includes multiple brain region images.
  • the fMRI image of the brain can be acquired by the fMRI image acquisition device described above.
  • the fMRI image of the brain to be classified includes images of multiple brain regions, and each brain region is stimulated by injection of different concentrations of the addictive substance. , There will be different reactions, and they will be displayed on the fMRI image.
  • the brain addiction trait structure map evaluation model of the preset addictive substance is based on the real fMRI images of the brain obtained under the influence of different concentrations of the preset addictive substance, and each of the fMRI images obtained through the sample generator corresponding to each concentration. Multiple synthetic fMRI images of the brain under the influence of concentration are obtained by training a preset classification model.
  • the preset addictive substance may be nicotine.
  • the real fMRI images of the brain obtained under the influence of different nicotine concentrations indicate the corresponding responses of each brain area after the stimulation of different concentrations of nicotine injection.
  • the sample generator corresponding to each concentration obtains multiple synthetic fMRI images of the brain under the influence of each concentration, and provides a simulated image of the brain area at that concentration.
  • the classification of nicotine addiction traits in rats can be used as an example.
  • an experimental control group of rats under the action of high and low concentrations of nicotine is used to collect fMRI images before and after injection of different concentrations of nicotine for a period of time.
  • the collected original data set can contain a total of 30 real fMRI images.
  • the 30 real fMRI images are divided into three groups, each with 10 real fMRI images, one group is the real fMRI image collected after high-concentration nicotine injection, and the other is Real fMRI images acquired after low-concentration nicotine injection, a group of real fMRI images acquired after injection of nicotine-free saline.
  • the sample generator corresponding to each concentration is used to generate a certain number of synthesized fMRI images at each concentration.
  • the number of synthesized fMRI images is subject to actual requirements, and there is no limitation here.
  • the brain addiction character structure map evaluation model of the preset addiction includes 3D convolutional layer, 3D average pooling layer, first 3D dense connection block, first and second-order pooling module, first transition layer, and second-order pooling module.
  • a softmax function is set after the fully connected layer, and a one-hot code composed of 0 and 1 is obtained.
  • the one-hot code can be used to represent the classification of the performance and mechanism of each region after the brain is injected with different concentrations of nicotine.
  • the gradient may be calculated according to the loss function C_loss of the classification model, and then the learnable parameters of the classification model network layer can be updated by the gradient descent method.
  • the loss function C_loss is the supervised loss, that is, the cross entropy of the classification task between the real fMRI image and the synthetic fMRI image, which can be expressed as:
  • y is the classification and labeling information
  • (x label , y) ⁇ P (x, y) means that the image label pair obeys the joint probability distribution of image x and label y
  • x label ) means under the condition of image Xlabel Is the conditional probability of labeling information y.
  • a brain fMRI image to be classified is acquired, and the brain fMRI image includes multiple brain region images.
  • the pre-determined addiction’s brain addiction trait structure map evaluation model is based on the preset Real fMRI images of the brain obtained under the influence of different concentrations of addictive substances, and multiple synthetic fMRI images of the brain under the influence of each concentration obtained by the sample generator corresponding to each concentration are obtained by training a preset classification model.
  • the brain addiction trait structure map evaluation model trained with the expanded training samples can more accurately evaluate the brain fMRI images to be classified, and improve the classification effect of the brain addiction trait structure map evaluation model.
  • Fig. 3 shows a schematic flow chart of another method for evaluating brain addiction structure atlas.
  • multiple synthetic fMRI images of the brain under the influence of each concentration obtained by the sample generator corresponding to each concentration include:
  • the sample generator is based on the real fMRI image of the brain under the corresponding concentration, and is obtained based on the training of the generated confrontation network.
  • the random noise vector may be a one-dimensional random noise vector with a Gaussian distribution
  • the sample generator may be a generator composed of 4 layers of 3D deconvolution layers.
  • the sample generator corresponding to each concentration is based on the real fMRI image collected after the injection of the concentration of nicotine.
  • the sample generator corresponding to the high concentration of nicotine corresponds to the injection of the high concentration of nicotine into After the rat, the real fMRI image of the rat’s brain was collected and obtained by training.
  • the sample generator generates multiple synthetic fMRI images of the brain under the influence of the corresponding concentration according to the random noise vector.
  • the difference between the synthetic fMRI image generated by the sample generator and the real fMRI image is very small, so the synthetic fMRI image and the real fMRI image can be used together to train the nicotine addiction trait classification model.
  • Fig. 4 shows a schematic flow chart of another method for evaluating brain addiction structure atlas.
  • the sample generator corresponding to each concentration can be obtained by training through the following steps:
  • Fig. 5 shows a schematic diagram of the structure of the generated confrontation network in a method for evaluating the brain addiction structure map.
  • the sample generator corresponding to each concentration is obtained through the training of the confrontation generation network.
  • the confrontation generation network includes a sample generator 52 and a noise vector 51 input to the sample generator 52.
  • the sample generator 52 generates a training fMRI image 53 according to the noise vector 51.
  • the sample generator 52 is not trained At this time, the generated training fMRI image 53 is very rough, and it can be easily distinguished as a composite image.
  • the first three deconvolution layers of the sample generator can use the vector convolution operation layer (Conv)-leaky corrected linear unit layer (Leaky Rectified Linear Unit, Leaky ReLU)-batch normalization layer (Batch Normalization) order , And remove Batch Normalization in the last layer, and use the hyperbolic tangent (Tanh) activation function.
  • Conv vector convolution operation layer
  • Leaky Rectified Linear Unit Leaky ReLU
  • Batch Normalization Batch Normalization
  • Tuh hyperbolic tangent activation function
  • the sample discriminator is used to determine the discrimination result of the training fMRI image according to the real fMRI image, and the discrimination result includes: true or false.
  • the sample discriminator 54 includes 4 layers of 3D convolutional layers and a fully connected layer.
  • the discriminator The 3D convolutional layer adopts the order of Conv—Leaky ReLU—Batch Normalization, and removes Batch Normalization in the first and last layers to maintain the independence of input and output.
  • the sample discriminator 54 performs feature extraction on the input training fMRI image 53 and the real fMRI image 54 through 4 layers of 3D convolutional layers, and then integrates the extracted local features through a fully connected layer to obtain the global training fMRI image 53 Features and global features of the real fMRI image 54.
  • the above global features are normalized to obtain the normalized feature value, and the normalized feature value is used as the sample discriminator 54's judgment on the input fMRI image, where if the judgment is true, Then the normalized eigenvalue is 1, otherwise, if the judgment is false, the normalized eigenvalue is 0.
  • Fig. 6 shows a schematic flow chart of a method for evaluating brain addiction structure atlas.
  • the sample generator at the corresponding concentration is obtained, including:
  • the real fMRI image can be preprocessed first.
  • the real fMRI image can be preprocessed for normalization, and the value of each voxel can be normalized to between [-1,1], and through rotation, Random cropping and other common computer vision processing methods are used to enhance the data in the real fMRI image.
  • the preprocessed real fMRI image and the training fMRI image are input into the sample discriminator in the form of high-order tensor respectively, and the discrimination result of the training fMRI image is obtained.
  • the generated confrontation network used needs to train the sample discriminator first to improve the discriminative ability. Its purpose is to discriminate the training fMRI image as false, and then train the sample generator to improve the generation ability. Its purpose is to make The sample discriminator discriminates the training fMRI image as true. The sample discriminator and the sample generator fight against the training. After a certain number of iterations, the sample discriminator discriminates the probability that the training fMRI image generated by the trained sample generator is true and false. Approaching 50%, that is, the sample discriminator cannot accurately identify the true and false of the training fMRI image, that is to say, the training fMRI image is realistic enough to be used as a training model for evaluating the structure of brain addiction traits.
  • S424 Iteratively train the sample discriminator and the sample generator until the loss function of the trained sample generator is less than the preset threshold, and the sample generator at the corresponding concentration is obtained.
  • the training of the sample generator and the sample discriminator are both implemented through a backpropagation algorithm.
  • the preset threshold can be set according to the required accuracy in actual applications, and there is no limitation here.
  • Fig. 7 shows a schematic flow chart of a method for evaluating brain addiction structure atlas.
  • the sample discriminator is trained so that the sample discriminator determines that the discrimination result of the training fMRI image is false, including:
  • the goal of the sample discriminator is to judge the real fMRI image as true, and the training fMRI image is judged as false. Therefore, its loss function consists of two parts, which can be expressed as:
  • x is the input image
  • D(x) is the probability that the sample discriminator judges the image x to be true
  • z is the random noise vector
  • G(z) represents the synthetic fMRI image generated from the random noise vector z
  • D(G( z)) represents the probability that the generated image G(z) is true by the sample discriminator
  • ⁇ d represents the parameters of the sample discriminator network layer
  • x ⁇ real means that the image x obeys the real data distribution
  • z ⁇ noise means the random noise vector z obeys Random normal distribution.
  • Fig. 8 shows a schematic flow chart of a method for evaluating brain addiction structure atlas.
  • the training sample generator enables the sample discriminator to determine that the discrimination result of the training fMRI image is true, including:
  • the gradient may be calculated according to the loss function G_loss of the sample generator, and then the learnable parameters of the sample generator network layer can be updated by the gradient descent method.
  • the goal of the sample generator is to generate training fMRI images that can deceive the sample discriminator by simulating real brain fMRI features.
  • the loss function consists of two parts, one is to deceive the sample discriminator to make it judge the training fMRI image as true, and the other is the reconstruction loss between the real fMRI image and the training fMRI image to make the generated image closer
  • the loss function G_loss of the sample generator can be expressed as:
  • x real represents the real fMRI image
  • ⁇ is the weight parameter of the L1 loss function
  • ⁇ g is the parameter of the sample generator network layer.
  • FIG. 9 shows the structure block diagram of the brain addiction trait structure map evaluation device provided by the embodiment of the present application. Apply for the relevant part of the embodiment.
  • the device includes:
  • the acquiring module 61 is configured to acquire a brain structure map to be classified, the brain structure map is a brain fMRI image, and includes a plurality of brain region images.
  • the determining module 62 is used to determine the brain addiction trait types of different regions in the fMRI image of the brain according to the evaluation model of the brain addiction trait structure map of the preset addiction, wherein the brain addiction trait structure of the preset addiction
  • the map evaluation model is based on the real fMRI images of the brain obtained under the influence of different concentrations of the preset addictive substance, and the multiple synthetic fMRI images of the brain under the influence of each concentration obtained through the sample generator corresponding to each concentration.
  • the classification model is obtained.
  • Fig. 10 shows a structural block diagram of another apparatus for evaluating brain addiction traits.
  • the device further includes a synthesis module 63 for inputting a random noise vector into a sample generator corresponding to each concentration, where the sample generator is based on the actual fMRI of the brain at the corresponding concentration. Image, based on the training of the generative confrontation network. The sample generator generates multiple synthetic fMRI images of the brain under the influence of the corresponding concentration according to the random noise vector.
  • Fig. 11 shows a structural block diagram of another apparatus for evaluating brain addiction traits.
  • the device further includes a training module 64, which is configured to use a preset sample generator to generate training fMRI images according to random noise vectors.
  • a training module 64 which is configured to use a preset sample generator to generate training fMRI images according to random noise vectors.
  • the training fMRI image the real fMRI image of the brain with the concentration to be trained, and the sample discriminator, the training sample generator obtains the sample generator at the corresponding concentration, where the sample discriminator is used to determine the discrimination of the training fMRI image according to the real fMRI image
  • the discrimination result includes: true or false.
  • the training module 64 is specifically configured to input the training fMRI image and the real fMRI image into the sample discriminator to obtain the discrimination result of the training fMRI image. According to the discrimination result of the training fMRI image, the sample generator or sample discriminator is trained until the loss function of the sample generator is less than the preset threshold, and the sample generator at the corresponding concentration is obtained.
  • the training module 64 is specifically configured to obtain the loss function of the sample discriminator according to the discrimination result if the sample discriminator determines that the discrimination result of the training fMRI image is true.
  • the sample discriminator is trained through the loss function of the sample discriminator, until the sample discriminator determines that the discriminant result of the training fMRI image is false.
  • the training module 64 is specifically configured to obtain the loss function of the sample generator according to the discrimination result if the sample discriminator determines that the discrimination result of the training fMRI image is false.
  • the sample generator is trained through the loss function of the sample generator until the sample discriminator determines that the discrimination result of the training fMRI image is true.
  • FIG. 12 shows a schematic structural diagram of the terminal device.
  • the terminal device 7 includes a memory 72, a processor 71, and a computer program 73 that is stored in the memory 7 and can run on the processor 71.
  • the processor 71 executes the computer program 73, the aforementioned brain addiction is realized. Structure map evaluation method.
  • the processor 71 may be a central processing unit (Central Processing Unit, CPU), and the processor 71 may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (ASICs). ), off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor or the like.
  • the memory 72 may be an internal storage unit of the terminal device 7 in some embodiments, such as a hard disk, flash memory, or memory of the terminal device 7. In other embodiments, the memory 72 may also be an external storage device of the terminal device 7, such as a plug-in hard disk equipped on the terminal device 7, a smart memory card (Smart Media Card, SMC), and a Secure Digital (SD). Card, Flash Card, etc. Further, the memory 72 may also include both an internal storage unit of the terminal device 7 and an external storage device.
  • the memory 72 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, for example, the program code of the computer program 73, videos, and the like.
  • the memory 72 can also be used to temporarily store data that has been output or will be output.
  • the embodiments of the present application also provide a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in each of the foregoing method embodiments can be realized.
  • the embodiments of the present application provide a computer program product.
  • the steps in the foregoing method embodiments can be realized when the mobile terminal is executed.
  • the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium.
  • the computer program can be stored in a computer-readable storage medium.
  • the computer program can be stored in a computer-readable storage medium.
  • the steps of the foregoing method embodiments can be implemented.
  • the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file, or some intermediate forms.
  • the computer-readable medium may at least include: any entity or device capable of carrying the computer program code to the photographing device/terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), and random access memory (RAM, Random Access Memory), electric carrier signal, telecommunications signal and software distribution medium.
  • ROM read-only memory
  • RAM random access memory
  • electric carrier signal telecommunications signal and software distribution medium.
  • U disk mobile hard disk, floppy disk or CD-ROM, etc.
  • computer-readable media cannot be electrical carrier signals and telecommunication signals.
  • the disclosed apparatus/network equipment and method may be implemented in other ways.
  • the device/network device embodiments described above are only illustrative.
  • the division of the modules or units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units.
  • components can be combined or integrated into another system, or some features can be omitted or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components displayed as units 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 units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Surgery (AREA)
  • General Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Pathology (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Neurology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Psychology (AREA)
  • Neurosurgery (AREA)
  • Physiology (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

Utilisée dans le domaine technique de l'intelligence artificielle, l'invention concerne un procédé d'évaluation de carte de structure de dépendance cérébrale et un appareil, le procédé comprenant les étapes consistant à : acquérir une image de structure cérébrale à catégoriser, l'image de structure cérébrale étant une image IRMf du cerveau comprenant de multiples images de région cérébrale (S21) ; selon un modèle d'évaluation de carte structurelle d'état de dépendance cérébrale pour un objet prédéfini de dépendance, déterminer des catégories d'état de dépendance cérébrale de différentes régions dans l'image IRMf du cerveau (S22). En utilisant un générateur d'échantillon correspondant à chaque concentration pour obtenir de multiples images IRMf intégrées d'un cerveau sous l'influence de chaque concentration lors de l'apprentissage du modèle d'évaluation de carte structurelle d'état de dépendance cérébrale prédéterminé, les images IRMf intégrées augmentant efficacement l'échantillon d'apprentissage, le modèle d'évaluation de carte structurelle d'état de dépendance cérébrale obtenu par apprentissage à l'aide de l'échantillon d'apprentissage augmenté peut évaluer plus précisément une image IRMf du cerveau devant être catégorisée, améliorant des résultats de catégorisation du modèle d'évaluation de carte structurelle d'état de dépendance cérébrale.
PCT/CN2020/129512 2019-12-16 2020-11-17 Procédé et appareil d'évaluation de carte de structure de dépendance cérébrale WO2021120961A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911294069.9 2019-12-16
CN201911294069.9A CN111067522A (zh) 2019-12-16 2019-12-16 大脑成瘾结构图谱评估方法及装置

Publications (1)

Publication Number Publication Date
WO2021120961A1 true WO2021120961A1 (fr) 2021-06-24

Family

ID=70315093

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/129512 WO2021120961A1 (fr) 2019-12-16 2020-11-17 Procédé et appareil d'évaluation de carte de structure de dépendance cérébrale

Country Status (2)

Country Link
CN (1) CN111067522A (fr)
WO (1) WO2021120961A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114463288A (zh) * 2022-01-18 2022-05-10 深圳市铱硙医疗科技有限公司 脑部医学影像评分方法、装置、计算机设备和存储介质

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111067522A (zh) * 2019-12-16 2020-04-28 中国科学院深圳先进技术研究院 大脑成瘾结构图谱评估方法及装置
CN111569225B (zh) * 2020-05-08 2021-07-06 中国科学技术大学 实时功能性磁共振神经反馈调节认知控制的系统及方法
CN113920123B (zh) * 2021-12-16 2022-03-15 中国科学院深圳先进技术研究院 一种成瘾性脑网络分析方法及装置
WO2023123380A1 (fr) * 2021-12-31 2023-07-06 深圳先进技术研究院 Procédé et système de génération de circuit neuronal d'addiction dynamique basés sur un apprentissage contrastif faiblement supervisé

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106510709A (zh) * 2016-09-23 2017-03-22 中国人民解放军第四军医大学 一种针对海洛因成瘾机制的大型脑网络分析方法
CN108596267A (zh) * 2018-05-03 2018-09-28 Oppo广东移动通信有限公司 一种图像重建方法、终端设备及计算机可读存储介质
CN110070935A (zh) * 2019-03-20 2019-07-30 中国科学院自动化研究所 基于对抗神经网络的医学图像合成方法、分类方法及装置
US20190329063A1 (en) * 2016-12-25 2019-10-31 The Medical Research, Infrastructure and Health Services Fund of the Tel Aviv Medical Center Method for modifying the therapeutic effects of drugs
CN110503654A (zh) * 2019-08-01 2019-11-26 中国科学院深圳先进技术研究院 一种基于生成对抗网络的医学图像分割方法、系统及电子设备
CN111067522A (zh) * 2019-12-16 2020-04-28 中国科学院深圳先进技术研究院 大脑成瘾结构图谱评估方法及装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101530494B1 (ko) * 2014-10-21 2015-06-30 충남대학교산학협력단 표준화된 성적 각성 유도와 기능적 자기 공명 영상을 이용한 성중독자 판별모형의 생성장치 및 판별모형의 도출방법
CN106250707A (zh) * 2016-08-12 2016-12-21 王双坤 一种基于深度学习算法处理头部结构像数据的方法
CN106823137B (zh) * 2016-12-30 2019-07-16 王征 一种优化神经调控的方法和装置
CN110503187B (zh) * 2019-07-26 2024-01-16 深圳万知达科技有限公司 一种用于功能核磁共振成像数据生成的生成对抗网络模型的实现方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106510709A (zh) * 2016-09-23 2017-03-22 中国人民解放军第四军医大学 一种针对海洛因成瘾机制的大型脑网络分析方法
US20190329063A1 (en) * 2016-12-25 2019-10-31 The Medical Research, Infrastructure and Health Services Fund of the Tel Aviv Medical Center Method for modifying the therapeutic effects of drugs
CN108596267A (zh) * 2018-05-03 2018-09-28 Oppo广东移动通信有限公司 一种图像重建方法、终端设备及计算机可读存储介质
CN110070935A (zh) * 2019-03-20 2019-07-30 中国科学院自动化研究所 基于对抗神经网络的医学图像合成方法、分类方法及装置
CN110503654A (zh) * 2019-08-01 2019-11-26 中国科学院深圳先进技术研究院 一种基于生成对抗网络的医学图像分割方法、系统及电子设备
CN111067522A (zh) * 2019-12-16 2020-04-28 中国科学院深圳先进技术研究院 大脑成瘾结构图谱评估方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ELLIOT A. STEIN, JOHN PANKIEWICZ, HAROLD H. HARSCH, JUNG-KI CHO, SCOTT A. FULLER, RAYMOND G. HOFFMANN, MARJORIE HAWKINS, STEPHEN M: "Nicotine-Induced Limbic Cortical Activation in the Human Brain: A Functional MRI Study", AM J PSYCHIATRY, vol. 155, no. 5, 1 August 1998 (1998-08-01), pages 1009 - 1015, XP055823162, DOI: 10.1176/ajp.155.8.1009 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114463288A (zh) * 2022-01-18 2022-05-10 深圳市铱硙医疗科技有限公司 脑部医学影像评分方法、装置、计算机设备和存储介质
CN114463288B (zh) * 2022-01-18 2023-01-10 深圳市铱硙医疗科技有限公司 脑部医学影像评分方法、装置、计算机设备和存储介质

Also Published As

Publication number Publication date
CN111067522A (zh) 2020-04-28

Similar Documents

Publication Publication Date Title
WO2021120961A1 (fr) Procédé et appareil d'évaluation de carte de structure de dépendance cérébrale
CN109583332B (zh) 人脸识别方法、人脸识别系统、介质及电子设备
CN111241989B (zh) 图像识别方法及装置、电子设备
CN108197592B (zh) 信息获取方法和装置
CN111932529B (zh) 一种图像分类分割方法、装置及系统
CN113822977A (zh) 图像渲染方法、装置、设备以及存储介质
CN108229375B (zh) 用于检测人脸图像的方法和装置
Öztürk et al. Cell‐type based semantic segmentation of histopathological images using deep convolutional neural networks
CN108509994B (zh) 人物图像聚类方法和装置
CN108388889B (zh) 用于分析人脸图像的方法和装置
CN113424222A (zh) 用于使用条件生成对抗网络提供中风病灶分割的系统和方法
CN108399401B (zh) 用于检测人脸图像的方法和装置
CN115272295A (zh) 基于时域-空域联合状态的动态脑功能网络分析方法及系统
Azaza et al. Context proposals for saliency detection
CN111488810A (zh) 人脸识别方法、装置、终端设备及计算机可读介质
CN115170401A (zh) 图像补全方法、装置、设备及存储介质
CN113850796A (zh) 基于ct数据的肺部疾病识别方法及装置、介质和电子设备
CN112258647A (zh) 地图重建方法及装置、计算机可读介质和电子设备
CN111461091B (zh) 万能指纹生成方法和装置、存储介质及电子装置
CN115719428A (zh) 基于分类模型的人脸图像聚类方法、装置、设备及介质
Fua et al. Modeling brain circuitry over a wide range of scales
CN114283114A (zh) 图像处理方法、装置、设备及存储介质
Chhabra et al. Comparison of different edge detection techniques to improve quality of medical images
CN112086174A (zh) 一种三维知识诊断模型构建方法和系统
CN110517239A (zh) 一种医学影像检测方法及装置

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: 20901921

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: 20901921

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205N DATED 10.01.2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20901921

Country of ref document: EP

Kind code of ref document: A1