CN110348500A - Sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery - Google Patents
Sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery Download PDFInfo
- Publication number
- CN110348500A CN110348500A CN201910581978.4A CN201910581978A CN110348500A CN 110348500 A CN110348500 A CN 110348500A CN 201910581978 A CN201910581978 A CN 201910581978A CN 110348500 A CN110348500 A CN 110348500A
- Authority
- CN
- China
- Prior art keywords
- sleep disturbance
- layer
- size
- obtains
- temperature data
- 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
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—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
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Heart & Thoracic Surgery (AREA)
- Veterinary Medicine (AREA)
- Pathology (AREA)
- Public Health (AREA)
- Surgery (AREA)
- Medical Informatics (AREA)
- Animal Behavior & Ethology (AREA)
- Multimedia (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery, include training and two stages of test, the purpose of training stage is the parted pattern for training the disaggregated model that can be realized automatic diagnosis sleep disturbance and automatic segmentation abnormal area, and the groundwork of test phase is the analysis for being directly split and classifying to infrared thermal imagery using trained model and generates result report.The present invention utilizes the untouchable and safe and healthy feature of IR thermal imaging inspection, in conjunction with the method for deep learning, the structure of neural network is built and adjusted for practical problem autonomous innovation to be adapted to carry out the automatic auxiliary diagnosis of sleep disturbance, can fast and effeciently be supplied to doctor with reference to diagnostic comments.
Description
Technical field
The invention mainly relates to sleep disturbance intelligent auxiliary diagnosis fields, more particularly to a kind of intelligence based on infrared thermal imagery
The method of energy auxiliary diagnosis sleep disturbance.
Background technique
Sleep disturbance refers to the exception of the exception of amount of sleep and matter of sleeping or certain clinical symptoms occurs in sleep, such as sleeps
Dormancy reduces or hypersomnia, somnambulism etc..It investigating according to the World Health Organization, about 1/3 people has sleeping problems in world wide, I
State-owned all kinds of sleep disturbance persons account for about the 38% of crowd, and higher than the ratio in the world 27%, sleep disturbance not only results in immune function
Can be impaired, can also be to cardiovascular health, metabolism, mental health, cancer susceptibility etc. has an impact.Infrared thermal imagery is as one
Kind green and healthy threaded non-contact detection instrument device, to the detection of sleep disturbance, there are also representative reflections, it is demonstrated experimentally that sleeping
The infrared thermal imagery abnormal expression of dormancy deficiency patient shows as frontal region in asymmetric, uneven distribution abnormal high fever expression, and can
Occurs the red abnormal high hot line of threadiness of single or double items between unilateral or bilateral brows and chignon line；Ocular is shown as
The round or semicircular abnormal high fever expression that endocanthion hot-zone range increases, often involves upper and lower eyelid.This explanation utilizes infrared
Thermal imagery can be used as the means of auxiliary diagnosis sleep disturbance, medically, manual sort usually require a great deal of time with
Energy, and subsidiary classification rapidly and efficiently may be implemented using deep learning algorithm as a result, playing actively for the work of doctor
Back work, improve doctor's working efficiency.
Therefore, it is necessary to improve to the prior art.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of efficiently sleep barriers based on deep learning and infrared thermal imagery
Hinder aided diagnosis method.
In order to solve the above technical problems, the present invention is provided and a kind of is assisted based on deep learning and the sleep disturbance of infrared thermal imagery
Diagnostic method, it is characterised in that: the following steps are included:
Step 1, after training being pre-processed with Infrared Thermogram, the head zone in Infrared Thermogram is identified, and
Be respectively fed to after the temperature data of head zone is normalized abnormal area parted pattern and sleep disturbance disaggregated model into
The training of row neural network obtains trained abnormal area parted pattern and sleep disturbance disaggregated model；
Step 2, it after actual patient Infrared Thermogram being carried out data prediction, identifies in actual patient Infrared Thermogram
The head zone of human body, and trained abnormal area segmentation mould is respectively fed to after the temperature data in the region is normalized
Type and sleep disturbance disaggregated model are tested, and segmentation result and classification results are obtained, and it is auxiliary that these results are generated result report
Doctor is helped to analyze.
As to the present invention is based on the improvement of deep learning and the sleep disturbance aided diagnosis method of infrared thermal imagery: the step
Rapid 1 specific step is as follows:
Step 1.1 pre-processes training with Infrared Thermogram, obtains pretreatment image.
Step 1.2 carries out head zone identification using Faster-RCNN network to pretreatment image, obtains head zone
Image；
The temperature data normalization of step 1.3, head zone image, obtains normalized temperature data；
Step 1.4 is trained come undated parameter sleep disturbance disaggregated model using normalized temperature data；It obtains
Trained sleep disturbance disaggregated model；
Step 1.5 is trained come undated parameter abnormal area parted pattern using normalized temperature data；It obtains
Trained abnormal area parted pattern；
Step 1.6 saves trained sleep disturbance disaggregated model and abnormal area parted pattern；
As to the present invention is based on the further improvements of deep learning and the sleep disturbance aided diagnosis method of infrared thermal imagery:
Specific step is as follows for the step 2:
Step 2.1 pre-processes actual patient image, obtains pretreatment image；
Step 2.2 carries out head zone identification using Faster-RCNN network to pretreatment image, obtains head zone
Image；
The temperature data normalization of step 2.3, head zone image, obtains normalized temperature data；
Temperature data after normalization is sent into trained sleep disturbance disaggregated model by step 2.4, obtains classification knot
Fruit；
Temperature data after normalization is sent into trained abnormal area parted pattern by step 2.5, obtains segmentation knot
Fruit；
Step 2.6 obtains result report according to classification results and segmentation result.
As to the present invention is based on the further improvements of deep learning and the sleep disturbance aided diagnosis method of infrared thermal imagery:
Sleep disturbance disaggregated model includes 4 layers of convolutional layer and 2 layers of full articulamentum, is first connect behind a convolutional layer
BatchNorm layers, then the nonlinear activation of activation primitive is carried out, loss function is cross entropy, learning rate 1e-5, optimisation strategy
For Adam gradient descent method, batch size is 64.
As to the present invention is based on the further improvements of deep learning and the sleep disturbance aided diagnosis method of infrared thermal imagery:
Abnormal area parted pattern uses FCN network, activation primitive Relu, and loss function is the cross entropy of Pixel-level,
Learning rate is 2e-5, and optimisation strategy is momentum gradient descent method, and batch size is 16.
As to the present invention is based on the further improvements of deep learning and the sleep disturbance aided diagnosis method of infrared thermal imagery:
The specific structure of sleep disturbance disaggregated model are as follows:
1st layer of input layer, Output Size 32x32x1；Level 2 volume lamination, Output Size 32x32x32, convolution kernel size
3x3, convolution nuclear volume 32, step-length 1；3rd layer of pond layer, Output Size 16x16x32, convolution kernel size 2x2, step-length 2；4th layer
Convolutional layer, Output Size 16x16x64, convolution kernel size 3x3, convolution nuclear volume 64, step-length 1；5th layer of pond layer, Output Size
8x8x64, convolution kernel size 2x2, step-length 2；6th layer of convolutional layer, Output Size 8x8x128, convolution kernel size 3x3, convolution nucleus number
Amount 128, step-length 1；7th layer of pond layer, Output Size 4x4x128, convolution kernel size 2x2, step-length 2；8th layer of convolutional layer, output
Size 4x4x256, convolution kernel size 3x3, convolution nuclear volume 256, step-length 1；9th layer of pond layer, Output Size 2x2x256, volume
Product core size 2x2, step-length 2；10th layer of full articulamentum, Output Size 1x1x1024；The full articulamentum of 11th layer, Output Size
1x1x2。
The present invention is based on the technical advantages of deep learning and the sleep disturbance aided diagnosis method of infrared thermal imagery are as follows:
The present invention includes training and two stages of test, and the purpose of training stage is to train to can be realized to diagnose automatically to sleep
The parted pattern of the disaggregated model of dormancy obstacle and automatic segmentation abnormal area, the groundwork of test phase are using trained
Analysis that model is directly split infrared thermal imagery and classifies simultaneously generates result report.The present invention utilizes IR thermal imaging inspection
Untouchable and safe and healthy feature is built and is adjusted for practical problem autonomous innovation in conjunction with the method for deep learning
The structure of whole neural network can fast and effeciently be supplied to doctor's ginseng to be adapted to carry out the automatic auxiliary diagnosis of sleep disturbance
Examine diagnostic comments.
Detailed description of the invention
Specific embodiments of the present invention will be described in further detail with reference to the accompanying drawing.
Fig. 1 is integrated stand composition of the invention；
Fig. 2 is the Faster-RCNN network structure for detecting head zone automatically；
Fig. 3 is the sorter network structure chart for subsidiary classification sleep disturbance；
Fig. 4 is to divide network structure for dividing the abnormal area of abnormal hot-zone.
Specific embodiment
The present invention is described further combined with specific embodiments below, but protection scope of the present invention is not limited in
This.
Embodiment 1, the sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery, as shown in Figs 1-4, this hair
Bright purpose is to provide a kind of method of auxiliary diagnosis sleep disturbance based on deep learning and infrared thermal imagery, this method include with
Lower step:
Step 1 carries out data prediction to Infrared Thermogram, obtains pretreatment image；
There is low definition in the Infrared Thermogram for the patient being collected into, therefore using the method pair of bicubic interpolation
Infrared Thermogram carries out super-resolution rebuilding, obtains higher image definition.In training, trained Infrared Thermogram is by curing
Raw relatively authoritative mark, mark include whether hot-zone is abnormal and whether have sleep disturbance etc..
Step 2 carries out head zone identification to pretreatment image, obtains head zone image；
Since the major part of sleep disturbance concern is human body head, it is therefore desirable to which what is handled and analyze is also to come from head
Temperature data, using existing object recognition technique Faster-RCNN network carry out head automatic identification, Faster-
RCNN includes three parts, i.e. the extraction conventional part of feature, the RPN network for automatically extracting area-of-interest and final
Part for classifying and detection block returns.
The groundwork for extracting the conventional part of feature is the characteristics of image for extracting infrared chart, the network packet of this part
Containing 4 sections of convolution, the size of a maximum pond layer reduction characteristic pattern is connect behind every section of convolution, therefore to the RPN net of next stage
Before network, the size of characteristic pattern has been reduced to the 1/16 of original image.
The groundwork of RPN network is to automatically extract interested region, its primary structure first passes through a 3x3
Convolution extracts feature, then generates scale difference for each point on characteristic pattern, and 9 different candidate frames of length-width ratio utilize
The position of actual frames is tagged to this 9 frames, is then trained, is adjusted to the position of candidate frame, eventually by non-
The method that maximum inhibits removes the higher frame of Duplication.
The main function of final classification Recurrent networks is the specific classification that classification is carried out to the candidate frame that RPN network generates
And the further intense adjustment of the position of frame, primary structure be using ROI-Pooling to the characteristic pattern of area-of-interest into
The consistent characteristic pattern of size is obtained behind row pond, then by 2 full articulamentums, a responsible classification, another is responsible for recurrence,
The loss function of classification task uses softmax, returns the loss function of task using smooth-L1.
The specific steps of this part work are as follows: it is tagged for the head zone of thermography first, as accomplishing fluently label
Training data, the i.e. position coordinates of head zone rectangle frame, then by accomplish fluently label training data be sent into Faster-RCNN into
Row training, can be obtained one after training can be with the network model of automatic identification head zone.
Network model after pretreatment image input training, obtains head zone image.
Step 3, the normalization of head zone image temperature data；Obtain normalized temperature data；
By the temperature data that thermal infrared imager provides, the temperature data of the head zone automatically identified is extracted
Come, and be normalized, data are limited between 0 to 1, normalization formula is as follows, wherein xnorm,xmin,xmaxRespectively
For temperature normalized value, temperature minimum value and temperature maximum:
Step 4, training sleep disturbance sorter network carry out sleep disturbance using the existing training data for accomplishing fluently label
The training of disaggregated model, which includes 4 layers of convolutional layer and 2 layers of full articulamentum, shown in table specific as follows:
Table 1
Wherein, BatchNorm layers are first connect behind each convolutional layer, then carry out the nonlinear activation of activation primitive, it is such
Benefit is all to draw each layer of distribution onto a normal distribution, can accelerate the training of network, and can improve over-fitting
Phenomenon.
Normalized temperature data is zoomed into 32x32 size, the input as training sleep disturbance sorter network.
Trained parameter setting is as follows: activation primitive is common Relu, and formula is as follows, and advantage is to enable to network fast
Speed convergence, and improve the problem of neural network gradient disappears.
Loss function is cross entropy, and learning rate 1e-5, optimisation strategy is Adam gradient descent method, and batch size is 64.
By the relatively authoritative mark of doctor, the infrared thermal imagery label of the patient with sleep disturbance is set as 1, it will be without sleep
The infrared thermal imagery label of the normal person of obstacle is set as 0, and network is instructed as training data by known data in this part
Practice.Sorter network parameter is updated as loss function using cross entropy formula.When training, normalized temperature data is inputted,
Classification results are obtained by Internet communication, and itself and actual classification label comparing calculation loss function are recycled into loss function
As a result gradient decline is carried out to update the parameter of network.Obtain the sleep disturbance sorter network of training completion.
Step 5, training abnormal area parted pattern are labeled with the data of abnormal hot-zone as instruction using known doctor
Practice data, carries out the training of abnormal area segmentation network, the basic model using FCN network as abnormal area segmentation network,
Building for the segmentation network suitable for the problem is carried out on this basis, the main thought of the network is to be based on full convolutional network,
The convolution of first half is for extracting characteristics of image, and image is restored to original image size using deconvolution by latter half, and to original
Scheme each pixel and carries out classification prediction.Specific structure is as shown in the table:
Table 2
Equally, plus BatchNorm layers of acceleration convergence after each convolution, last network output is to each picture of original image
The class probability of element, i.e., each pixel is the probability of abnormal hot-zone point, and training parameter is provided that
Activation primitive is Relu, and loss function is the cross entropy of Pixel-level, learning rate 2e-5, and optimisation strategy is
Momentum gradient descent method, batch size are 16.Training when, input normalized temperature data, by before neural network to biography
Acquisition one is broadcast with input matrix probability graph of the same size, wherein it is abnormal hot-zone pixel that each pixel, which represents the point,
Probability value.The corresponding label of the input is the abnormal hot-zone figure marked according to doctor, is the pixel for belonging to abnormal hot-zone on the figure
Value be 1, remaining position is 0, the output of network and label is carried out to the calculating of loss function, and pass through the reversed biography of gradient decline
Broadcast update network parameter.
Step 6 preserves training to convergent sorter network and abnormal area segmentation network, for using when test.
Step 7 tests actual patient Infrared Thermogram, when actual patient is tested, is not no label, needs
It is predicted using trained network.Successively carry out data prediction, head zone automatic identification and head zone temperature
Data normalization is spent, it is similar with work (Step 1: two, three) of training stage, it does not illustrate excessively.Next the temperature that will be handled well
Degree obtains the probability with sleep disturbance according to trained disaggregated model is sent into, while being sent into trained parted pattern and generating
Potential exception thermal region, obtains segmentation result and classification results, these results are generated result report auxiliary doctor and are divided
Analysis, a foundation as auxiliary diagnosis.
The above list is only a few specific embodiments of the present invention for finally, it should also be noted that.Obviously, this hair
Bright to be not limited to above embodiments, acceptable there are many deformations.Those skilled in the art can be from present disclosure
All deformations for directly exporting or associating, are considered as protection scope of the present invention.
Claims (6)
1. the sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery, it is characterised in that: the following steps are included:
Step 1, after training being pre-processed with Infrared Thermogram, the head zone in Infrared Thermogram is identified, and correct
The temperature data in portion region is respectively fed to abnormal area parted pattern after being normalized and sleep disturbance disaggregated model carries out mind
Training through network obtains trained abnormal area parted pattern and sleep disturbance disaggregated model；
Step 2, after actual patient Infrared Thermogram being carried out data prediction, human body in actual patient Infrared Thermogram is identified
Head zone, and be respectively fed to after the temperature data in the region is normalized trained abnormal area parted pattern and
Sleep disturbance disaggregated model is tested, and segmentation result and classification results are obtained, these results are generated result report auxiliary doctor
Life is analyzed.
2. the sleep disturbance aided diagnosis method according to claim 1 based on deep learning and infrared thermal imagery, feature
Be: specific step is as follows for the step 1:
Step 1.1 pre-processes training with Infrared Thermogram, obtains pretreatment image.
Step 1.2 carries out head zone identification using Faster-RCNN network to pretreatment image, obtains head zone image；
The temperature data normalization of step 1.3, head zone image, obtains normalized temperature data；
Step 1.4 is trained come undated parameter sleep disturbance disaggregated model using normalized temperature data；It is trained
Good sleep disturbance disaggregated model；
Step 1.5 is trained come undated parameter abnormal area parted pattern using normalized temperature data；It is trained
Good abnormal area parted pattern；
Step 1.6 saves trained sleep disturbance disaggregated model and abnormal area parted pattern.
3. the sleep disturbance aided diagnosis method according to claim 2 based on deep learning and infrared thermal imagery, feature
Be: specific step is as follows for the step 2:
Step 2.1 pre-processes actual patient image, obtains pretreatment image；
Step 2.2 carries out head zone identification using Faster-RCNN network to pretreatment image, obtains head zone image；
The temperature data normalization of step 2.3, head zone image, obtains normalized temperature data；
Temperature data after normalization is sent into trained sleep disturbance disaggregated model by step 2.4, obtains classification results；
Temperature data after normalization is sent into trained abnormal area parted pattern by step 2.5, obtains segmentation result；
Step 2.6 obtains result report according to classification results and segmentation result.
4. the sleep disturbance aided diagnosis method according to claim 3 based on deep learning and infrared thermal imagery, feature
It is:
Sleep disturbance disaggregated model includes 4 layers of convolutional layer and 2 layers of full articulamentum, first connects BatchNorm layers behind a convolutional layer,
The nonlinear activation of activation primitive is carried out again, and loss function is cross entropy, and learning rate 1e-5, optimisation strategy is under Adam gradient
Drop method, batch size are 64.
5. the sleep disturbance aided diagnosis method according to claim 4 based on deep learning and infrared thermal imagery, feature
It is:
Abnormal area parted pattern uses FCN network, activation primitive Relu, and loss function is the cross entropy of Pixel-level, study
Rate is 2e-5, and optimisation strategy is momentum gradient descent method, and batch size is 16.
6. the sleep disturbance aided diagnosis method according to claim 5 based on deep learning and infrared thermal imagery, feature
It is:
The specific structure of sleep disturbance disaggregated model are as follows:
1st layer of input layer, Output Size 32x32x1；Level 2 volume lamination, Output Size 32x32x32, convolution kernel size 3x3, volume
Product nuclear volume 32, step-length 1；3rd layer of pond layer, Output Size 16x16x32, convolution kernel size 2x2, step-length 2；4th layer of convolution
Layer, Output Size 16x16x64, convolution kernel size 3x3, convolution nuclear volume 64, step-length 1；5th layer of pond layer, Output Size
8x8x64, convolution kernel size 2x2, step-length 2；6th layer of convolutional layer, Output Size 8x8x128, convolution kernel size 3x3, convolution nucleus number
Amount 128, step-length 1；7th layer of pond layer, Output Size 4x4x128, convolution kernel size 2x2, step-length 2；8th layer of convolutional layer, output
Size 4x4x256, convolution kernel size 3x3, convolution nuclear volume 256, step-length 1；9th layer of pond layer, Output Size 2x2x256, volume
Product core size 2x2, step-length 2；10th layer of full articulamentum, Output Size 1x1x1024；The full articulamentum of 11th layer, Output Size
1x1x2。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910581978.4A CN110348500A (en) | 2019-06-30 | 2019-06-30 | Sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910581978.4A CN110348500A (en) | 2019-06-30 | 2019-06-30 | Sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110348500A true CN110348500A (en) | 2019-10-18 |
Family
ID=68177697
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910581978.4A Pending CN110348500A (en) | 2019-06-30 | 2019-06-30 | Sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110348500A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111248868A (en) * | 2020-02-20 | 2020-06-09 | 长沙湖湘医疗器械有限公司 | Quick eye movement sleep analysis method, system and equipment |
| CN111915616A (en) * | 2020-05-26 | 2020-11-10 | 华瑞新智科技（北京）有限公司 | Method and device for infrared temperature measurement based on weak supervision image segmentation |
| CN111931785A (en) * | 2020-06-19 | 2020-11-13 | 国网山西省电力公司吕梁供电公司 | Edge detection method for infrared image target of power equipment |
| CN112102946A (en) * | 2019-11-21 | 2020-12-18 | 北京健康有益科技有限公司 | Sleep quality detection method and device based on deep learning |
| CN112907608A (en) * | 2021-03-17 | 2021-06-04 | 北京智象云中医科技有限公司 | Traditional Chinese medicine infrared image human body segmentation algorithm based on deep learning |
| CN113221859A (en) * | 2021-06-22 | 2021-08-06 | 浙江工业大学 | Sleep disorder distinguishing method and system based on micro-expression analysis |
| CN113303807A (en) * | 2021-05-18 | 2021-08-27 | 浙江大学 | Electrocardio data generation method |
| CN114129175A (en) * | 2021-11-19 | 2022-03-04 | 江苏科技大学 | LSTM and BP based motor imagery electroencephalogram signal classification method |
| CN116051555A (en) * | 2023-03-31 | 2023-05-02 | 深圳市冠禹半导体有限公司 | Wafer temperature distribution detection system and method |
| CN116313164A (en) * | 2023-05-22 | 2023-06-23 | 亿慧云智能科技（深圳）股份有限公司 | Anti-interference sleep monitoring method, device, equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107595243A (en) * | 2017-07-28 | 2018-01-19 | 深圳和而泰智能控制股份有限公司 | A kind of illness appraisal procedure and terminal device |
| CN107610773A (en) * | 2017-09-12 | 2018-01-19 | 北京即刻叁维数据科技股份有限公司 | A kind of vascular dissection aided diagnosis method based on sustainer medical image |
| CN108257135A (en) * | 2018-02-01 | 2018-07-06 | 浙江德尚韵兴图像科技有限公司 | The assistant diagnosis system of medical image features is understood based on deep learning method |
| CN108257674A (en) * | 2018-01-24 | 2018-07-06 | 龙马智芯（珠海横琴）科技有限公司 | Disease forecasting method and apparatus, equipment, computer readable storage medium |
| CN108523856A (en) * | 2017-03-02 | 2018-09-14 | 成都那美克科技有限公司 | Infrared thermal imaging sleep state survey meter |
| CN109559295A (en) * | 2018-06-04 | 2019-04-02 | 新影智能科技(昆山)有限公司 | Image analysis system, method, computer readable storage medium and electric terminal |
| CN109691986A (en) * | 2018-12-25 | 2019-04-30 | 合肥镭智光电科技有限公司 | Sleep detection system and detection method |
-
2019
- 2019-06-30 CN CN201910581978.4A patent/CN110348500A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108523856A (en) * | 2017-03-02 | 2018-09-14 | 成都那美克科技有限公司 | Infrared thermal imaging sleep state survey meter |
| CN107595243A (en) * | 2017-07-28 | 2018-01-19 | 深圳和而泰智能控制股份有限公司 | A kind of illness appraisal procedure and terminal device |
| CN107610773A (en) * | 2017-09-12 | 2018-01-19 | 北京即刻叁维数据科技股份有限公司 | A kind of vascular dissection aided diagnosis method based on sustainer medical image |
| CN108257674A (en) * | 2018-01-24 | 2018-07-06 | 龙马智芯（珠海横琴）科技有限公司 | Disease forecasting method and apparatus, equipment, computer readable storage medium |
| CN108257135A (en) * | 2018-02-01 | 2018-07-06 | 浙江德尚韵兴图像科技有限公司 | The assistant diagnosis system of medical image features is understood based on deep learning method |
| CN109559295A (en) * | 2018-06-04 | 2019-04-02 | 新影智能科技(昆山)有限公司 | Image analysis system, method, computer readable storage medium and electric terminal |
| CN109691986A (en) * | 2018-12-25 | 2019-04-30 | 合肥镭智光电科技有限公司 | Sleep detection system and detection method |
Non-Patent Citations (3)
| Title |
|---|
| AMAN DALMIA,ET AL.: "Exploring Deep Learning Networks for Tumour Segmentation in Infrared Images", 《QIRT 2018》 * |
| 朱莉，等.: "一种红外热图像目标区域分割的深度学习算法", 《西安电子科技大学学报》 * |
| 虢慧芝，等.: "短期睡眠不足住院患者的红外热成像表达与特性分析", 《中华生物医学工程杂志》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112102946A (en) * | 2019-11-21 | 2020-12-18 | 北京健康有益科技有限公司 | Sleep quality detection method and device based on deep learning |
| CN112102946B (en) * | 2019-11-21 | 2021-08-03 | 北京健康有益科技有限公司 | Sleep quality detection method and device based on deep learning |
| CN111248868A (en) * | 2020-02-20 | 2020-06-09 | 长沙湖湘医疗器械有限公司 | Quick eye movement sleep analysis method, system and equipment |
| CN111915616A (en) * | 2020-05-26 | 2020-11-10 | 华瑞新智科技（北京）有限公司 | Method and device for infrared temperature measurement based on weak supervision image segmentation |
| CN111931785A (en) * | 2020-06-19 | 2020-11-13 | 国网山西省电力公司吕梁供电公司 | Edge detection method for infrared image target of power equipment |
| CN112907608A (en) * | 2021-03-17 | 2021-06-04 | 北京智象云中医科技有限公司 | Traditional Chinese medicine infrared image human body segmentation algorithm based on deep learning |
| CN113303807B (en) * | 2021-05-18 | 2022-06-14 | 浙江大学 | Electrocardio data generation method |
| CN113303807A (en) * | 2021-05-18 | 2021-08-27 | 浙江大学 | Electrocardio data generation method |
| CN113221859A (en) * | 2021-06-22 | 2021-08-06 | 浙江工业大学 | Sleep disorder distinguishing method and system based on micro-expression analysis |
| CN113221859B (en) * | 2021-06-22 | 2024-03-29 | 浙江工业大学 | Sleep disorder discriminating method and system based on microexpressive analysis |
| CN114129175A (en) * | 2021-11-19 | 2022-03-04 | 江苏科技大学 | LSTM and BP based motor imagery electroencephalogram signal classification method |
| CN116051555A (en) * | 2023-03-31 | 2023-05-02 | 深圳市冠禹半导体有限公司 | Wafer temperature distribution detection system and method |
| CN116313164A (en) * | 2023-05-22 | 2023-06-23 | 亿慧云智能科技（深圳）股份有限公司 | Anti-interference sleep monitoring method, device, equipment and storage medium |
| CN116313164B (en) * | 2023-05-22 | 2023-08-22 | 亿慧云智能科技（深圳）股份有限公司 | Anti-interference sleep monitoring method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110348500A (en) | Sleep disturbance aided diagnosis method based on deep learning and infrared thermal imagery | |
| CN105825511B (en) | A kind of picture background clarity detection method based on deep learning | |
| CN105160317B (en) | One kind being based on area dividing pedestrian gender identification method | |
| CN107491726A (en) | A kind of real-time expression recognition method based on multi-channel parallel convolutional neural networks | |
| CN106709477A (en) | Face recognition method and system based on adaptive score fusion and deep learning | |
| CN108416314B (en) | Picture important face detection method | |
| CN107945153A (en) | A kind of road surface crack detection method based on deep learning | |
| CN110503053A (en) | Human motion recognition method based on cyclic convolution neural network | |
| Mondal et al. | Detection and classification technique of Yellow Vein Mosaic Virus disease in okra leaf images using leaf vein extraction and Naive Bayesian classifier | |
| CN109920538B (en) | Zero sample learning method based on data enhancement | |
| CN108596256B (en) | Object recognition classifier construction method based on RGB-D | |
| CN111666845B (en) | Small sample deep learning multi-mode sign language recognition method based on key frame sampling | |
| CN109325915A (en) | A kind of super resolution ratio reconstruction method for low resolution monitor video | |
| Al-Masawabe et al. | Papaya Maturity Classifications using Deep Convolutional Neural Networks | |
| Guo et al. | Quality grading of jujubes using composite convolutional neural networks in combination with RGB color space segmentation and deep convolutional generative adversarial networks | |
| Choudhury et al. | Quality evaluation in guavas using deep learning architectures: an experimental review | |
| Zhang et al. | Identification of jujube trees diseases using neural network | |
| CN103235943A (en) | Principal component analysis-based (PCA-based) three-dimensional (3D) face recognition system | |
| CN111368663A (en) | Method, device, medium and equipment for recognizing static facial expressions in natural scene | |
| CN116884036A (en) | Live pig posture detection method, device, equipment and medium based on YOLOv5DA | |
| Khan et al. | Screening fundus images to extract multiple ocular features: A unified modeling approach | |
| Yang et al. | Identification and detection for plant disease based on image segmentation and deep learning | |
| CN112270221A (en) | Garment personalized recommendation method fusing four-season color theory | |
| Wang et al. | A study on emotion classification of image based on BP neural network | |
| Latumakulita et al. | Human Facial Expressions Identification using Convolutional Neural Network with VGG16 Architecture. |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191018 |