CN113080847A - Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph - Google Patents

Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph Download PDF

Info

Publication number
CN113080847A
CN113080847A CN202110286243.6A CN202110286243A CN113080847A CN 113080847 A CN113080847 A CN 113080847A CN 202110286243 A CN202110286243 A CN 202110286243A CN 113080847 A CN113080847 A CN 113080847A
Authority
CN
China
Prior art keywords
cognitive impairment
mild cognitive
matrix
graph
representing
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.)
Granted
Application number
CN202110286243.6A
Other languages
Chinese (zh)
Other versions
CN113080847B (en
Inventor
周宇涛
安兴伟
明东
郭恒言
赵津笛
钟文潇
柯余峰
刘爽
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.)
Tianjin University
Original Assignee
Tianjin University
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 Tianjin University filed Critical Tianjin University
Priority to CN202110286243.6A priority Critical patent/CN113080847B/en
Publication of CN113080847A publication Critical patent/CN113080847A/en
Application granted granted Critical
Publication of CN113080847B publication Critical patent/CN113080847B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • A61B5/40—Detecting, measuring or recording for evaluating the nervous system
    • A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
    • A61B5/4088—Diagnosing of monitoring cognitive diseases, e.g. Alzheimer, prion diseases or dementia
    • 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
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Neurology (AREA)
  • General Health & Medical Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Molecular Biology (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Physiology (AREA)
  • Psychiatry (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Evolutionary Computation (AREA)
  • Hospice & Palliative Care (AREA)
  • Developmental Disabilities (AREA)
  • Child & Adolescent Psychology (AREA)
  • Neurosurgery (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Signal Processing (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Psychology (AREA)
  • Radiology & Medical Imaging (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

The invention discloses a device for diagnosing mild cognitive impairment based on a bidirectional long-short term memory model of a graph, which comprises: the data preprocessing module is used for carrying out binarization processing on the adjacent matrix through a given threshold value, and the functional connection matrix after binarization processing can reflect abnormal functional connections between brain areas of the brains of patients with mild cognitive impairment; the characteristic extraction and modeling module is used for dividing the whole brain into different brain areas by using an AAL brain template according to the preprocessed data, and extracting BOLD signals of the different brain areas respectively; a training model and prediction result module for sequentially obtaining H after forward GC-LSTM cyclic processing1To HTSequentially obtaining H after reverse GC-LSTM cyclic treatment1' to HT', forward high state H to be at time ttAnd reverse hidden state H'T+1‑tSpliced to form ytAnd performing classification prediction on the currently input tested object through the softmax layer.

Description

Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph
Technical Field
The invention relates to the field of artificial neural network diagnosis, in particular to a device for diagnosing mild cognitive impairment based on a bidirectional long-short term memory model of a graph.
Background
Alzheimer's Disease (AD) is an irreversible, degenerative neurological disorder that affects primarily the elderly, and dementia is a common major form. MCI (mild cognitive impairment) is a transitional stage of normal aging and Alzheimer disease, prevents the disease from further worsening by accurately diagnosing mild cognitive impairment and implementing effective human intervention means, and can delay or even prevent the possibility of converting into the Alzheimer disease. In the early stage of mild cognitive impairment (eMCI), it is difficult to observe subtle structural changes in the brain by structural magnetic resonance imaging (srmri), while cognitive ability is not significantly degraded, but functional changes occur significantly for specific brain regions. Since fMRI (functional magnetic resonance imaging) is characterized by non-invasiveness, no radiation, high spatial resolution, and high effectiveness in detecting neurodegenerative diseases, it is a reliable means to study mild cognitive impairment using fMRI.
At present, the solution indexes for analyzing mild cognitive impairment by using fMRI mainly include low-frequency oscillation Amplitude (ALFF), local coherence (ReHo), and Functional Connectivity (FC). However, these indexes have limitations to some extent, and deep learning techniques have been widely used in various fields, and an attempt to identify MCI using a deep learning algorithm can greatly improve classification performance. Therefore, the diagnosis performance of the MCI can be greatly improved by combining the deep learning method.
Disclosure of Invention
The invention provides a device for diagnosing mild cognitive impairment based on a bidirectional long and short term memory model of a graph, which constructs a dynamic function connection matrix for BOLD (blood oxygen level dependence) signals in fMRI data by using a sliding window, and classifies and identifies a tested object by means of the bidirectional long and short term memory artificial neural network model of the graph, which is described in detail as follows:
an apparatus for diagnosing mild cognitive impairment based on a graph-based two-way long-short term memory model, the apparatus comprising:
the data preprocessing module is used for carrying out binarization processing on the adjacent matrix through a given threshold value, and the functional connection matrix after binarization processing can reflect abnormal functional connections between brain areas of the brains of patients with mild cognitive impairment;
the characteristic extraction and modeling module is used for dividing the whole brain into different brain areas by using an AAL brain template according to the preprocessed data, and extracting BOLD signals of the different brain areas respectively;
a training model and prediction result module for sequentially obtaining H after forward GC-LSTM cyclic processing1To HTH 'is obtained after reverse GC-LSTM cyclic treatment'1To H'TForward high state H to be at time ttAnd reverse hidden state H'T+1-tSpliced to form ytClassifying and predicting the currently input tested object through the softmax layer;
and a module for diagnosing whether the subject suffers from mild cognitive impairment, wherein the module is used for diagnosing whether the subject suffers from mild cognitive impairment.
Wherein, the graph convolution long-term and short-term memory unit is as follows:
ht-1indicating the hidden state, c, at time t-1t-1Indicates the cell state, A, at time t-1tRepresenting an adjacency matrix, XtRepresenting a feature matrix, + representing matrix addition, + representing matrix multiplication, # representing a graph convolution operation, # representing a nonlinear activation function, tanh representing a hyperbolic tangent function,hthigh state, c at time ttIndicating the cell state at time t.
The technical scheme provided by the invention has the beneficial effects that:
1. the method can effectively improve the accuracy and systematicness of the identification of the mild cognitive impairment, provides a plurality of targeted auxiliary diagnosis suggestions for clinicians, further provides effective treatment for patients, reduces the risk of the MCI patients transforming into AD, and obtains considerable social benefit and economic benefit;
2. the method can solve the problems that the traditional machine learning has low accuracy rate for detecting and identifying mild cognitive impairment and doctors cannot diagnose a large amount of data, and can realize more accurate diagnosis by combining dynamic function connection with a bidirectional long-term and short-term memory algorithm based on a graph model;
3. the invention can more fully utilize the acquired data, has stronger robustness and can also migrate to other brain functional diseases.
Drawings
FIG. 1 is a schematic diagram of an apparatus for diagnosing mild cognitive impairment based on a graph-based two-way long-short term memory model;
FIG. 2 is a schematic diagram of the structural framework of the Bi-GC-LSTM model;
FIG. 3 is a schematic diagram of the internal structure of GC-LSTM.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention are described in further detail below.
Example 1
An apparatus for diagnosing mild cognitive impairment based on a graph-based two-way long-short term memory model, see fig. 1, the apparatus comprising:
the embodiment of the invention provides a device for predicting mild cognitive impairment by using functional magnetic resonance imaging (fMRI), dynamic functional connectivity (dFC) and a bidirectional neural long-short term memory artificial neural network (Bi-GC-LSTM). Among them, fMRI can evaluate blood oxygen level-dependent (BOLD-level-dependent, BOLD) signals in the brain and reveal activation and inactivation status of different brain regions, and can analyze the possible differences between Normal Control (NC) and Mild Cognitive Impairment (MCI) patients in brain functional connectivity.
The embodiment of the invention carries out diagnosis and identification by the following steps: the device comprises a data preprocessing module, a feature extraction and modeling module, a training model and prediction result module and a module for diagnosing whether the tested object suffers from mild cognitive impairment. The technical process comprises the following steps:
firstly, preprocessing collected fMRI data of a normal control group and a mild cognitive impairment patient; secondly, dividing the brain into different brain areas by using a priori templates of the brain, and respectively carrying out slicing processing on the whole time sequence of each brain area by using a sliding time window strategy; then, respectively constructing a functional connection matrix among the brain areas in different time windows for each tested object; finally, the Bi-GC-LSTM model can be used for diagnosing whether the test object suffers from MCI and providing useful diagnosis basis for doctors.
Example 2
A flow chart of an apparatus for diagnosing mild cognitive impairment based on dynamic functional connectivity of functional magnetic resonance imaging is shown in figure 1.
1. First, collect fMRI data of the subject;
2. then, extracting the BOLD signals of 116 brain areas by using an AAL (automatic anatomical labeling) template through preprocessing, dividing the BOLD signals into a plurality of equal-length sub-segments by using a sliding window strategy, calculating all different brain areas by using a Pearson correlation coefficient, and generating a functional connection matrix in a corresponding time window. The strength of functional connectivity between different brain regions of the brain reflects the degree of correlation between brain regions. The adjacency matrix is binarized by a given threshold τ. The functional connection matrix after binarization processing can reflect the abnormalities of functional connection between brain areas of the brains of patients with mild cognitive impairment.
The embodiment of the invention needs to collect the data of the rs-fMRI of the testee with normal aging and mild cognitive impairment. Compared with sMRI, fMRI can observe the functional change of different brain areas, and simultaneously compared with the functional magnetic resonance technology in a task state, the functional resonance imaging technology in a resting state can effectively reduce the experimental time without setting a complex experiment, thereby reducing the requirement on the tested object and simplifying the operation process.
1) Data pre-processing
The obtained original data is subjected to a series of preprocessing steps, and the main purpose is to reduce physiological noise and machine noise generated in the data acquisition process, improve the signal-to-noise ratio of the data and facilitate subsequent analysis. The pretreatment steps of the embodiment of the invention sequentially comprise: removing data of the first 5 time points in the original data, correcting a time layer, correcting a head movement, standardizing a space, smoothing the space, removing linear drift, filtering, removing covariates and the like.
2) Feature extraction
The data after preprocessing are divided into 116 different brain areas by using an AAL brain template, and BOLD signals of the different brain areas are respectively extracted. The BOLD signal of each brain region is then divided into K equal-length non-overlapping sub-segments using a sliding window strategy. The sub-segment partition of the sliding window can be expressed as:
Figure BDA0002980601240000041
where T represents the number of time points of fMRI data acquisition, L represents the length of each sub-segment, S represents the step size of the sliding, and K represents the number of sub-segments.
Calculating a functional connection between an ith brain region and a jth brain region in an ith time window using Pearson's correlation coefficients
Figure BDA0002980601240000042
Obtaining a functional connection matrix C in the l time windowlFor each time by Fisher Z transformThe functional connection matrix within the window is normalized.
To function connection matrix ClThresholding is carried out, and the adjacency matrix of the mth tested object in the ith time window is obtained by giving a threshold value tau
Figure BDA0002980601240000043
The thresholding operation can be expressed as
Figure BDA0002980601240000044
Wherein the content of the first and second substances,
Figure BDA0002980601240000045
indicating that the tested brain area i is functionally connected with the brain area j in the ith time window,
Figure BDA0002980601240000046
it indicates whether there is a connection between the brain regions i and j in the ith time window, 1 indicates that there is a connection between the brain regions, and 0 indicates that there is no connection between the brain regions.
Connecting matrix C due to original functionlIs a symmetric matrix, so C remainslThe elements of the upper and lower triangles and convert them into line vectors
Figure BDA0002980601240000047
For generating the feature matrix X.
3. Construction of Bi-GC-LSTM neural network model
The following describes the processing flow of the Bi-GC-LSTM model of the present invention in detail, and the framework of the model structure is shown in FIG. 2, first by initializing hidden state h0And cell state c0Then at t0Receives the output h from the last unit at the momentt-1，ct-1And the input feature matrix X of the current time point ttAnd the adjacent matrix AtSequentially obtaining H after forward GC-LSTM (graph convolution long short term memory unit) cyclic processing1To HTThen sequentially obtaining the product after reverse GC-LSTM cyclic treatmentTo H'1To H'TForward high state H to be at time ttAnd reverse hidden state H'T+1-tSpliced to form ytThen, classification prediction is carried out on the currently input tested object through the softmax layer.
Wherein H1To HTRepresents a hidden state, H ', of the output at each time t in the forward direction'1To H'TRepresenting the back propagation of the hidden state output at each time instant t.
FIG. 3 shows a GC-LSTM structure in which h ist-1Indicating the hidden state, c, at time t-1 (i.e., the previous time)t-1Indicates the cell state, A, at time t-1tRepresenting an adjacency matrix, XtRepresenting a feature matrix, + representing matrix addition, + representing matrix multiplication, + representing a graph convolution operation, [ sigma ] representing a nonlinear activation function, tanh representing a hyperbolic tangent function, htHigh state, c representing time t (i.e., current time)tIndicating the cell state at time t.
Therein, a graph convolution neural network definition undirected graph G (V, E) is used to describe the connection between 116 brain regions. v. ofie.V denotes the ith brain region, eijE represents the functional connection strength between the ith and jth brain regions. The graph neural network can comprehensively consider the influence of the relation between nodes and the connection of edges on the classification prediction of the tested object by aggregating the adjacent nodes of one node and combining the structure of the graph. The graph convolution network can be expressed as:
Hl+1＝f(Hl,A) (3)
Figure BDA0002980601240000051
Figure BDA0002980601240000052
where H denotes the characteristics of each layer, for an input layer H0Where X, I denotes an identity matrix, a denotes an adjacency matrix,
Figure BDA0002980601240000053
is that
Figure BDA0002980601240000054
A degree matrix of (a), a represents a nonlinear activation function, W represents a weight matrix,
Figure BDA0002980601240000055
to add the adjacency matrix after self-rotation, f (.) is the feature extraction function. Due to the respective incoming corresponding feature matrix X at time ttAnd an adjacency matrix AtIn order to comprehensively consider the connection relationship between different brain regions, the X is transmittedtAnd AtPerforming a graph convolution operation and assigning a weight matrix W1While also converting Xt、AtAnd h of the highest state at the previous momentt-1Respectively with a weight matrix W2、W3、W4Matrix multiplication is carried out, a bias b is added, and f is obtained through nonlinear activation function and hyperbolic tangent function processing respectivelyt，it，otAnd gtC is generated by updating the cell state and the hidden state at time t by the equations (10) and (11), respectivelytAnd htAs input for the next moment.
Figure BDA0002980601240000056
Figure BDA0002980601240000057
Figure BDA0002980601240000058
Figure BDA0002980601240000061
ct＝ft·ct-1+it·gt (10)
ht＝ot·tanh(ct) (11)
Wherein, w11，w12……w44As a weight matrix, b1……b4Is an offset amount, ftTo forget the door, itTo input gate, gtAs candidate memory cells, otIs an output gate.
Because the human brain continuously forms higher level cognitive and guidance state transitions through contextual background information during the process of processing information, rather than generating a single output at the end of a scan.
To effectively solve this problem, the embodiment of the present invention generates Y ═ Y by combining the hidden states output at all times1,y2…yT]The output of each cell is regarded as equally important information, then the whole cell is classified and predicted through softmax, and the final judgment result is output, so that the Bi-GC-LSTM can reasonably utilize two independent effective models of the information before and after the hidden layer in the opposite directions.
The invention provides a device for diagnosing mild cognitive impairment by dynamic functional connectivity of functional magnetic resonance imaging, which is characterized in that a series of preprocessing operations such as time layer correction, head motion correction, normalization, spatial smoothing, filtering and the like are carried out on original fMRI data, then BOLD signals of different brain areas are respectively extracted according to a brain template, then a dynamic functional connectivity matrix is constructed by using a sliding window strategy, and an adjacency matrix and a characteristic matrix are respectively input into a model, so that the model can fully utilize a connection structure between different brain areas and characteristic information of the brain areas, and thus, the mild cognitive impairment is accurately and objectively identified.
The method effectively improves the accuracy and the simplicity of recognizing the mild cognitive impairment and obtains considerable social and economic benefits. The preferred embodiment is intended for patent assignment, technology collaboration or product development. A system developed based on this technique can be applied to assist clinicians in diagnosing mild cognitive impairment.
In the embodiment of the present invention, except for the specific description of the model of each device, the model of other devices is not limited, as long as the device can perform the above functions.
Those skilled in the art will appreciate that the drawings are only schematic illustrations of preferred embodiments, and the above-described embodiments of the present invention are merely provided for description and do not represent the merits of the embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (3)

1. An apparatus for diagnosing mild cognitive impairment based on a graph-based two-way long-short term memory model, the apparatus comprising:
the data preprocessing module is used for carrying out binarization processing on the adjacent matrix through a given threshold value, and the functional connection matrix after binarization processing can reflect abnormal functional connections between brain areas of the brains of patients with mild cognitive impairment;
the characteristic extraction and modeling module is used for dividing the whole brain into different brain areas by using an AAL brain template according to the preprocessed data, and extracting BOLD signals of the different brain areas respectively;
a training model and prediction result module for sequentially obtaining H after forward GC-LSTM cyclic processing1To HTSequentially obtaining H after reverse GC-LSTM cyclic treatment1'to H'TForward high steady state H at time ttAnd reverse hidden stateH'T+1-tSpliced to form ytClassifying and predicting the currently input tested object through the softmax layer;
and a module for diagnosing whether the subject suffers from mild cognitive impairment, wherein the module is used for diagnosing whether the subject suffers from mild cognitive impairment.
2. The apparatus for diagnosing mild cognitive impairment based on the graph-based two-way long-short term memory model according to claim 1, wherein the graph convolution long-short term memory unit is:
ht-1indicating the hidden state, c, at time t-1t-1Indicates the cell state, A, at time t-1tRepresenting an adjacency matrix, XtRepresenting a feature matrix, + representing matrix addition, + representing matrix multiplication, + representing a graph convolution operation, [ sigma ] representing a nonlinear activation function, tanh representing a hyperbolic tangent function, htHigh state, c at time ttIndicating the cell state at time t.
3. The device for diagnosing mild cognitive impairment according to the graph-based two-way long-short term memory model of claim 1,
ct＝ft·ct-1+it·gt
ht＝ot·tanh(ct)
Figure FDA0002980601230000011
Figure FDA0002980601230000012
Figure FDA0002980601230000013
Figure FDA0002980601230000014
wherein, b1......b4Is an offset amount, ftTo forget the door, itTo input gate, gtAs candidate memory cells, otIs an output gate.
CN202110286243.6A 2021-03-17 2021-03-17 Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph Active CN113080847B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110286243.6A CN113080847B (en) 2021-03-17 2021-03-17 Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110286243.6A CN113080847B (en) 2021-03-17 2021-03-17 Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph

Publications (2)

Publication Number Publication Date
CN113080847A true CN113080847A (en) 2021-07-09
CN113080847B CN113080847B (en) 2022-11-29

Family

ID=76668693

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110286243.6A Active CN113080847B (en) 2021-03-17 2021-03-17 Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph

Country Status (1)

Country Link
CN (1) CN113080847B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115222688A (en) * 2022-07-12 2022-10-21 广东技术师范大学 Medical image classification method based on graph network time sequence

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005099576A2 (en) * 2004-04-13 2005-10-27 Cambridge University Technical Services Limited Methods for assessing neurocognitive disorders
CN110060773A (en) * 2019-04-22 2019-07-26 东华大学 Alzheimer's disease progression of the disease forecasting system based on two-way LSTM
CN110189302A (en) * 2019-05-07 2019-08-30 上海联影智能医疗科技有限公司 Brain image analysis method, computer equipment and readable storage medium storing program for executing
CN111009324A (en) * 2019-11-25 2020-04-14 东北大学 Mild cognitive impairment auxiliary diagnosis system and method based on brain network multi-feature analysis
US20200126221A1 (en) * 2018-10-19 2020-04-23 University Of Louisville Research Foundation, Inc. Computer aided diagnosis system for mild cognitive impairment
CN111657935A (en) * 2020-05-11 2020-09-15 浙江大学 Epilepsia electroencephalogram recognition system based on hierarchical graph convolutional neural network, terminal and storage medium
CN111985245A (en) * 2020-08-21 2020-11-24 江南大学 Attention cycle gating graph convolution network-based relation extraction method and system
CN112163426A (en) * 2020-09-30 2021-01-01 中国矿业大学 Relationship extraction method based on combination of attention mechanism and graph long-time memory neural network

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005099576A2 (en) * 2004-04-13 2005-10-27 Cambridge University Technical Services Limited Methods for assessing neurocognitive disorders
US20200126221A1 (en) * 2018-10-19 2020-04-23 University Of Louisville Research Foundation, Inc. Computer aided diagnosis system for mild cognitive impairment
CN110060773A (en) * 2019-04-22 2019-07-26 东华大学 Alzheimer's disease progression of the disease forecasting system based on two-way LSTM
CN110189302A (en) * 2019-05-07 2019-08-30 上海联影智能医疗科技有限公司 Brain image analysis method, computer equipment and readable storage medium storing program for executing
CN111009324A (en) * 2019-11-25 2020-04-14 东北大学 Mild cognitive impairment auxiliary diagnosis system and method based on brain network multi-feature analysis
CN111657935A (en) * 2020-05-11 2020-09-15 浙江大学 Epilepsia electroencephalogram recognition system based on hierarchical graph convolutional neural network, terminal and storage medium
CN111985245A (en) * 2020-08-21 2020-11-24 江南大学 Attention cycle gating graph convolution network-based relation extraction method and system
CN112163426A (en) * 2020-09-30 2021-01-01 中国矿业大学 Relationship extraction method based on combination of attention mechanism and graph long-time memory neural network

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115222688A (en) * 2022-07-12 2022-10-21 广东技术师范大学 Medical image classification method based on graph network time sequence
CN115222688B (en) * 2022-07-12 2023-01-10 广东技术师范大学 Medical image classification method based on graph network time sequence

Also Published As

Publication number Publication date
CN113080847B (en) 2022-11-29

Similar Documents

Publication Publication Date Title
CN110188836B (en) Brain function network classification method based on variational self-encoder
KR102125127B1 (en) Method of brain disorder diagnosis via deep learning
CN110598793B (en) Brain function network feature classification method
CN110731773B (en) Abnormal electrocardiogram screening method based on fusion of global and local depth features of electrocardiogram
CN113724880A (en) Abnormal brain connection prediction system, method and device and readable storage medium
CN109567789B (en) Electrocardiogram data segmentation processing method and device and readable storage medium
CN106650818B (en) Resting state function magnetic resonance image data classification method based on high-order hyper-network
CN111863244B (en) Functional connection mental disease classification method and system based on sparse pooling graph convolution
CN114041795B (en) Emotion recognition method and system based on multi-mode physiological information and deep learning
CN112043260B (en) Electrocardiogram classification method based on local mode transformation
CN111932522B (en) Alzheimer's disease classifier based on brain imaging big data deep learning
CN115272295A (en) Dynamic brain function network analysis method and system based on time domain-space domain combined state
CN109935321B (en) Risk prediction system for converting depression patient into bipolar affective disorder based on functional nuclear magnetic resonance image data
CN113080847B (en) Device for diagnosing mild cognitive impairment based on bidirectional long-short term memory model of graph
CN114881105A (en) Sleep staging method and system based on transformer model and contrast learning
CN114241240A (en) Method and device for classifying brain images, electronic equipment and storage medium
Liu et al. Semantic segmentation of qrs complex in single channel ecg with bidirectional lstm networks
Sarraf et al. Recent applications of deep learning algorithms in medical image analysis
CN116759076A (en) Unsupervised disease diagnosis method and system based on medical image
CN115909438A (en) Pain expression recognition system based on depth time-space domain convolutional neural network
Singh et al. Autism Spectrum Disorder Detection using theDeep Learning Approaches
CN110363096B (en) Brain time signal processing method based on hidden Markov model
Bharani et al. Diabetes Mellitus Diagnosis based on Tongue Images using Machine Learning
CN116665906B (en) Resting state functional magnetic resonance brain age prediction method based on similarity twin network
CN114913587B (en) Non-contact heart rate measurement uncertainty quantization method based on Bayesian deep learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
CB03 Change of inventor or designer information

Inventor after: An Xingwei

Inventor after: Zhou Yutao

Inventor after: Ming Dong

Inventor after: Guo Hengyan

Inventor after: Zhao Jindi

Inventor after: Zhong Wenxiao

Inventor after: Ke Yufeng

Inventor after: Liu Shuang

Inventor before: Zhou Yutao

Inventor before: An Xingwei

Inventor before: Ming Dong

Inventor before: Guo Hengyan

Inventor before: Zhao Jindi

Inventor before: Zhong Wenxiao

Inventor before: Ke Yufeng

Inventor before: Liu Shuang

CB03 Change of inventor or designer information
GR01 Patent grant
GR01 Patent grant