WO2025129461A1 - 一种基于小样本数据增强的自闭症风险评估方法及装置 - Google Patents

一种基于小样本数据增强的自闭症风险评估方法及装置 Download PDF

Info

Publication number
WO2025129461A1
WO2025129461A1 PCT/CN2023/139971 CN2023139971W WO2025129461A1 WO 2025129461 A1 WO2025129461 A1 WO 2025129461A1 CN 2023139971 W CN2023139971 W CN 2023139971W WO 2025129461 A1 WO2025129461 A1 WO 2025129461A1
Authority
WO
WIPO (PCT)
Prior art keywords
brain
risk assessment
data enhancement
channel
connectivity map
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/CN2023/139971
Other languages
English (en)
French (fr)
Inventor
徐永杰
王怡珊
李烨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to PCT/CN2023/139971 priority Critical patent/WO2025129461A1/zh
Publication of WO2025129461A1 publication Critical patent/WO2025129461A1/zh
Anticipated expiration legal-status Critical
Pending legal-status Critical Current

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems

Definitions

  • the present application belongs to the field of deep learning technology, and in particular relates to an autism risk assessment method and device based on small sample data enhancement.
  • Autism spectrum disorder is a neurodevelopmental disorder in which the functional connectivity of the brain of autistic patients is significantly different from that of normal people. Autism can cause persistent defects in patients' social communication skills and behaviors. As of 2020, the prevalence of autism in children aged 6-12 in China is approximately 0.70%, which means that one in every 143 children suffers from autism. People with autism will impose a serious economic burden on families and society. Traditional clinical diagnosis of autism uses observation scales and behavioral methods to evaluate and analyze autism, which requires doctors to have a high level of professionalism and has low diagnostic efficiency.
  • Autism risk assessment can be achieved by building an autism recognition model.
  • EEG Electronic EEG
  • ASD Automatic Spectrum Disorders
  • a domestic study extracted four entropy features from EEG (Electroencephalogram) signals of ASD (Autistic Spectrum Disorders) children and normal children, and used independent sample t-test to analyze the differences between the groups.
  • the support vector machine learning algorithm was used to establish a classification model for each entropy measurement in different brain regions, with a classification accuracy of up to 84.55%.
  • Another foreign study converted EEG time domain signals into frequency domain signals through discrete Fourier transform, designed a reinforcement learning algorithm to quickly update and reconstruct the convolutional neural network model, and used the final convolutional neural network model for classification, with a recognition accuracy of up to 92.63%.
  • the existing autism detection methods based on EEG signals generally use the original signal or spectrum characteristics of EEG signals, which is limited by the number of subjects.
  • Most existing technologies use the data of subjects at different time segments as independent samples for repeated use, and only use part of the EEG data for analysis each time, which may miss more important EEG information, and this analysis method may cause data pollution, resulting in distorted analysis results; or reuse a subject's data to increase it to multiple samples, resulting in too high similarity between samples.
  • the present application provides an autism risk assessment method and device based on small sample data enhancement, aiming to solve at least one of the above-mentioned technical problems in the prior art to a certain extent.
  • a method for autism risk assessment based on small sample data enhancement comprising:
  • the multi-channel EEG signals are calculated using the Pearson correlation coefficient to obtain the correlation between the channels, generate a brain functional connection matrix, and construct a brain temporal connectivity map according to the timing information of the multi-channel EEG signals;
  • the brain temporal connectivity map is subjected to data enhancement processing using a generative adversarial network, and the data-enhanced brain temporal connectivity map is input into a trained deep convolutional network and long short-term memory network joint model for classification, and the classifier outputs the subject's autism risk assessment result.
  • the technical solution adopted by the embodiment of the present application also includes: after collecting the multi-channel EEG signal of the subject, it also includes:
  • Preprocessing the multi-channel EEG signals includes: filtering the multi-channel EEG signals using an anti-aliasing filter, digitizing and amplifying all EEG signals at a set sampling rate;
  • the amplified multi-channel EEG signal is corrected using a baseline correction algorithm, and the multi-channel EEG signal is artifact-removed using an artifact detection algorithm to obtain a preprocessed multi-channel EEG signal.
  • the technical solution adopted in the embodiment of the present application also includes: using the Pearson correlation coefficient to calculate the multi-channel EEG signal to obtain the correlation between each channel, generate a brain functional connection matrix, and construct a brain time series connectivity map according to the time series information of the multi-channel EEG signal, specifically:
  • each channel and other channels in a subject's EEG segment is calculated by Pearson's correlation coefficient to obtain a functional connectivity matrix of a set size
  • the technical solution adopted by the embodiment of the present application also includes: after constructing the brain time series connectivity map according to the time series information of the multi-channel EEG signal, it also includes:
  • the brain temporal connectivity map is divided into a training set and a validation set according to a set ratio.
  • the technical solution adopted by the embodiment of the present application also includes: using the generative adversarial network to perform data enhancement processing on the brain temporal connectivity map, specifically:
  • the generative adversarial network includes a generator and a discriminator.
  • the generator includes three deconvolution layers, which map the noise vector to the brain temporal connectivity map through the deconvolution layers to generate a new brain temporal connectivity map.
  • the discriminator uses three convolution layers to determine the authenticity of the generated brain temporal connectivity map, and after the generator and the discriminator iterate and confront a set number of times, the brain temporal connectivity map generated by the generator is used as an expanded pseudo sample.
  • the technical solution adopted by the embodiment of the present application also includes: the brain temporal connectivity map after data enhancement processing is input into the trained deep convolutional network and long short-term memory network joint model for classification, and the classifier outputs the autism risk assessment result of the subject, specifically:
  • the deep convolutional network part of the joint model of the deep convolutional network and the long short-term memory network includes 5 convolutional layers and 4 pooling layers.
  • the local features of the input data are learned through the convolutional layers, and the convolutional layers are activated by the ReLU function;
  • the long short-term memory network includes a hidden layer, a recurrent network layer and a fully connected layer.
  • the input data is successively reshaped by the hidden layer and two recurrent network layers, and then passes through two fully connected layers, and finally the classifier is connected to output a binary classification result, and the binary classification result includes two categories: "autism" and "health".
  • an autism risk assessment device based on small sample data enhancement comprising:
  • Signal acquisition module used to collect multi-channel EEG signals of subjects
  • a connectivity map construction module is used to calculate the multi-channel EEG signals using the Pearson correlation coefficient to obtain the correlation between the channels, generate a brain functional connection matrix, and construct a brain temporal connectivity map according to the timing information of the multi-channel EEG signals;
  • a data enhancement module used for performing data enhancement processing on the brain temporal connectivity map by using a generative adversarial network
  • Risk assessment module used to input the brain temporal connectivity map after data enhancement processing into the trained deep convolutional network and long short-term memory network joint model for classification, and the classifier outputs the subject's autism risk assessment result.
  • the technical solution adopted by the embodiment of the present application also includes: the connectivity map construction module calculates the multi-channel EEG signal using the Pearson correlation coefficient to obtain the correlation between each channel, generates a brain functional connection matrix, and constructs a brain time series connectivity map according to the time series information of the multi-channel EEG signal. Specifically:
  • each channel and other channels in a subject's EEG segment is calculated by Pearson's correlation coefficient to obtain a functional connectivity matrix of a set size
  • the EEG signals of a set number of time segments collected from a subject are organized into a brain functional connection matrix of a set size in chronological order, and the brain functional connection matrix is converted into a brain temporal connectivity map according to the gray value distribution.
  • a computer device includes a processor and a memory coupled to the processor, wherein:
  • the memory stores program instructions for implementing the autism risk assessment method based on small sample data enhancement
  • the processor is used to execute the program instructions stored in the memory to control an autism risk assessment method based on small sample data enhancement.
  • a storage medium storing program instructions executable by a processor, wherein the program instructions are used to execute the autism risk assessment method based on small sample data enhancement.
  • the autism risk assessment method and device based on small sample data enhancement of the embodiments of the present application calculates the brain functional connection matrix using EEG signals, and constructs a brain temporal connectivity map based on the time information of the EEG signals, and uses a generative adversarial network to perform sample expansion on the brain temporal connectivity map, and then inputs it into a joint model of a deep convolutional network and a long short-term memory network for classification, and the classifier outputs the autism risk assessment result.
  • the perspective of brain functional connectivity takes into account the differences between autistic patients and healthy people, while taking into account the time information of brain functional connectivity and EEG signals, which can more accurately assess the risk of autism in subjects.
  • FIG1 is a flow chart of an autism risk assessment method based on small sample data enhancement according to an embodiment of the present application
  • FIG2 is a brain temporal connectivity diagram constructed in an embodiment of the present application.
  • FIG3 is a schematic diagram of a model structure of a generative adversarial network according to an embodiment of the present application.
  • FIG4 is a schematic diagram of the structure of a joint model of a deep convolutional network and a long short-term memory network according to an embodiment of the present application;
  • FIG5 is a schematic diagram of the structure of an autism risk assessment device based on small sample data enhancement according to an embodiment of the present application
  • FIG6 is a schematic diagram of the structure of a computer device according to an embodiment of the present application.
  • FIG. 7 is a schematic diagram of the structure of a storage medium according to an embodiment of the present application.
  • first”, “second”, “third” in this application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features.
  • the features defined as “first”, “second”, “third” can expressly or implicitly include at least one of the features.
  • the meaning of “multiple” is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
  • all directional indications (such as up, down, left, right, front, back%) are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly.
  • Figure 1 is a flow chart of the autism risk assessment method based on small sample data enhancement in an embodiment of the present application.
  • the autism risk assessment method based on small sample data enhancement in an embodiment of the present application includes the following steps:
  • S100 Collect multi-channel EEG signals of the subjects
  • the collected multi-channel EEG signal has 125 channels, and the specific number of channels can be set according to the actual application scenario.
  • the preprocessing process of the signal includes: filtering the multi-channel EEG signal with an anti-aliasing filter, digitizing and amplifying all EEG signals at a set sampling rate. Finally, the amplified multi-channel EEG signal is corrected using a baseline correction algorithm, and the multi-channel EEG signal is artifact-removed using an artifact detection algorithm to obtain the preprocessed multi-channel EEG signal.
  • Pearson product-moment correlation coefficient is used to calculate the preprocessed multi-channel EEG signals to obtain the correlation between each channel, generate the brain functional connection matrix, and construct the brain temporal connectivity map in time sequence;
  • the correlation between each channel and other channels in an EEG segment of the subject is calculated by the Pearson correlation coefficient, thereby obtaining a 125 ⁇ 125 functional connection matrix.
  • the upper triangular matrix of the functional connection matrix is expanded into one dimension, and then the SelectKBest algorithm is used as a feature screening algorithm to screen the functional connection features, and the set number of main functional connection features is retained. Assuming that the main functional connection features retained are 400, the EEG signals of 40 time segments collected by a subject are organized into a 400 ⁇ 40 brain functional connection matrix in chronological order, and then the brain functional connection matrix is converted into a brain temporal connectivity map according to the gray value distribution, and all brain temporal connectivity maps are divided into training sets and verification sets according to the set ratio.
  • the division ratio of the training set is 80%
  • the division ratio of the verification set is 20%, which can be set according to the actual application scenario.
  • Figure 2 it is a brain temporal connectivity map constructed by an embodiment of the present application, wherein (a) is a brain temporal connectivity map of autistic patients, and (b) is a brain temporal connectivity map of a healthy control group; from Figure (2), it can be seen that the brain temporal connectivity maps of autistic patients and the healthy control group have detailed differences.
  • the phase locking value can also be used as a connection indicator for constructing a brain temporal connectivity map.
  • S130 using a generative adversarial network (DCGAN) to perform data enhancement processing on the brain temporal connectivity map in the training set to generate an expanded training set;
  • DCGAN generative adversarial network
  • the embodiment of the present application uses pytorch to design a generative adversarial network (DCGAN), and uses the generative adversarial network as a data enhancement method to expand the sample size of the brain temporal connectivity map.
  • DCGAN generative adversarial network
  • the model structure of the generative adversarial network of the embodiment of the present application is shown in Figure 3.
  • the generative adversarial network includes two parts: a generator and a discriminator.
  • the generator part consists of 3 layers of deconvolution layers, which maps the 100x1 noise vector to the 400x40x1 brain temporal connectivity map through deconvolution to generate a new brain temporal connectivity map; in the discriminator part, 3 layers of convolution layers are used to discriminate the authenticity of the generated brain temporal connectivity map. After the generator and the discriminator iteratively confront each other 100 times, the brain temporal connectivity map generated by the generator is used as an expanded pseudo sample. In the embodiment of the present application, data enhancement processing is only performed on the training set, which will not affect the results of the verification set classification experiment.
  • S140 inputting the expanded training set into the constructed deep convolutional network and long short-term memory network joint model (CNN-LSTM) for classification, and the classifier outputs the autism risk assessment result of the subject;
  • CNN-LSTM long short-term memory network joint model
  • the deep convolutional network part includes 5 convolutional layers and 4 pooling layers.
  • the local features of the input data are learned through convolutional layers of size 3X3 and 2x2.
  • the convolutional layer is activated by the ReLU function.
  • the hidden layer dimension size of the long short-term memory network is set to 64, the number of layers of the recurrent network is set to 2 layers, and the fully connected layer is set to 2 layers. Since the constructed brain temporal connectivity map is a grayscale image, the initial input data size is 400 ⁇ 40 ⁇ 1.
  • the image size becomes 400 ⁇ 1 ⁇ 1, and then the image size becomes 64 ⁇ 1 ⁇ 1 after the hidden layer shaping and two recurrent network layers in sequence, and then after two fully connected layers, the classifier is finally connected to output the final binary classification result.
  • the joint model of deep convolutional network and long short-term memory network is experimentally verified using the 5-fold cross-validation technique to obtain the optimal joint model of deep convolutional network and long short-term memory network.
  • the binary classification results include two types of results: "autism” and "health”. If the classification result is "autism", it means that the subject has a high risk of autism and needs to be diagnosed more professionally in the next step. It can be understood that the embodiment of the present application only takes the autism risk assessment based on small sample data enhancement as an example, and the embodiment of the present application is also applicable to other medical images with small sample sizes and assessment classifications of images with spatiotemporal characteristics.
  • the autism risk assessment method based on small sample data enhancement of the embodiment of the present application calculates the brain function connection matrix using EEG signals, and constructs a brain temporal connectivity map according to the time information of EEG signals, and uses a generative adversarial network (DCGAN) to expand the brain temporal connectivity map after sample expansion, and inputs it into a deep convolutional network and a long short-term memory network joint model (CNN-LSTM) for classification, and outputs autism risk assessment results by a classifier.
  • the embodiment of the present application considers the difference between autistic patients and healthy people from the perspective of brain functional connectivity, and considers the time information of brain functional connectivity and EEG signals at the same time, and can more accurately assess autism risk to the subject.
  • the EEG signal of each subject's overall time is used as a sample, which is also more in line with the actual diagnosis scene.
  • Sample expansion is carried out by designing a generative adversarial network model, which effectively solves the problem that the sample size is generally less.
  • Autism risk assessment is carried out by constructing a deep convolutional network and a long short-term memory network joint model, and more comprehensive EEG signal features can be considered, further improving the accuracy of autism risk assessment.
  • FIG5 is a schematic diagram of the structure of an autism risk assessment device based on small sample data enhancement according to an embodiment of the present application.
  • the autism risk assessment device 40 based on small sample data enhancement according to an embodiment of the present application comprises:
  • Signal acquisition module 41 used for acquiring multi-channel EEG signals of the subject
  • Connectivity map construction module 42 used to calculate the multi-channel EEG signals using the Pearson correlation coefficient to obtain the correlation between the channels, generate a brain functional connection matrix, and construct a brain time-series connectivity map according to the time-series information of the multi-channel EEG signals;
  • Risk assessment module 44 used to perform data enhancement processing on the brain temporal connectivity map using a generative adversarial network, and input the brain temporal connectivity map after data enhancement processing into a trained deep convolutional network and long short-term memory network joint model for classification, and the classifier outputs the subject's autism risk assessment result.
  • the device provided in the embodiment of the present application can be applied in the aforementioned method embodiment.
  • a processor 52 connected to the memory 51;
  • the processor 52 is used to call the executable program instructions stored in the memory 51 and perform the following steps: collecting multi-channel EEG signals of the subject; calculating the multi-channel EEG signals using the Pearson correlation coefficient to obtain the correlation between each channel, generating a brain functional connection matrix, and constructing a brain temporal connectivity map according to the timing information of the multi-channel EEG signals; using a generative adversarial network to perform data enhancement processing on the brain temporal connectivity map, and inputting the brain temporal connectivity map after data enhancement processing into a trained deep convolutional network and long short-term memory network joint model for classification, and the classifier outputs the subject's autism risk assessment result.
  • the processor 52 may also be referred to as a CPU (Central Processing Unit).
  • the processor 52 may be an integrated circuit chip with signal processing capabilities.
  • the processor 52 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices.
  • DSP digital signal processor
  • ASIC application-specific integrated circuit
  • FPGA field-programmable gate array
  • a general purpose processor may be a microprocessor or any conventional processor.
  • Figure 7 is a schematic diagram of the structure of the storage medium of the embodiment of the present application.
  • the storage medium of the embodiment of the present application stores a program instruction 61 that can implement the following steps: collect the multi-channel EEG signal of the subject; calculate the multi-channel EEG signal using the Pearson correlation coefficient, obtain the correlation between each channel, generate a brain functional connection matrix, and construct a brain temporal connectivity map according to the timing information of the multi-channel EEG signal; use a generative adversarial network to perform data enhancement processing on the brain temporal connectivity map, and input the brain temporal connectivity map after the data enhancement processing into the trained deep convolutional network and long short-term memory network joint model for classification, and output the autism risk assessment result of the subject by the classifier.
  • the program instruction 61 can be stored in the above-mentioned storage medium in the form of a software product, including several instructions to make a computer device (which can be a personal computer, a server, or a network computer device, etc.) or a processor (processor) perform all or part of the steps of each implementation method of the present application.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program instructions, or terminal computer devices such as computers, servers, mobile phones, tablets, etc.
  • the server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as big data and artificial intelligence platforms.
  • cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as big data and artificial intelligence platforms.
  • each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of software functional units. The above is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the description and drawings of this application, or directly or indirectly used in other related technical fields, is also included in the patent protection scope of the present application.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Molecular Biology (AREA)
  • Psychiatry (AREA)
  • Psychology (AREA)
  • Physics & Mathematics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biophysics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Veterinary Medicine (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

本申请涉及一种基于小样本数据增强的自闭症风险评估方法及装置。所述方法包括：采集受试者的多通道脑电信号；利用皮尔森相关系数对多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按多通道脑电信号的时序信息构建大脑时序连通图；利用生成对抗网络对所述大脑时序连通图进行数据增强处理，并将数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，由分类器输出受试者的自闭症风险评估结果。本申请实施例从大脑功能连通性的角度考虑了自闭症患者与健康人群之间存在的差异性，同时考虑了大脑功能连接性与脑电信号所具有的时间信息，能更准确地对受试者进行自闭症风险评估。

Description

一种基于小样本数据增强的自闭症风险评估方法及装置 技术领域
本申请属于深度学习技术领域，特别涉及一种基于小样本数据增强的自闭症风险评估方法及装置。
背景技术
自闭症谱系障碍是一种神经发育障碍性疾病，自闭症患者的大脑功能连通性与正常人相比有明显的差异。自闭症会导致患者的社交沟通技巧和行为持续存在缺陷。截至2020年，在中国6-12岁儿童中的自闭症患病率约为0.70％，相当于每143名儿童中就有一名患有自闭症。自闭症人群会给家庭与社会造成严重的经济负担，传统的自闭症临床诊断方案采取观察量表和行为学等方式进行自闭症的评估和分析，需要医生具有较高的专业水平，且诊断效率较低。
对于自闭症风险评估可以通过构建自闭症识别模型来实现，目前有很多研究者利用脑电信号与机器学习和深度学习的方法构建自闭症的识别模型。例如，国内有研究从ASD(Autistic Spectrum Disorders,泛自闭症障碍症候群)儿童和正常儿童的EEG(Electroencephalogram,脑电波)信号中提取四种熵特征,并应用独立样本t检验分析组间差异。最后利用支持向量机学习算法为不同脑区的每种熵测量建立分类模型,分类准确率最高到84.55％。国外另一项研究通过离散傅里叶变换将脑电时域信号转换为频域信号，设计强化学习算法快速更新和重建卷积神经网络模型，利用最终的卷积神经网络模型进行分类，最高能够达到92.63％的识别精度。
然后，现有基于EEG信号的自闭症检测方法普遍是利用EEG信号的原始信号或者频谱特征，受限于受试者数量，普遍存在实验受试者数量少的状况，现 有技术大都是采用受试者在不同时间片段数据都作为独立样本进行重复利用，每次只利用部分片段的脑电数据进行分析，可能会遗漏较为重要的脑电信息，并且这种分析方式可能会造成数据污染，导致分析结果失真；或复用一个受试者数据增加为多个样本，导致样本之间的相似性太高。另外，现有技术中很少有研究利用EEG数据统计受试者的大脑功能连接特征以及自闭症分析。
发明内容
本申请提供了一种基于小样本数据增强的自闭症风险评估方法及装置，旨在至少在一定程度上解决现有技术中的上述技术问题之一。
为了解决上述问题，本申请提供了如下技术方案：
一种基于小样本数据增强的自闭症风险评估方法，包括：
采集受试者的多通道脑电信号；
利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；
利用生成对抗网络对所述大脑时序连通图进行数据增强处理，并将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。
本申请实施例采取的技术方案还包括：所述采集受试者的多通道脑电信号之后，还包括：
对所述多通道脑电信号进行预处理；所述预处理过程包括：使用抗混叠滤波器对所述多通道脑电信号进行滤波，以设定采样率对所有脑电信号进行数字化并放大；
使用基线校正算法对放大后的多通道脑电信号进行校正，并利用伪影检测算法对多通道脑电信号进行伪影去除，得到预处理后的多通道脑电信号。
本申请实施例采取的技术方案还包括：所述利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图，具体为：
通过皮尔森相关系数计算受试者一个脑电片段中每个通道与其他通道之间的相关性，得到一个设定大小的功能连接矩阵；
将所述功能连接矩阵的上三角矩阵展开成一维，并利用SelectKBest算法作为特征筛选算法对功能连接特征进行筛选，保留设定数量的主要功能连接特征；
将一个受试者所采集的设定数量的时间片段的脑电信号按照时间顺序组成设定大小的大脑功能连接矩阵，并根据灰度值分布将所述大脑功能连接矩阵转化为大脑时序连通图。
本申请实施例采取的技术方案还包括：所述按所述多通道脑电信号的时序信息构建大脑时序连通图之后，还包括：
将所述大脑时序连通图按照设定比例划分为训练集和验证集。
本申请实施例采取的技术方案还包括：所述利用生成对抗网络对所述大脑时序连通图进行数据增强处理，具体为：
所述生成对抗网络包括生成器和判别器，所述生成器包括3层反卷积层，将噪声矢量通过反卷积层映射到大脑时序连通图中，生成新的大脑时序连通图；所述判别器使用3层卷积层判别生成的大脑时序连通图真伪，并经过生成器与判别器迭代对抗设定次数后，将所述生成器生成的大脑时序连通图作为扩充的伪样本。
本申请实施例采取的技术方案还包括：所述将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果，具体为：
所述深度卷积网络与长短期记忆网络联合模型的深度卷积网络部分包括5个卷积层和4个池化层，通过卷积层学习输入数据的局部特征，所述卷积层由ReLU函数激活；所述长短期记忆网络包括隐藏层、循环网络层和全连接层，输入数据经过输入维数转换后，依次经过隐藏层的整形和两个循环网络层，然后经过两个全连接层，最后连接分类器输出二值分类结果，所述二值分类结果包括“自闭症”和“健康”两类。
本申请实施例采取的另一技术方案为：一种基于小样本数据增强的自闭症风险评估装置，包括：
信号采集模块：用于采集受试者的多通道脑电信号；
连通图构建模块：用于利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；
数据增强模块：用于利用生成对抗网络对所述大脑时序连通图进行数据增强处理；
风险评估模块：用于将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。
本申请实施例采取的技术方案还包括：所述连通图构建模块利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图， 具体为：
通过皮尔森相关系数计算受试者一个脑电片段中每个通道与其他通道之间的相关性，得到一个设定大小的功能连接矩阵；
将所述功能连接矩阵的上三角矩阵展开成一维，并利用Se l ectKBest算法作为特征筛选算法对功能连接特征进行筛选，保留设定数量的主要功能连接特征；
将一个受试者所采集的设定数量的时间片段的脑电信号按照时间顺序组成设定大小的大脑功能连接矩阵，并根据灰度值分布将所述大脑功能连接矩阵转化为大脑时序连通图。
本申请实施例采取的又一技术方案为：一种计算机设备，所述计算机设备包括处理器、与所述处理器耦接的存储器，其中，
所述存储器存储有用于实现所述基于小样本数据增强的自闭症风险评估方法的程序指令；
所述处理器用于执行所述存储器存储的所述程序指令以控制基于小样本数据增强的自闭症风险评估方法。
本申请实施例采取的又一技术方案为：一种存储介质，存储有处理器可运行的程序指令，所述程序指令用于执行所述基于小样本数据增强的自闭症风险评估方法。
相对于现有技术，本申请实施例产生的有益效果在于：本申请实施例的基于小样本数据增强的自闭症风险评估方法及装置利用脑电信号计算出大脑功能连接矩阵，并根据脑电信号的时间信息构建大脑时序连通图，利用生成对抗网络对大脑时序连通图进行样本扩充后，输入到深度卷积网络与长短期记忆网络联合模型进行分类，并由分类器输出自闭症风险评估结果。本申请实施例从 大脑功能连通性的角度考虑了自闭症患者与健康人群之间存在的差异性，同时考虑了大脑功能连接性与脑电信号所具有的时间信息，能更准确地对受试者进行自闭症风险评估。将每一个受试者整体时间的脑电信号作为一个样本也更加符合实际诊断场景。通过设计生成对抗网络模型进行样本扩充，有效解决了样本量普遍较少的问题。通过构建了深度卷积网络与长短期记忆网络联合模型进行自闭症风险评估，能够考虑更全面的脑电信号特征，进一步提高自闭症风险评估的精度。
附图说明
图1是本申请实施例的基于小样本数据增强的自闭症风险评估方法的流程图；
图2是本申请实施例构建的大脑时序连通图；
图3为本申请实施例的生成对抗网络的模型结构示意图；
图4为本申请实施例的深度卷积网络与长短期记忆网络联合模型结构示意图；
图5为本申请实施例的基于小样本数据增强的自闭症风险评估装置结构示意图；
图6为本申请实施例的计算机设备结构示意图；
图7为本申请实施例的存储介质的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅是本申请的一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请中的术语“第一”、“第二”、“第三”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”、“第三”的特征可以明示或者隐含地包括至少一个该特征。本申请的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。本申请实施例中所有方向性指示(诸如上、下、左、右、前、后……)仅用于解释在某一特定姿态(如附图所示)下各部件之间的相对位置关系、运动情况等，如果该特定姿态发生改变时，则该方向性指示也相应地随之改变。此外，术语“包括”和“具有”以及它们任何变形，意图在于覆盖不排他的包含。例如包含了一系列步骤或单元的过程、方法、系统、产品或计算机设备没有限定于已列出的步骤或单元，而是可选地还包括没有列出的步骤或单元，或可选地还包括对于这些过程、方法、产品或计算机设备固有的其它步骤或单元。
在本文中提及“实施例”意味着，结合实施例描述的特定特征、结构或特性可以包含在本申请的至少一个实施例中。在说明书中的各个位置出现该短语并不一定均是指相同的实施例，也不是与其它实施例互斥的独立的或备选的实施例。本领域技术人员显式地和隐式地理解的是，本文所描述的实施例可以与其它实施例相结合。
具体地，请参阅图1，是本申请实施例的基于小样本数据增强的自闭症风险评估方法的流程图。本申请实施例的基于小样本数据增强的自闭症风险评估方法包括以下步骤：
S100：采集受试者的多通道脑电信号；
本步骤中，采集的多通道脑电信号具有125个通道，具体通道数可根据实际应用场景进行设定。
S110：对多通道脑电信号进行预处理；
本步骤中，由于在脑电信号采集过程中不可避免会有噪声引入，因此在预处理阶段需对眼电信号等噪声进行消除处理。本申请实施例中，多通道脑电信 号的预处理过程包括：使用抗混叠滤波器对多通道脑电信号进行滤波，以设定采样率对所有脑电信号进行数字化并放大。最后使用基线校正算法对放大后的多通道脑电信号进行校正，并利用伪影检测算法对多通道脑电信号进行伪影去除，得到预处理后的多通道脑电信号。
S120：利用皮尔森相关系数(Pearson product-moment correlation coefficient)对预处理后的多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按时序构建大脑时序连通图；
本步骤中，通过皮尔森相关系数计算受试者一个脑电片段中每个通道与其他通道之间的相关性，从而得到一个125×125的功能连接矩阵。将功能连接矩阵的上三角矩阵展开成一维，然后利用SelectKBest算法作为特征筛选算法对功能连接特征进行筛选，保留设定数量的主要功能连接特征。假设保留的主要功能连接特征为400个，将一个受试者所采集的40个时间片段的脑电信号按照时间顺序组成400×40的大脑功能连接矩阵，然后根据灰度值分布将大脑功能连接矩阵转化为大脑时序连通图，并将所有大脑时序连通图按照设定比例划分为训练集和验证集，本申请实施例中，训练集的划分比例为80％，验证集的划分比例为20％，具体可根据实际应用场景进行设定。具体如图2所示，是本申请实施例构建的大脑时序连通图，其中(a)为自闭症患者的大脑时序连通图，(b)为健康对照组的大脑时序连通图；由图(2)可以看到自闭症患者与健康对照组的大脑时序连通图具有明细差异。可以理解，在本申请其他实施例中，还可以采用相位锁定值作为构建大脑时序连通图的连接指标。
S130：利用生成对抗网络(DCGAN)对训练集中的大脑时序连通图进行数据增强处理，生成扩充后的训练集；
本步骤中，由于自闭症受试者样本数量普遍不足，本申请实施例利用pytorch设计生成对抗网络(DCGAN)，将生成对抗网络作为一种数据增强的方法对大脑时序连通图进行样本量扩充。具体的，本申请实施例的生成对抗网络的模型结构如图3所示，生成对抗网络包括生成器和判别器两个部分，在生成 器部分由3层反卷积层组成，将100x1噪声矢量通过反卷积映射到400x40x1的大脑时序连通图中，生成新的大脑时序连通图；在判别器部分用3层卷积层判别生成的大脑时序连通图真伪。经过生成器与判别器迭代对抗100次后，将生成器生成的大脑时序连通图作为扩充的伪样本。本申请实施例中，仅针对训练集进行数据增强处理，不会影响验证集分类实验的结果。
S140：将扩充后的训练集输入到构建的深度卷积网络与长短期记忆网络联合模型(CNN-LSTM)中进行分类，并由分类器输出受试者的自闭症风险评估结果；
本步骤中，利用pytorch搭建深度卷积网络与长短期记忆网络联合模型，深度卷积网络与长短期记忆网络联合模型结构如图4所示。其中，深度卷积网络部分包括5个卷积层和4个池化层，通过大小为3X3和2x2的卷积层学习输入数据的局部特征，卷积层由ReLU函数激活。长短期记忆网络的隐藏层维度大小设置为64，循环网络的层数设置为2层，全连接层设置为2层。由于构建的大脑时序连通图是灰度值图像，因此起始的输入数据尺寸为400×40×1。经过输入维数转换后，图像尺寸变为400×1×1，随后依次经过隐藏层的整形和两个循环网络层后图像尺寸变为64×1×1，然后经过两个全连接层最后连接分类器输出最终的二值分类结果，利用5倍交叉验证技术对深度卷积网络与长短期记忆网络联合模型进行实验验证，得到最优的深度卷积网络与长短期记忆网络联合模型。所述二值分类结果包括“自闭症”和“健康”两类结果，如果分类结果为“自闭症”，表示受试者患有自闭症的风险较高，需要进行下一步更加专业的诊断。可以理解，本申请实施例仅以基于小样本数据增强的自闭症风险评估为例，本申请实施例同样适用于其他同样存在样本量较小的医学图像以及具有时空特征图像的评估分类。
为了验证本申请实施例的可行性和有效性，通过实验对本申请实施例的识别精度进行了评估，本发明通过实验证明方法可行，利用自闭症患者与健康对照组数据进行测试可达到的识别精度如下表1所示：
表1自闭症患者与健康对照组数据进行测试的识别精度
实验结果表明，本申请实施例可以快速准确地对受试者患有自闭症的风险进行评估，其中在静息态脑电信号数据集中，利用本申请实施例的深度卷积网络与长短期记忆网络联合模型进行测试得到的识别精度最高可达81.08％。
基于上述，本申请实施例的基于小样本数据增强的自闭症风险评估方法利用脑电信号计算出大脑功能连接矩阵，并根据脑电信号的时间信息构建大脑时序连通图，利用生成对抗网络(DCGAN)对大脑时序连通图进行样本扩充后，输入到深度卷积网络与长短期记忆网络联合模型(CNN-LSTM)进行分类，并由分类器输出自闭症风险评估结果。本申请实施例从大脑功能连通性的角度考虑了自闭症患者与健康人群之间存在的差异性，同时考虑了大脑功能连接性与脑电信号所具有的时间信息，能更准确地对受试者进行自闭症风险评估。将每一个受试者整体时间的脑电信号作为一个样本也更加符合实际诊断场景。通过设计生成对抗网络模型进行样本扩充，有效解决了样本量普遍较少的问题。通过构建了深度卷积网络与长短期记忆网络联合模型进行自闭症风险评估，能够考虑更全面的脑电信号特征，进一步提高自闭症风险评估的精度。
请参阅图5，为本申请实施例的基于小样本数据增强的自闭症风险评估装置结构示意图。本申请实施例的基于小样本数据增强的自闭症风险评估装置40包括：
信号采集模块41：用于采集受试者的多通道脑电信号；
连通图构建模块42：用于利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；
数据增强模块43：用于利用生成对抗网络对所述大脑时序连通图进行数据增强处理；
风险评估模块44：用于利用生成对抗网络对所述大脑时序连通图进行数据增强处理，并将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。
需要说明的是，上述装置/单元之间的信息交互、执行过程等内容，由于与本申请方法实施例基于同一构思，其具体功能及带来的技术效果，具体可参见方法实施例部分，此处不再赘述。
本申请实施例提供的装置可以应用在前述方法实施例中，详情参见上述方法实施例的描述，在此不再赘述。
请参阅图6，为本申请实施例的计算机设备结构示意图。该计算机设备50包括：
存储有可执行程序指令的存储器51；
与存储器51连接的处理器52；
处理器52用于调用存储器51中存储的可执行程序指令并执行以下步骤：采集受试者的多通道脑电信号；利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；利用生成对抗网络对所述大脑时序连通图进行数据增强处理，并将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。
其中，处理器52还可以称为CPU(Central Processing Unit，中央处理单元)。处理器52可能是一种集成电路芯片，具有信号的处理能力。处理器52还可以是通用处理器、数字信号处理器(DSP)、专用集成电路(ASIC)、现成可编程门阵列(FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器 件、分立硬件组件。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
请参阅图7，图7为本申请实施例的存储介质的结构示意图。本申请实施例的存储介质存储有能够实现以下步骤的程序指令61：采集受试者的多通道脑电信号；利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；利用生成对抗网络对所述大脑时序连通图进行数据增强处理，并将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。其中，该程序指令61可以以软件产品的形式存储在上述存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络计算机设备等)或处理器(processor)执行本申请各个实施方式方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序指令的介质，或者是计算机、服务器、手机、平板等终端计算机设备。其中，服务器可以是独立的服务器，也可以是提供云服务、云数据库、云计算、云函数、云存储、网络服务、云通信、中间件服务、域名服务、安全服务、内容分发网络(Content Delivery Network，CDN)、以及大数据和人工智能平台等基础云计算服务的云服务器。
在本申请所提供的几个实施例中，应该理解到，所揭露的系统，装置和方法，可以通过其它的方式实现。例如，以上所描述的系统实施例仅仅是示意性的，例如，单元的划分，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，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
另外，在本申请各个实施例中的各功能单元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，也可以采用软件功能单元的形式实现。以上仅为本申请的实施方式，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (10)

  1. 一种基于小样本数据增强的自闭症风险评估方法，其特征在于，包括：
    采集受试者的多通道脑电信号；
    利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；
    利用生成对抗网络对所述大脑时序连通图进行数据增强处理，并将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。
  2. 根据权利要求1所述的基于小样本数据增强的自闭症风险评估方法，其特征在于，所述采集受试者的多通道脑电信号之后，还包括：
    对所述多通道脑电信号进行预处理；所述预处理过程包括：使用抗混叠滤波器对所述多通道脑电信号进行滤波，以设定采样率对所有脑电信号进行数字化并放大；
    使用基线校正算法对放大后的多通道脑电信号进行校正，并利用伪影检测算法对多通道脑电信号进行伪影去除，得到预处理后的多通道脑电信号。
  3. 根据权利要求2所述的基于小样本数据增强的自闭症风险评估方法，其特征在于，所述利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图，具体为：
    通过皮尔森相关系数计算受试者一个脑电片段中每个通道与其他通道之间的相关性，得到一个设定大小的功能连接矩阵；
    将所述功能连接矩阵的上三角矩阵展开成一维，并利用SelectKBest算法作为特征筛选算法对功能连接特征进行筛选，保留设定数量的主要功能连接特征；
    将一个受试者所采集的设定数量的时间片段的脑电信号按照时间顺序组成设定大小的大脑功能连接矩阵，并根据灰度值分布将所述大脑功能连接矩阵转化为大脑时序连通图。
  4. 根据权利要求1至3任一项所述的基于小样本数据增强的自闭症风险评估方法，其特征在于，所述按所述多通道脑电信号的时序信息构建大脑时序连通图之后，还包括：
    将所述大脑时序连通图按照设定比例划分为训练集和验证集。
  5. 根据权利要求1至4任一项所述的基于小样本数据增强的自闭症风险评估方法，其特征在于，所述利用生成对抗网络对所述大脑时序连通图进行数据增强处理，具体为：
    所述生成对抗网络包括生成器和判别器，所述生成器包括3层反卷积层，将噪声矢量通过反卷积层映射到大脑时序连通图中，生成新的大脑时序连通图；所述判别器使用3层卷积层判别生成的大脑时序连通图真伪，并经过生成器与判别器迭代对抗设定次数后，将所述生成器生成的大脑时序连通图作为扩充的伪样本。
  6. 根据权利要求5所述的基于小样本数据增强的自闭症风险评估方法，其特征在于，所述将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果，具体为：
    所述深度卷积网络与长短期记忆网络联合模型的深度卷积网络部分包括5个卷积层和4个池化层，通过卷积层学习输入数据的局部特征，所述卷积层由 ReLU函数激活；所述长短期记忆网络包括隐藏层、循环网络层和全连接层，输入数据经过输入维数转换后，依次经过隐藏层的整形和两个循环网络层，然后经过两个全连接层，最后连接分类器输出二值分类结果，所述二值分类结果包括“自闭症”和“健康”两类。
  7. 一种基于小样本数据增强的自闭症风险评估装置，其特征在于，包括：
    信号采集模块：用于采集受试者的多通道脑电信号；
    连通图构建模块：用于利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图；
    数据增强模块：用于利用生成对抗网络对所述大脑时序连通图进行数据增强处理；
    风险评估模块：用于将所述数据增强处理后的大脑时序连通图输入到训练好的深度卷积网络与长短期记忆网络联合模型中进行分类，并由分类器输出受试者的自闭症风险评估结果。
  8. 根据权利要求7所述的基于小样本数据增强的自闭症风险评估装置，其特征在于，所述连通图构建模块利用皮尔森相关系数对对所述多通道脑电信号进行计算，得到各个通道之间的相关性，生成大脑功能连接矩阵，并按所述多通道脑电信号的时序信息构建大脑时序连通图，具体为：
    通过皮尔森相关系数计算受试者一个脑电片段中每个通道与其他通道之间的相关性，得到一个设定大小的功能连接矩阵；
    将所述功能连接矩阵的上三角矩阵展开成一维，并利用SelectKBest算法作为特征筛选算法对功能连接特征进行筛选，保留设定数量的主要功能连接特征；
    将一个受试者所采集的设定数量的时间片段的脑电信号按照时间顺序组成 设定大小的大脑功能连接矩阵，并根据灰度值分布将所述大脑功能连接矩阵转化为大脑时序连通图。
  9. 一种计算机设备，其特征在于，所述计算机设备包括处理器、与所述处理器耦接的存储器，其中，
    所述存储器存储有用于实现权利要求1-6任一项所述的基于小样本数据增强的自闭症风险评估方法的程序指令；
    所述处理器用于执行所述存储器存储的所述程序指令以控制基于小样本数据增强的自闭症风险评估方法。
  10. 一种存储介质，其特征在于，存储有处理器可运行的程序指令，所述程序指令用于执行权利要求1至6任一项所述基于小样本数据增强的自闭症风险评估方法。
PCT/CN2023/139971 2023-12-19 2023-12-19 一种基于小样本数据增强的自闭症风险评估方法及装置 Pending WO2025129461A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2023/139971 WO2025129461A1 (zh) 2023-12-19 2023-12-19 一种基于小样本数据增强的自闭症风险评估方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2023/139971 WO2025129461A1 (zh) 2023-12-19 2023-12-19 一种基于小样本数据增强的自闭症风险评估方法及装置

Publications (1)

Publication Number Publication Date
WO2025129461A1 true WO2025129461A1 (zh) 2025-06-26

Family

ID=96136204

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/139971 Pending WO2025129461A1 (zh) 2023-12-19 2023-12-19 一种基于小样本数据增强的自闭症风险评估方法及装置

Country Status (1)

Country Link
WO (1) WO2025129461A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN121190464A (zh) * 2025-11-07 2025-12-23 上海市普陀区中心医院 一种应对耳鼻喉病变图像的智能分类与识别方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110301431A1 (en) * 2010-06-05 2011-12-08 The Board Of Trustees Of The Leland Stanford Junior University Methods of classifying cognitive states and traits and applications thereof
CN112001306A (zh) * 2020-08-21 2020-11-27 西安交通大学 基于深度卷积对抗生成神经网络的脑电信号解码方法
CN113768465A (zh) * 2021-09-15 2021-12-10 安徽师范大学 基于卷积网络和长短期记忆网络的脑网络特征提取方法
CN114241240A (zh) * 2021-12-15 2022-03-25 中国科学院深圳先进技术研究院 脑部图像的分类方法、装置、电子设备及存储介质
CN115414041A (zh) * 2022-08-26 2022-12-02 深圳先进技术研究院 基于脑电数据的自闭症评估装置、方法、终端设备和介质
CN116313090A (zh) * 2023-03-16 2023-06-23 上海外国语大学 一种基于静息态脑电数据睡眠障碍风险评估方法和系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110301431A1 (en) * 2010-06-05 2011-12-08 The Board Of Trustees Of The Leland Stanford Junior University Methods of classifying cognitive states and traits and applications thereof
CN112001306A (zh) * 2020-08-21 2020-11-27 西安交通大学 基于深度卷积对抗生成神经网络的脑电信号解码方法
CN113768465A (zh) * 2021-09-15 2021-12-10 安徽师范大学 基于卷积网络和长短期记忆网络的脑网络特征提取方法
CN114241240A (zh) * 2021-12-15 2022-03-25 中国科学院深圳先进技术研究院 脑部图像的分类方法、装置、电子设备及存储介质
CN115414041A (zh) * 2022-08-26 2022-12-02 深圳先进技术研究院 基于脑电数据的自闭症评估装置、方法、终端设备和介质
CN116313090A (zh) * 2023-03-16 2023-06-23 上海外国语大学 一种基于静息态脑电数据睡眠障碍风险评估方法和系统

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN121190464A (zh) * 2025-11-07 2025-12-23 上海市普陀区中心医院 一种应对耳鼻喉病变图像的智能分类与识别方法

Similar Documents

Publication Publication Date Title
CN113768519B (zh) 基于深度学习和静息态脑电数据分析病人意识水平的方法
Ganesan et al. Computer-aided diabetic retinopathy detection using trace transforms on digital fundus images
CN111862009B (zh) 一种眼底oct图像的分类方法及计算机可读存储介质
CN111223160A (zh) 图像重建方法、装置、设备、系统及计算机可读存储介质
CN111481192A (zh) 一种基于改进U-Net的心电信号R波检测方法
Xu et al. Interpatient ECG arrhythmia detection by residual attention CNN
CN118873153B (zh) 基于分组余弦卷积网络的脑电异常检测及定位方法及系统
CN113076878A (zh) 基于注意力机制卷积网络结构的体质辨识方法
WO2024244880A1 (zh) 超声病灶标准切面选取方法及系统
CN119989089A (zh) 一种基于gnn的多模态脑成像自闭症谱系障碍检测方法
CN117455774B (zh) 一种基于差分输出的图像重建方法及系统
WO2025129461A1 (zh) 一种基于小样本数据增强的自闭症风险评估方法及装置
CN115067963A (zh) 心电信号处理方法、装置、设备及存储介质
CN118708937A (zh) 一种基于脑电信号的跨任务认知负荷识别方法及系统
CN116889405B (zh) 一种注意缺陷多动障碍辅助诊断方法及系统
CN111784637B (zh) 一种预后特征可视化方法、系统、设备和存储介质
Tao et al. DDLA: a double deep latent autoencoder for diabetic retinopathy diagnose based on continuous glucose sensors
CN115375628B (zh) 一种基于as-octa的图像噪声去除方法及系统
CN116881762A (zh) 一种基于动态脑网络特征的情绪识别方法
CN111798455A (zh) 一种基于全卷积密集空洞网络的甲状腺结节实时分割方法
CN109685143A (zh) 一种甲状腺锝扫图像的识别模型构建方法及装置
CN113506274A (zh) 基于视觉显著性差异图的用于人体认知状况的检测系统
CN117352135A (zh) 基于深度学习的糖网识别方法和装置、设备及存储介质
Shi et al. Sleep stage classification with multi-resolution temporal modeling and adaptive feature fusion
CN117952930A (zh) 一种脑部精神疾病的图像处理方法和装置

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: 23961806

Country of ref document: EP

Kind code of ref document: A1