CN111914952B - AD characteristic parameter screening method and system based on deep neural network - Google Patents
AD characteristic parameter screening method and system based on deep neural network Download PDFInfo
- Publication number
- CN111914952B CN111914952B CN202010847474.5A CN202010847474A CN111914952B CN 111914952 B CN111914952 B CN 111914952B CN 202010847474 A CN202010847474 A CN 202010847474A CN 111914952 B CN111914952 B CN 111914952B
- Authority
- CN
- China
- Prior art keywords
- stimulation
- value
- input layer
- neural network
- model
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012216 screening Methods 0.000 title claims abstract description 24
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 19
- 230000000638 stimulation Effects 0.000 claims abstract description 50
- 210000002569 neuron Anatomy 0.000 claims abstract description 39
- 230000008859 change Effects 0.000 claims abstract description 28
- 238000003062 neural network model Methods 0.000 claims abstract description 27
- 210000004556 brain Anatomy 0.000 claims description 11
- 239000002207 metabolite Substances 0.000 claims description 7
- 230000001629 suppression Effects 0.000 claims description 4
- 230000007423 decrease Effects 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 abstract description 14
- 238000004364 calculation method Methods 0.000 abstract description 3
- 238000005457 optimization Methods 0.000 description 5
- 208000024827 Alzheimer disease Diseases 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 201000010099 disease Diseases 0.000 description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- BTCSSZJGUNDROE-UHFFFAOYSA-N gamma-aminobutyric acid Chemical class NCCCC(O)=O BTCSSZJGUNDROE-UHFFFAOYSA-N 0.000 description 3
- 238000002595 magnetic resonance imaging Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 229960003692 gamma aminobutyric acid Drugs 0.000 description 2
- 238000004611 spectroscopical analysis Methods 0.000 description 2
- 208000024891 symptom Diseases 0.000 description 2
- 206010008096 Cerebral atrophy Diseases 0.000 description 1
- -1 GLX Chemical compound 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
The invention discloses an AD characteristic parameter screening method and system based on a deep neural network. The method comprises the following steps: acquiring AD initial characteristic parameters; taking the initial characteristic parameters as input of a deep neural network model; performing input layer node stimulation on the deep learning neural network model to obtain a neuron node value corresponding to stimulation of an input layer and a relative change value of model output values before and after stimulation; linearly fitting the neuron node value of the corresponding stimulation of the input layer and the relative change value of the model output value before and after the stimulation; and screening the initial characteristic parameters according to the fitting result to determine AD characteristic parameters. According to the method, the input layer parameter value is changed, and the linear fitting is carried out on the corresponding prediction result to obtain the high correlation quantity about output, so that compared with the existing method, the method has the advantages that the operation quantity is low, complex calculation such as covariance is not needed, and the like, and the AD characteristic parameters can be screened more quickly and accurately by the adopted model.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to an AD characteristic parameter screening method and system based on a deep neural network.
Background
With the application development of artificial intelligence in the medical field, the deep learning technology has good effects on classification and feature extraction of complex problems in constructing an unconventional model. The deep learning technology is adopted to classify the Alzheimer's disease (Alzheimer disease, AD) to obtain higher classification accuracy, and a solution is provided for multi-mode information fusion effect, super-parameter and kernel function selection. Since the research on specific characteristic parameters leading to AD conditions is not clear, the setting of characteristic parameters becomes a major problem in the classification of AD.
Super-parameter optimization is always a difficult problem for limiting the improvement of the network model performance, the current neural network model is more and more complex, the types of super-parameters are more and more, and proper super-parameter combinations cannot be selected according to the relation among the super-parameters. The current super-parameter optimization method is a Bayesian optimization method, good decision combination can be found from a decision space under the actual evaluation times as much as possible, the main idea is to construct a proxy model of the whole problem process according to historical data without evaluating the actual problem, determine the next sampling point through uncertainty of the prediction of the proxy model, and find an approximate optimal solution after continuous iteration. The main proxy model in Bayesian optimization is Gaussian Process (GPs) at present, but generally only a single problem is optimized at a time, or the problems of fully utilizing data information in a certain time are ensured by parallel running at the cost of hardware. Single-task learning solely starts from zero, ignores related information of other similar tasks to deeply study data features, and often encounters problems of large noise, higher data dimension, smaller data size, and the like, which have a larger influence on the results. A large amount of observation data is usually required to train to obtain a single-task agent model with enough accuracy, but the requirement is difficult to reach in real life, so that the model trained according to the data has certain limitation, and the model prediction is not accurate enough. With the increase of data volume, the covariance function calculation complexity in the Gaussian process increases exponentially, which results in the problems of high calculation cost, long running time and the like.
Disclosure of Invention
The invention aims to provide an AD characteristic parameter screening method and system based on a deep neural network, which can screen data with larger and sensitive model output value change through node stimulation, namely, can screen AD characteristic parameters capable of more accurately classifying AD symptoms.
In order to achieve the above object, the present invention provides the following solutions:
an AD characteristic parameter screening method based on a deep neural network comprises the following steps:
acquiring AD initial characteristic parameters; taking the initial characteristic parameters as input of a deep neural network model;
performing input layer node stimulation on the deep learning neural network model to obtain a neuron node value corresponding to stimulation of an input layer and a relative change value of model output values before and after stimulation;
linearly fitting the neuron node value of the corresponding stimulation of the input layer and the relative change value of the model output value before and after the stimulation;
and screening the initial characteristic parameters according to the fitting result to determine AD characteristic parameters.
Optionally, the AD initial characteristic parameters include brain region voxels, metabolite concentration, and individual characteristics.
Optionally, the performing input layer node stimulation on the deep learning neural network model includes:
changes are made according to 0.6-1.4 times of the characteristic value of the original neuron node.
Optionally, the fitting result represents a degree of influence of a change in the input layer neuron node on an output value of the deep learning neural network model.
The invention also provides an AD characteristic parameter screening system based on the deep neural network, which comprises the following steps:
the initial characteristic parameter acquisition module is used for acquiring AD initial characteristic parameters; taking the initial characteristic parameters as input of a deep neural network model;
the stimulation module is used for carrying out input layer node stimulation on the deep learning neural network model to obtain a neuron node value corresponding to the stimulation of the input layer and a relative change value of the output values of the model before and after the stimulation;
the fitting module is used for linearly fitting the neuron node value of the corresponding stimulation of the input layer and the relative change value of the model output value before and after the stimulation;
and the screening module is used for screening the initial characteristic parameters according to the fitting result and determining AD characteristic parameters.
Optionally, the AD initial characteristic parameters include brain region voxels, metabolite concentration, and individual characteristics.
Optionally, the performing input layer node stimulation on the deep learning neural network model includes:
changes are made according to 0.6-1.4 times of the characteristic value of the original neuron node.
Optionally, the fitting result represents a degree of influence of a change in the input layer neuron node on an output value of the deep learning neural network model.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides an AD characteristic parameter screening method and system based on a deep neural network, which are characterized in that the high correlation quantity about output is obtained by changing the parameter value of an input layer and performing linear fitting on a corresponding prediction result.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of an AD feature parameter screening method based on a deep neural network according to an embodiment of the invention;
FIG. 2 is a schematic diagram of an SAE neural network structure model according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of forward promotion of the first 1-20 data according to an embodiment of the present invention;
FIG. 4 is a diagram of the reverse suppression of the first 1-20 data according to an embodiment of the present invention;
fig. 5 is a block diagram of an AD feature parameter screening system based on a deep neural network according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide an AD characteristic parameter screening method and system based on a deep neural network, which can screen data with larger and sensitive model output value change through node stimulation, namely, can screen AD characteristic parameters capable of more accurately classifying AD symptoms.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
As shown in fig. 1, an AD feature parameter screening method based on a deep neural network includes:
step 101: acquiring AD initial characteristic parameters; and taking the initial characteristic parameters as input of a deep neural network model. The AD initial characteristic parameters comprise brain region voxels, metabolite concentration and individual characteristics.
Human body parameters related to Alzheimer's disease (Alzheimer disease, AD) are studied through a deep learning neural network, and according to the previous study, the magnetic resonance image (Magnetic Resonance Imaging, MRI) technology can be known to accurately acquire the change of brain region voxels (volumes) caused by cerebral atrophy, thereby helping the diagnosis of AD. Magnetic resonance spectroscopy (Magnetic Resonance spectroscopy, MRS) techniques can accurately detect GABA metabolite concentrations, thereby aiding in the diagnosis of AD. Therefore, the voxel of the brain 279 brain regions obtained from the MRI image and the metabolite concentration of GABA, GLX, NAA and the like in the MRS data, and the personal characteristic parameters of age, sex and the like are set as super parameters of the neural network, the labels of the model are set to 1 and 0,1 represents an AD patient, and 0 represents a healthy individual. An SAE neural network model is adopted as the AD classification model. The modeling result is: average, variance and maximum errors were 0.0111, 0.0001 and 0.0356 respectively, and AUC values for the model were 0.9750.
Step 102: and performing input layer node stimulation on the deep learning neural network model to obtain a neuron node value corresponding to the stimulation of the input layer and a relative change value of the model output values before and after the stimulation.
And performing input layer node stimulation on the deep learning neural network model, namely traversing each node of the input layer, and changing according to 0.6 to 1.4 times of the characteristic value of the original neuron node to obtain a neuron node value corresponding to stimulation of the input layer and a relative change value of the output value of the model before and after the stimulation. Taking the SAE neural network model shown in fig. 2 as an example: the neuron node value x1 of the stimulus input layer is input according to the values of 0.6x1, 0.8x1, 1.0x1, 1.2x1 and 1.4x1, and x2, … and xm are unchanged, so that an output result of changing x1 is obtained. And x2, … and xm are sequentially carried out according to x1 operation, and only one parameter is changed each time to obtain respective output layer prediction results.
Step 103: and linearly fitting the neuron node value of the corresponding stimulation of the input layer and the relative change value of the model output value before and after the stimulation.
And (3) performing linear fitting on the neuron node value of the corresponding stimulus of the input layer obtained in the step (102) and the relative change value of the model output value before and after the stimulus, and expressing the change degree by using the absolute value of the slope. The more the model output value tends to be 1, the greater the probability of representing a disease, the more the model output value tends to be 0, the less the probability of representing a disease.
Taking x1 as an example, the fitting result is:
x2, …, xm are performed sequentially according to x1 operation.
Step 104: and screening the initial characteristic parameters according to the fitting result to determine AD characteristic parameters.
And (3) according to the linear fitting result of the relative change value of the neuron node value of the corresponding stimulus of the input layer and the model output value before and after the stimulus, which is obtained in the step (103), the larger the slope is, the larger the influence of the change of the neuron node of the input layer on the model output value is, and the input layer parameters with large influence on the model output value are filtered out.
The forward promotion probability is referred to when the model output value increases with an increase in the input layer neuron node value, i.e., the model output value tends to be 1, and the reverse suppression probability is referred to when the model output value decreases with an increase in the input layer neuron node value, i.e., the model output value tends to be 0.
After the stimulation of the neuron nodes, type data with high forward promotion disease probability and type data with reverse inhibition disease probability (20 data types in forward and reverse directions respectively, left brain and right brain) are respectively taken. The selection results are shown in fig. 3-4.
And (3) verifying an optimization result:
in order to verify that the extracted 40-dimensional data can effectively classify AD, the extracted 40-dimensional data is subjected to SAE depth neural network modeling, the modeling mode is unchanged, and the modeling result is as follows: average, variance and maximum errors were 0.0053, 4.13E-0.5 and 0.02, respectively, and the AUC value of the model was 0.9750.
40 pieces of multi-mode data, MRS data (GABA) and age data are effectively screened out through the method for modeling, and classification performance of the model is improved, so that the method is effective and feasible.
As shown in fig. 5, the present invention further provides an AD feature parameter screening system based on a deep neural network, including:
an initial characteristic parameter obtaining module 501, configured to obtain an AD initial characteristic parameter; and taking the initial characteristic parameters as input of a deep neural network model. The AD initial characteristic parameters comprise brain region voxels, metabolite concentration and individual characteristics.
The stimulation module 502 is configured to perform input layer node stimulation (changing according to 0.6-1.4 times of the characteristic value of the original neuron node) on the deep learning neural network model, so as to obtain a neuron node value corresponding to the input layer stimulation and a relative change value of the model output values before and after the stimulation.
And a fitting module 503, configured to linearly fit the neuron node value of the input layer corresponding to the stimulus and the relative variation value of the model output value before and after the stimulus.
And a screening module 504, configured to screen the initial feature parameters according to the fitting result, and determine AD feature parameters. The fitting result represents the degree of influence of the change of the input layer neuron nodes on the output value of the deep learning neural network model.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (2)
1. The AD characteristic parameter screening method based on the deep neural network is characterized by comprising the following steps of:
acquiring AD initial characteristic parameters; taking the initial characteristic parameters as input of a deep neural network model; the AD initial characteristic parameters comprise brain region voxels, metabolite concentration and individual characteristics;
performing input layer node stimulation on the deep neural network model to obtain a neuron node value corresponding to stimulation of an input layer and a relative change value of model output values before and after the stimulation; the method specifically comprises the following steps: performing input layer node stimulation on the deep neural network model, namely traversing each node of the input layer, and changing according to 0.6 to 1.4 times of the characteristic value of the original neuron node to obtain a neuron node value of the input layer corresponding to stimulation and a relative change value of the model output values before and after the stimulation;
linearly fitting the neuron node value of the corresponding stimulation of the input layer and the relative change value of the model output value before and after the stimulation;
screening the initial characteristic parameters according to the fitting result to determine AD characteristic parameters; the fitting result represents the influence degree of the change of the neuron nodes of the input layer on the output value of the deep neural network model; the method specifically comprises the following steps: the forward promotion probability is referred to when the model output value increases with an increase in the input layer neuron node value, i.e., the model output value tends to be 1, and the reverse suppression probability is referred to when the model output value decreases with an increase in the input layer neuron node value, i.e., the model output value tends to be 0.
2. An AD feature parameter screening system based on a deep neural network, comprising:
the initial characteristic parameter acquisition module is used for acquiring AD initial characteristic parameters; taking the initial characteristic parameters as input of a deep neural network model; the AD initial characteristic parameters comprise brain region voxels, metabolite concentration and individual characteristics;
the stimulation module is used for carrying out input layer node stimulation on the deep neural network model to obtain a neuron node value corresponding to the stimulation of the input layer and a relative change value of the output values of the model before and after the stimulation; the method specifically comprises the following steps: performing input layer node stimulation on the deep neural network model, namely traversing each node of the input layer, and changing according to 0.6 to 1.4 times of the characteristic value of the original neuron node to obtain a neuron node value of the input layer corresponding to stimulation and a relative change value of the model output values before and after the stimulation;
the fitting module is used for linearly fitting the neuron node value of the corresponding stimulation of the input layer and the relative change value of the model output value before and after the stimulation;
the screening module is used for screening the initial characteristic parameters according to the fitting result to determine AD characteristic parameters; the fitting result represents the influence degree of the change of the neuron nodes of the input layer on the output value of the deep neural network model; the method specifically comprises the following steps: the forward promotion probability is referred to when the model output value increases with an increase in the input layer neuron node value, i.e., the model output value tends to be 1, and the reverse suppression probability is referred to when the model output value decreases with an increase in the input layer neuron node value, i.e., the model output value tends to be 0.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010847474.5A CN111914952B (en) | 2020-08-21 | 2020-08-21 | AD characteristic parameter screening method and system based on deep neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010847474.5A CN111914952B (en) | 2020-08-21 | 2020-08-21 | AD characteristic parameter screening method and system based on deep neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111914952A CN111914952A (en) | 2020-11-10 |
CN111914952B true CN111914952B (en) | 2024-03-08 |
Family
ID=73279765
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010847474.5A Active CN111914952B (en) | 2020-08-21 | 2020-08-21 | AD characteristic parameter screening method and system based on deep neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111914952B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112712891B (en) * | 2020-12-28 | 2022-02-08 | 聊城大学 | Brain function connection network learning method without hyper-parameters |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1987905A (en) * | 2006-12-14 | 2007-06-27 | 东华大学 | Nerve network input parameter screening method based on fuzzy logic |
CN104933722A (en) * | 2015-06-29 | 2015-09-23 | 电子科技大学 | Image edge detection method based on Spiking-convolution network model |
US9280745B1 (en) * | 2015-07-08 | 2016-03-08 | Applied Underwriters, Inc. | Artificial intelligence expert system for screening |
CN106874581A (en) * | 2016-12-30 | 2017-06-20 | 浙江大学 | A kind of energy consumption of air conditioning system in buildings Forecasting Methodology based on BP neural network model |
CN107505837A (en) * | 2017-07-07 | 2017-12-22 | 浙江大学 | A kind of semi-supervised neural network model and the soft-measuring modeling method based on the model |
CN107767946A (en) * | 2017-09-26 | 2018-03-06 | 浙江工业大学 | Breast cancer diagnosis system based on PCA and PSO KELM models |
CN108898160A (en) * | 2018-06-01 | 2018-11-27 | 中国人民解放军战略支援部队信息工程大学 | Breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features |
CN110443364A (en) * | 2019-06-21 | 2019-11-12 | 深圳大学 | A kind of deep neural network multitask hyperparameter optimization method and device |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11301750B2 (en) * | 2017-03-31 | 2022-04-12 | Ecole Polytechnique Federale De Lausanne (Epfl) | Simplification of neural models that include arborized projections |
-
2020
- 2020-08-21 CN CN202010847474.5A patent/CN111914952B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1987905A (en) * | 2006-12-14 | 2007-06-27 | 东华大学 | Nerve network input parameter screening method based on fuzzy logic |
CN104933722A (en) * | 2015-06-29 | 2015-09-23 | 电子科技大学 | Image edge detection method based on Spiking-convolution network model |
US9280745B1 (en) * | 2015-07-08 | 2016-03-08 | Applied Underwriters, Inc. | Artificial intelligence expert system for screening |
CN106874581A (en) * | 2016-12-30 | 2017-06-20 | 浙江大学 | A kind of energy consumption of air conditioning system in buildings Forecasting Methodology based on BP neural network model |
CN107505837A (en) * | 2017-07-07 | 2017-12-22 | 浙江大学 | A kind of semi-supervised neural network model and the soft-measuring modeling method based on the model |
CN107767946A (en) * | 2017-09-26 | 2018-03-06 | 浙江工业大学 | Breast cancer diagnosis system based on PCA and PSO KELM models |
CN108898160A (en) * | 2018-06-01 | 2018-11-27 | 中国人民解放军战略支援部队信息工程大学 | Breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features |
CN110443364A (en) * | 2019-06-21 | 2019-11-12 | 深圳大学 | A kind of deep neural network multitask hyperparameter optimization method and device |
Non-Patent Citations (3)
Title |
---|
Classification of Alzheimer’s Disease Based on Deep Learning of Brain Structural and Metabolic Data;Huiquan Wang 等;《Frontiers in Aging Neuroscience》;第14卷;第1-10页 * |
Discrimination of EMG Signals Using a Neuromorphic Implementation of a Spiking Neural Network;Elisa Donati 等;《IEEE Transactions on Biomedical Circuits and Systems》;第13卷(第5期);第795 - 803页 * |
基于深度学习的阿尔兹海默症分类模型研究;甘琦睿;《中国优秀硕士学位论文全文数据库 (医药卫生科技辑)》(第02期);第E071-50页 * |
Also Published As
Publication number | Publication date |
---|---|
CN111914952A (en) | 2020-11-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kang et al. | Identifying early mild cognitive impairment by multi-modality MRI-based deep learning | |
Rahman et al. | Ambiguous medical image segmentation using diffusion models | |
Robinson et al. | Dynamic functional connectivity using state-based dynamic community structure: Method and application to opioid analgesia | |
CN107242873B (en) | Brain network construction method based on functional magnetic resonance psychophysiological interaction | |
Warren et al. | Functional magnetic resonance imaging, deep learning, and Alzheimer's disease: A systematic review | |
CN109065128A (en) | A kind of sparse brain network establishing method of weighted graph regularization | |
CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
CN105117731A (en) | Community partition method of brain functional network | |
CN112488976B (en) | Multi-modal medical image fusion method based on DARTS network | |
Bonacchi et al. | Role of artificial intelligence in MS clinical practice | |
CN110889496B (en) | Human brain effect connection identification method based on countermeasure generation network | |
CN115272295A (en) | Dynamic brain function network analysis method and system based on time domain-space domain combined state | |
Khan et al. | MRI imaging, comparison of MRI with other modalities, noise in MRI images and machine learning techniques for noise removal: a review | |
CN111914952B (en) | AD characteristic parameter screening method and system based on deep neural network | |
US20210313046A1 (en) | Qualitative and quantitative MRI using deep learning | |
Mouches et al. | Unifying brain age prediction and age-conditioned template generation with a deterministic autoencoder | |
Shehata et al. | A deep learning-based cad system for renal allograft assessment: diffusion, bold, and clinical biomarkers | |
Selvathi et al. | Brain region segmentation using convolutional neural network | |
KR102439319B1 (en) | Method and apparatus for processing nuclear magnetic resonance and magnetic resonance spectroscopy data based on artificial neural network | |
CN108805181A (en) | A kind of image classification device and sorting technique based on more disaggregated models | |
Lim et al. | Fitting a directional microstructure model to diffusion-relaxation mri data with self-supervised machine learning | |
CN114266738A (en) | Longitudinal analysis method and system for mild brain injury magnetic resonance image data | |
CN111476726B (en) | Unsupervised two-photon calcium imaging denoising method and device based on antagonistic neural network | |
CN117197637B (en) | Constant model construction method, application and device for neural image brain network efficiency index | |
CN114862834B (en) | Resting state functional magnetic resonance image data classification method |
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 | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20240124 Address after: No.324, jingwuweiqi Road, Jinan City, Shandong Province Applicant after: Shandong Provincial Hospital Affiliated to Shandong First Medical University Country or region after: China Address before: No.324, jingwuweiqi Road, Jinan City, Shandong Province Applicant before: SHANDONG MEDICAL IMAGING Research Institute Country or region before: China |
|
GR01 | Patent grant | ||
GR01 | Patent grant |