CN109785346A - Monitoring model training method and device based on tongue phase partitioning technique - Google Patents
Monitoring model training method and device based on tongue phase partitioning technique Download PDFInfo
- Publication number
- CN109785346A CN109785346A CN201910075133.8A CN201910075133A CN109785346A CN 109785346 A CN109785346 A CN 109785346A CN 201910075133 A CN201910075133 A CN 201910075133A CN 109785346 A CN109785346 A CN 109785346A
- Authority
- CN
- China
- Prior art keywords
- tongue
- fatty liver
- patient
- phase images
- tongue phase
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The embodiment of the present application provides a kind of monitoring model training method and device based on tongue phase partitioning technique, by extracting each fatty liver associated region marked in the first tongue phase images sample having and the second tongue phase images sample for being marked with non-fat hepatopath, and according to the fatty liver associated region training depth convolutional neural networks in each first tongue phase images sample of extraction and the second tongue phase images sample, fatty liver monitoring model is obtained.Then, the fatty liver associated region in the tongue phase images of patient to be measured is input in fatty liver monitoring model, obtains the confidence level that the patient to be measured suffers from fatty liver, the fatty liver situation of the patient to be measured is reflected with the confidence level to suffer from fatty liver by the patient to be measured.The Classification and Identification of the tongue phase images of patient is carried out by using deep learning algorithm as a result, so that whether fast slowdown monitoring patient suffers from fatty liver, fatty liver monitoring can be carried out at any time by carrying out interrogation treatment without patient to hospital, be used convenient for patient.
Description
Technical field
This application involves field of computer technology, in particular to a kind of monitoring model based on tongue phase partitioning technique
Training method and device.
Background technique
Therapy of combing traditional Chinese and Western medicine is the important topic of medicine for a long time, how by the approach application of Chinese medicine to doctor trained in Western medicine, is
Numerous people provides the medical services of efficient quick, becomes the problem of educational circles.Discovery and diagnosis for fatty liver use at present
Way be still to go whether to measure with fatty liver by certain body indexs based on the diagnosis and treatment data such as blood test.But
In the formation of early stage fatty liver, patient is simultaneously ignorant, and also there will be no motivations, and hospital to be gone to carry out the coherence checks such as blood, causes to work as
When making a definite diagnosis fatty liver, conditions of patients is very serious.
Summary of the invention
In order to overcome above-mentioned deficiency in the prior art, being designed to provide for the application is a kind of based on tongue phase partitioning technique
Monitoring model training method and device, to solve or improve the above problem.
To achieve the goals above, the embodiment of the present application the technical solution adopted is as follows:
In a first aspect, the embodiment of the present application provides the monitoring model training method based on tongue phase partitioning technique, it is applied to electricity
Sub- equipment, which comprises
Tongue phase images sample set is obtained, the tongue phase images sample set includes multiple first tongues for being marked with Patients with Fatty Liver
Phase images sample and the multiple second tongue phase images samples for being marked with non-fat hepatopath；
Extract the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample, and according to mentioning
Fatty liver associated region training depth convolutional Neural in each first tongue phase images sample and the second tongue phase images sample taken
Network obtains fatty liver monitoring model, wherein extracts the rouge in each first tongue phase images sample and the second tongue phase images sample
The step of fat liver associated region, specifically includes:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase height
Direction is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, obtains subregion tongue phasor
Decent；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second
The overlapping region being located between two second areas at edge in region is associated with as the fatty liver of the tongue phase images sample
Region obtains the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample to extract；
Fatty liver associated region in the tongue phase images of patient to be measured is input in the fatty liver monitoring model, is obtained
The confidence level that the patient to be measured suffers from fatty liver.
In a kind of possible embodiment, the step of the acquisition tongue phase images sample set, comprising:
It obtains each patient and stretches out the patient image acquired when tongue；
Tongue position in each patient image is marked, and each patient image is carried out using facefinder
Face datection obtains multiple face key feature points；
Tongue image is intercepted from each patient image according to the multiple face key feature points detected；
Each tongue image of interception is input in tongue parted pattern trained in advance, obtains each tongue image
Tongue tab area；
By the pixel value of each pixel of each tongue image and each pixel of corresponding tongue tab area
Pixel value carries out product calculation, to obtain tongue phase images sample set.
In a kind of possible embodiment, the method also includes:
Training obtains the tongue parted pattern in advance, specifically includes:
Tongue image in each patient image is carried out to carry out tongue area marking, generates the tongue in each patient image
Head tab area；
It is input with each patient image, take the tongue tab area of each patient image as output iteration instruction
Practice depth convolutional neural networks, and exports the tongue parted pattern when reaching trained termination condition.
In a kind of possible embodiment, multiple face key feature points that the basis detects are schemed from each patient
The step of tongue image is intercepted as in, comprising:
Lower jaw cusp and upper lip midpoint are obtained from the multiple face key feature points；
Corresponding rectangular area is determined from each patient image according to the lower jaw cusp and upper lip midpoint；Wherein,
The abscissa of the central point of the rectangular area be the lower jaw cusp abscissa and the upper lip midpoint abscissa it
Half, the height of the sum of the ordinate of the half of sum, the ordinate that ordinate is the lower jaw cusp and the upper lip midpoint
For 2 times of the difference of the ordinate of the ordinate and lower jaw cusp at the upper lip midpoint, width be the upper lip midpoint
Ordinate and 1.5 times of difference of ordinate of the lower jaw cusp；
The rectangular area in each patient image is intercepted to intercept tongue image from each patient image.
In a kind of possible embodiment, the method also includes:
The step of generating the fatty liver monitoring result of the patient to be measured according to the confidence level that the patient to be measured suffers from fatty liver, tool
Body includes:
Whether the confidence level for judging that the patient to be measured suffers from fatty liver is greater than setting confidence level；
If the confidence level that the patient to be measured suffers from fatty liver is greater than setting confidence level, generates the patient to be measured and suffer from fatty liver
Fatty liver prediction result；
If the confidence level that the patient to be measured suffers from fatty liver generates the patient to be measured and is not suffering from rouge no more than setting confidence level
The fatty liver prediction result of fat liver.
Second aspect, the embodiment of the present application also provide a kind of monitoring model training device based on tongue phase partitioning technique, answer
For electronic equipment, described device includes:
Module is obtained, for obtaining tongue phase images sample set, the tongue phase images sample set includes being marked with fatty liver trouble
Multiple first tongue phase images samples of person and the multiple second tongue phase images samples for being marked with non-fat hepatopath；
Training module is extracted, for extracting the fatty liver in each first tongue phase images sample and the second tongue phase images sample
Associated region, and according to the fatty liver associated region in each first tongue phase images sample of extraction and the second tongue phase images sample
Training depth convolutional neural networks, obtain fatty liver monitoring model, wherein the extraction training module is especially by following manner
Extract the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase height
Direction is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, obtains subregion tongue phasor
Decent；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second
The overlapping region being located between two second areas at edge in region is associated with as the fatty liver of the tongue phase images sample
Region obtains the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample to extract；
Input module is input to the fatty liver prison for the fatty liver associated region in the tongue phase images by patient to be measured
It surveys in model, obtains the confidence level that the patient to be measured suffers from fatty liver.
The third aspect, the embodiment of the present application also provide a kind of readable storage medium storing program for executing, are stored thereon with computer program, described
Computer program, which is performed, realizes the above-mentioned monitoring model training method based on tongue phase partitioning technique.
In terms of existing technologies, the application has the advantages that
The embodiment of the present application provides a kind of monitoring model training method and device based on tongue phase partitioning technique, passes through extraction
Each the second tongue phase images sample for marking the first tongue phase images sample having and being marked with non-fat hepatopath
In fatty liver associated region, and according to the fat in each first tongue phase images sample of extraction and the second tongue phase images sample
Liver associated region trains depth convolutional neural networks, obtains fatty liver monitoring model.It then, will be in the tongue phase images of patient to be measured
Fatty liver associated region be input in fatty liver monitoring model, the confidence level that the patient to be measured suffers from fatty liver is obtained, to pass through
The confidence level that the patient to be measured suffers from fatty liver reflects the fatty liver situation of the patient to be measured.It is calculated as a result, by using deep learning
Method carries out the Classification and Identification of the tongue phase images of patient, so that whether fast slowdown monitoring patient suffers from fatty liver, without patient to hospital
Fatty liver monitoring can be carried out at any time by carrying out interrogation treatment, be used convenient for patient.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only some embodiments of the application, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is that the process of the monitoring model training method provided by the embodiments of the present application based on tongue phase partitioning technique is illustrated
Figure；
Fig. 2 is that multiple face key feature points that basis provided by the embodiments of the present application detects are intercepted from patient image
Tongue image schematic diagram；
Fig. 3 is tongue tab area schematic diagram provided by the embodiments of the present application；
Fig. 4 is the tongue phase images sample schematic diagram provided by the embodiments of the present application being partitioned into；
Fig. 5 is the segmentation schematic diagram of tongue phase images sample provided by the embodiments of the present application；
Fig. 6 is that the functional module of the monitoring model training device provided by the embodiments of the present application based on tongue phase partitioning technique is shown
It is intended to；
Fig. 7 is provided by the embodiments of the present application for realizing the above-mentioned monitoring model training method based on tongue phase partitioning technique
Electronic equipment structural schematic block diagram.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiment is some embodiments of the present application, instead of all the embodiments.Usually herein
The component of the embodiment of the present application described and illustrated in place's attached drawing can be arranged and be designed with a variety of different configurations.
Therefore, the detailed description of the embodiments herein provided in the accompanying drawings is not intended to limit below claimed
Scope of the present application, but be merely representative of the selected embodiment of the application.Based on the embodiment in the application, this field is common
Technical staff's all other embodiment obtained without creative labor belongs to the application protection
Range.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.
Referring to Fig. 1, being the one of the monitoring model training method provided by the embodiments of the present application based on tongue phase partitioning technique
Kind flow diagram, it should be understood that in other embodiments, the monitoring model training based on tongue phase partitioning technique of the present embodiment
The sequence of method part step can be exchanged with each other according to actual needs or part steps therein also can be omitted or
It deletes.The detailed step of the monitoring model training method based on tongue phase partitioning technique is described below.
Step S210 obtains tongue phase images sample set.
In the present embodiment, the tongue phase images sample set include be marked with Patients with Fatty Liver multiple first tongue phasors it is decent
Sheet and the multiple second tongue phase images samples for being marked with non-fat hepatopath.
In detail, as an example, firstly, obtaining each patient stretches out the patient image acquired when tongue.For example, can
To allow patient to stretch out tongue when acquiring patient image and be directed at camera, guarantee that patient's tongue appears in preparation as far as possible and adopts with this
Then the centre of the patient image of collection acquires the patient image of each patient by camera.
Then, the tongue position in each patient image is marked, and each patient is schemed using facefinder
As carrying out Face datection, multiple face key feature points are obtained, wherein the face key feature points may include that face is main
68 key points.
Then, tongue image is intercepted from each patient image according to the multiple face key feature points detected.For example,
Lower jaw cusp and upper lip midpoint can be obtained from the multiple face key feature points, then according to the lower jaw cusp and
Upper lip midpoint determines corresponding rectangular area from each patient image.Wherein, the horizontal seat of the central point of the rectangular area
It is designated as the half of the sum of the abscissa of the lower jaw cusp and the abscissa at the upper lip midpoint, ordinate is the lower jaw point
Half, the ordinate highly for the upper lip midpoint and the institute of the sum of the ordinate of the ordinate and the upper lip midpoint put
State the vertical seat of 2 times of the difference of the ordinate of lower jaw cusp, the ordinate that width is the upper lip midpoint and the lower jaw cusp
1.5 times of the difference of mark.Namely shown in Fig. 2, for tongue image is intercepted from each patient image through the above steps, it is assumed that lower jaw
The coordinate of cusp is (x1, y1), the coordinate at upper lip midpoint is (x2, y2), then correspondence can be determined in each patient image
Rectangular area, it is highly 2* (y2-y1) that the coordinate of the central point of the rectangular area, which is ((x1+x2)/2, (y1+y2)/2), wide
Degree is 1.5* (y2-y1).
Then, on the basis of the above, the rectangular area in each patient image is intercepted to intercept from each patient image
Tongue image, and each tongue image of interception is input in tongue parted pattern trained in advance, obtain each tongue figure
The tongue tab area of picture.
In detail, it also needs training in advance to obtain the tongue parted pattern before this, specifically includes: to each patient
Tongue image in image carries out carrying out tongue area marking, generates the tongue tab area in each patient image, such as Fig. 3
Shown, the white area in the center Fig. 3 is tongue tab area.It then, is input, with described each with each patient image
The tongue tab area of a patient image is output repetitive exercise depth convolutional neural networks, and when reaching trained termination condition
Export the tongue parted pattern.
On the basis of the above, by the way that each tongue image of interception to be input in tongue parted pattern trained in advance,
The tongue tab area of available each tongue image.
Then, by each pixel of the pixel value of each pixel of each tongue image and corresponding tongue tab area
The pixel value of point carries out product calculation, tongue phase images sample shown in Fig. 4 can be obtained, to obtain tongue phase images sample set.
Step S220 extracts the fatty liver association area in each first tongue phase images sample and the second tongue phase images sample
Domain, and it is deep according to the fatty liver associated region training in each first tongue phase images sample of extraction and the second tongue phase images sample
Convolutional neural networks are spent, fatty liver monitoring model is obtained.
In detail, to be thought according to theory of traditional Chinese medical science, Fatty Liver Disease is generally associated with the part specific region in patient's tongue phase,
The specific region is had found after studying for a long period of time through inventor.As an example, for each first tongue phase images sample
With the second tongue phase images sample, which is equally divided into five first areas, in tongue phase in tongue phase short transverse
It is equally divided into four second areas in width direction, obtains subregion tongue phase images sample, it is decent then to extract the subregion tongue phasor
Two second of edge are located in the first area of centre three in five first areas and four second areas in this
Fatty liver associated region of the overlapping region as the tongue phase images sample between region obtains each first tongue phasor to extract
Fatty liver associated region in decent and the second tongue phase images sample.Such as the segmentation of tongue phase images sample that Fig. 5 is shown
Schematic diagram, fill part is the fatty liver association area in each first tongue phase images sample and the second tongue phase images sample in figure
Domain.
On the basis of the above, according to the fat in each first tongue phase images sample of extraction and the second tongue phase images sample
Liver associated region trains depth convolutional neural networks, obtains fatty liver monitoring model.
Fatty liver associated region in the tongue phase images of patient to be measured is input to the fatty liver and monitors mould by step S230
In type, the confidence level that the patient to be measured suffers from fatty liver is obtained.
Through the above steps, the classification for the tongue phase images that the present embodiment carries out patient by using deep learning algorithm is known
Not, so that whether fast slowdown monitoring patient suffers from fatty liver, fat can be carried out at any time by carrying out interrogation treatment without patient to hospital
Liver monitoring, uses convenient for patient.
As an implementation, still eferring to Fig. 1, step S240, the confidence level to be suffered from fatty liver according to the patient to be measured
Generate the fatty liver prediction result of the patient to be measured.
It is alternatively possible to which whether the confidence level for judging that the patient to be measured suffers from fatty liver is greater than setting confidence level, if this is to be measured
The confidence level that patient suffers from fatty liver is greater than setting confidence level, then generates the fatty liver prediction knot that the patient to be measured suffers from fatty liver
Fruit；If the confidence level that the patient to be measured suffers from fatty liver generates the patient to be measured and does not suffer from fatty liver no more than setting confidence level
Fatty liver prediction result.
For example, the confidence level that the patient to be measured suffers from fatty liver is greater than 0.5 if setting confidence level as 0.5, then generating should should be to
Survey the fatty liver prediction result that patient suffers from fatty liver；If the confidence level that the patient to be measured suffers from fatty liver is not more than 0.5, generating should
The fatty liver prediction result that the patient to be measured does not suffer from fatty liver.
Further, referring to Fig. 6, the embodiment of the present application also provides a kind of monitoring model instruction based on tongue phase partitioning technique
Practice device 200, the function that should be realized based on the monitoring model training device 200 of tongue phase partitioning technique can correspond to above-mentioned based on tongue
The step of monitoring model training method of phase partitioning technique executes.As shown in fig. 6, the message forwarding controller 200 can wrap
It includes and obtains module 210, extracts training module 220 and input module 230, separately below to the prison based on tongue phase partitioning technique
The function of surveying each functional module of model training apparatus 200 is described in detail.
Module 210 is obtained, for obtaining tongue phase images sample set, the tongue phase images sample set includes being marked with fatty liver
Multiple first tongue phase images samples of patient and the multiple second tongue phase images samples for being marked with non-fat hepatopath；
Training module 220 is extracted, for extracting the rouge in each first tongue phase images sample and the second tongue phase images sample
Fat liver associated region, and be associated with according to the fatty liver in each first tongue phase images sample of extraction and the second tongue phase images sample
Regional training depth convolutional neural networks, obtain fatty liver monitoring model, wherein the extraction training module is especially by following
Mode extracts the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase height
Direction is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, obtains subregion tongue phasor
Decent；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second
The overlapping region being located between two second areas at edge in region is associated with as the fatty liver of the tongue phase images sample
Region obtains the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample to extract；
Input module 230 is input to the fat for the fatty liver associated region in the tongue phase images by patient to be measured
In liver monitoring model, the confidence level that the patient to be measured suffers from fatty liver is obtained.
In a kind of possible embodiment, the acquisition module 210 is decent especially by following manner acquisition tongue phasor
This collection:
It obtains each patient and stretches out the patient image acquired when tongue；
Tongue position in each patient image is marked, and each patient image is carried out using facefinder
Face datection obtains multiple face key feature points；
Tongue image is intercepted from each patient image according to the multiple face key feature points detected；
Each tongue image of interception is input in tongue parted pattern trained in advance, obtains each tongue image
Tongue tab area；
By the pixel value of each pixel of each tongue image and each pixel of corresponding tongue tab area
Pixel value carries out product calculation, to obtain tongue phase images sample set.
In a kind of possible embodiment, the acquisition module 210 is especially by following manner from each patient image
Middle interception tongue image:
Lower jaw cusp and upper lip midpoint are obtained from the multiple face key feature points；
Corresponding rectangular area is determined from each patient image according to the lower jaw cusp and upper lip midpoint；Wherein,
The abscissa of the central point of the rectangular area be the lower jaw cusp abscissa and the upper lip midpoint abscissa it
Half, the height of the sum of the ordinate of the half of sum, the ordinate that ordinate is the lower jaw cusp and the upper lip midpoint
For 2 times of the difference of the ordinate of the ordinate and lower jaw cusp at the upper lip midpoint, width be the upper lip midpoint
Ordinate and 1.5 times of difference of ordinate of the lower jaw cusp；
The rectangular area in each patient image is intercepted to intercept tongue image from each patient image.
In a kind of possible embodiment, the extraction training module 220 extracts each the especially by following manner
Fatty liver associated region in one tongue phase images sample and the second tongue phase images sample:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase height
Direction is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, obtains subregion tongue phasor
Decent；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second
The overlapping region being located between two second areas at edge in region is associated with as the fatty liver of the tongue phase images sample
Region obtains the fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample to extract.
In a kind of possible embodiment, the monitoring model training device 200 based on tongue phase partitioning technique may be used also
To include generation module 240, the fat of the patient to be measured is generated specifically for the confidence level to suffer from fatty liver according to the patient to be measured
Liver monitoring result, specifically includes:
Whether the confidence level for judging that the patient to be measured suffers from fatty liver is greater than setting confidence level；
If the confidence level that the patient to be measured suffers from fatty liver is greater than setting confidence level, generates the patient to be measured and suffer from fatty liver
Fatty liver prediction result；
If the confidence level that the patient to be measured suffers from fatty liver generates the patient to be measured and is not suffering from rouge no more than setting confidence level
The fatty liver prediction result of fat liver.
It is understood that the concrete operation method of each functional module in the present embodiment can refer to above method embodiment
The detailed description of middle corresponding steps, it is no longer repeated herein.
Further, referring to Fig. 7, being used for the above-mentioned monitoring based on tongue phase partitioning technique to be provided by the embodiments of the present application
The structural schematic block diagram of the electronic equipment 100 of model training method.In the present embodiment, the electronic equipment 100 can be by bus
110 make general bus architecture to realize.According to the concrete application of electronic equipment 100 and overall design constraints condition,
Bus 110 may include any number of interconnection bus and bridge joint.Bus 110 by various circuit connections together, these circuits
Including processor 120, storage medium 130 and bus interface 140.Optionally, bus interface 140 can be used in electronic equipment 100
Network adapter 150 is waited and is connected via bus 110.Network adapter 150 can be used for realizing physical layer in electronic equipment 100
Signal processing function, and sending and receiving for radiofrequency signal is realized by antenna.User interface 160 can connect external equipment,
Such as: keyboard, display, mouse or control stick etc..Bus 110 can also connect various other circuits, such as timing source, periphery
Equipment, voltage regulator or management circuit etc., these circuits are known in the art, therefore are no longer described in detail.
It can replace, electronic equipment 100 may also be configured to generic processing system, such as be commonly referred to as chip, the general place
Reason system includes: to provide the one or more microprocessors of processing function, and provide at least part of of storage medium 130
External memory, it is all these all to be linked together by external bus architecture and other support circuits.
Alternatively, following realize can be used in electronic equipment 100: having processor 120, bus interface 140, user
The ASIC (specific integrated circuit) of interface 160；And it is integrated at least part of the storage medium 130 in one single chip, or
Following realize can be used in person, electronic equipment 100: one or more FPGA (field programmable gate array), PLD are (programmable
Logical device), controller, state machine, gate logic, discrete hardware components, any other suitable circuit or be able to carry out this
Any combination of the circuit of various functions described in application in the whole text.
Wherein, processor 120 is responsible for management bus 110 and general processing (is stored on storage medium 130 including executing
Software).One or more general processors and/or application specific processor can be used to realize in processor 120.Processor 120
Example includes microprocessor, microcontroller, dsp processor and the other circuits for being able to carry out software.It should be by software broadly
It is construed to indicate instruction, data or any combination thereof, regardless of being called it as software, firmware, middleware, microcode, hard
Part description language or other.
Storage medium 130 is illustrated as separating with processor 120 in Fig. 7, however, those skilled in the art be easy to it is bright
White, storage medium 130 or its arbitrary portion can be located at except electronic equipment 100.For example, storage medium 130 may include
Transmission line, the carrier waveform modulated with data, and/or the computer product that separates with radio node, these media can be with
It is accessed by processor 120 by bus interface 140.Alternatively, storage medium 130 or its arbitrary portion can integrate everywhere
It manages in device 120, for example, it may be cache and/or general register.
Above-described embodiment can be performed in the processor 120, specifically, can store in the storage medium 130 described
Based on the monitoring model training device 200 of tongue phase partitioning technique, the processor 120 can be used for executing described based on tongue phase point
The monitoring model training device 200 of area's technology.
Further, the embodiment of the present application also provides a kind of nonvolatile computer storage media, the computer is deposited
Storage media is stored with computer executable instructions, which can be performed the base in above-mentioned any means embodiment
In the monitoring model training method of tongue phase partitioning technique.
In conclusion the embodiment of the present application provides a kind of monitoring model training method and dress based on tongue phase partitioning technique
It sets, by extracting each the second tongue for marking the first tongue phase images sample having and being marked with non-fat hepatopath
Fatty liver associated region in phase images sample, and it is decent according to each first tongue phase images sample and the second tongue phasor of extraction
Fatty liver associated region training depth convolutional neural networks in this, obtain fatty liver monitoring model.Then, by patient's to be measured
Fatty liver associated region in tongue phase images is input in fatty liver monitoring model, obtains the confidence that the patient to be measured suffers from fatty liver
Degree, the fatty liver situation of the patient to be measured is reflected with the confidence level to suffer from fatty liver by the patient to be measured.As a result, by using depth
The Classification and Identification that learning algorithm carries out the tongue phase images of patient is spent, so that whether fast slowdown monitoring patient suffers from fatty liver, without suffering from
Person carries out interrogation treatment to hospital can carry out fatty liver monitoring at any time, use convenient for patient.
In embodiment provided herein, it should be understood that disclosed device and method, it can also be by other
Mode realize.Device and method embodiment described above is only schematical, for example, flow chart and frame in attached drawing
Figure shows the system frame in the cards of the system of multiple embodiments according to the application, method and computer program product
Structure, function and operation.In this regard, each box in flowchart or block diagram can represent a module, section or code
A part, a part of the module, section or code includes one or more for implementing the specified logical function
Executable instruction.It should also be noted that function marked in the box can also be with not in some implementations as replacement
It is same as the sequence marked in attached drawing generation.For example, two continuous boxes can actually be basically executed in parallel, they have
When can also execute in the opposite order, this depends on the function involved.It is also noted that in block diagram and or flow chart
Each box and the box in block diagram and or flow chart combination, can function or movement as defined in executing it is dedicated
Hardware based system realize, or can realize using a combination of dedicated hardware and computer instructions.
In addition, each functional module in each embodiment of the application can integrate one independent portion of formation together
Point, it is also possible to modules individualism, an independent part can also be integrated to form with two or more modules.
It can replace, can be realized wholly or partly by software, hardware, firmware or any combination thereof.When
When using software realization, can entirely or partly it realize in the form of a computer program product.The computer program product
Including one or more computer instructions.It is all or part of when loading on computers and executing the computer program instructions
Ground is generated according to process or function described in the embodiment of the present application.The computer can be general purpose computer, special purpose computer,
Computer network or other programmable devices.The computer instruction may be stored in a computer readable storage medium, or
Person is transmitted from a computer readable storage medium to another computer readable storage medium, for example, the computer instruction
Wired (such as coaxial cable, optical fiber, digital subscriber can be passed through from a web-site, computer, server or data center
Line (DSL)) or wireless (such as infrared, wireless, microwave etc.) mode to another web-site, computer, server or data
It is transmitted at center.The computer readable storage medium can be any usable medium that computer can access and either wrap
The data storage devices such as electronic equipment, server, the data center integrated containing one or more usable mediums.The usable medium
It can be magnetic medium, (for example, floppy disk, hard disk, tape), optical medium (for example, DVD) or semiconductor medium (such as solid-state
Hard disk Solid State Disk (SSD)) etc..
It should be noted that, in this document, term " including ", " including " or its any other variant are intended to non-row
Its property includes, so that the process, method, article or equipment for including a series of elements not only includes those elements, and
And further include the other elements being not explicitly listed, or further include for this process, method, article or equipment institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence " including one ... ", it is not excluded that including institute
State in the process, method, article or equipment of element that there is also other identical elements.
It is obvious to a person skilled in the art that the application is not limited to the details of above-mentioned exemplary embodiment, Er Qie
In the case where without departing substantially from spirit herein or essential characteristic, the application can be realized in other specific forms.Therefore, no matter
From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and scope of the present application is by appended power
Benefit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent elements of the claims
Variation is included in the application.Any reference signs in the claims should not be construed as limiting the involved claims.
Claims (10)
1. a kind of monitoring model training method based on tongue phase partitioning technique, which is characterized in that be applied to electronic equipment, the side
Method includes:
Tongue phase images sample set is obtained, the tongue phase images sample set includes the multiple first tongue phasors for being marked with Patients with Fatty Liver
Decent and it is marked with multiple second tongue phase images samples of non-fat hepatopath；
The fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample is extracted, and according to extraction
Fatty liver associated region training depth convolutional neural networks in each first tongue phase images sample and the second tongue phase images sample,
Obtain fatty liver monitoring model, wherein extract the fatty liver in each first tongue phase images sample and the second tongue phase images sample
The step of associated region, specifically includes:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase short transverse
It is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, it is decent to obtain subregion tongue phasor
This；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second areas
In be located at fatty liver associated region of the overlapping region as the tongue phase images sample between two second areas at edge,
The fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample is obtained to extract；
Fatty liver associated region in the tongue phase images of patient to be measured is input in the fatty liver monitoring model, obtain this to
Survey the confidence level that patient suffers from fatty liver.
2. the monitoring model training method according to claim 1 based on tongue phase partitioning technique, which is characterized in that described to obtain
The step of taking tongue phase images sample set, comprising:
It obtains each patient and stretches out the patient image acquired when tongue；
Tongue position in each patient image is marked, and face is carried out to each patient image using facefinder
Detection, obtains multiple face key feature points；
Tongue image is intercepted from each patient image according to the multiple face key feature points detected；
Each tongue image of interception is input in tongue parted pattern trained in advance, obtains the tongue of each tongue image
Tab area；
By the pixel of the pixel value of each pixel of each tongue image and each pixel of corresponding tongue tab area
Value carries out product calculation, to obtain tongue phase images sample set.
3. the monitoring model training method according to claim 2 based on tongue phase partitioning technique, which is characterized in that the side
Method further include:
Training obtains the tongue parted pattern in advance, specifically includes:
Tongue image in each patient image is carried out to carry out tongue area marking, generates the tongue mark in each patient image
Infuse region；
It is input with each patient image, take the tongue tab area of each patient image as output repetitive exercise depth
Convolutional neural networks are spent, and export the tongue parted pattern when reaching trained termination condition.
4. the monitoring model training method according to claim 2 based on tongue phase partitioning technique, which is characterized in that described
The step of tongue image is intercepted from each patient image according to the multiple face key feature points detected, comprising:
Lower jaw cusp and upper lip midpoint are obtained from the multiple face key feature points；
Corresponding rectangular area is determined from each patient image according to the lower jaw cusp and upper lip midpoint；Wherein, described
The abscissa of the central point of rectangular area is the sum of the abscissa of the lower jaw cusp and the abscissa at the upper lip midpoint
Half, ordinate are the half of the sum of the ordinate of the lower jaw cusp and the ordinate at the upper lip midpoint, are highly institute
State 2 times of the difference of the ordinate at upper lip midpoint and the ordinate of the lower jaw cusp, width be the vertical of the upper lip midpoint
1.5 times of the difference of the ordinate of coordinate and the lower jaw cusp；
The rectangular area in each patient image is intercepted to intercept tongue image from each patient image.
5. the monitoring model training method based on tongue phase partitioning technique described in any one of -4 according to claim 1, special
Sign is, the method also includes:
The step of generating the fatty liver monitoring result of the patient to be measured according to the confidence level that the patient to be measured suffers from fatty liver, it is specific to wrap
It includes:
Whether the confidence level for judging that the patient to be measured suffers from fatty liver is greater than setting confidence level；
If the confidence level that the patient to be measured suffers from fatty liver is greater than setting confidence level, the rouge that the patient to be measured suffers from fatty liver is generated
Fat liver prediction result；
If the confidence level that the patient to be measured suffers from fatty liver generates the patient to be measured and does not suffer from fatty liver no more than setting confidence level
Fatty liver prediction result.
6. a kind of monitoring model training device based on tongue phase partitioning technique, which is characterized in that be applied to electronic equipment, the dress
It sets and includes:
Module is obtained, for obtaining tongue phase images sample set, the tongue phase images sample set includes being marked with Patients with Fatty Liver
Multiple first tongue phase images samples and the multiple second tongue phase images samples for being marked with non-fat hepatopath；
Training module is extracted, for extracting the association of the fatty liver in each first tongue phase images sample and the second tongue phase images sample
Region, and according to the fatty liver associated region training in each first tongue phase images sample of extraction and the second tongue phase images sample
Depth convolutional neural networks obtain fatty liver monitoring model, wherein the extraction training module is extracted especially by following manner
Fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase short transverse
It is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, it is decent to obtain subregion tongue phasor
This；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second areas
In be located at fatty liver associated region of the overlapping region as the tongue phase images sample between two second areas at edge,
The fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample is obtained to extract；
Input module is input to the fatty liver monitoring mould for the fatty liver associated region in the tongue phase images by patient to be measured
In type, the confidence level that the patient to be measured suffers from fatty liver is obtained.
7. the monitoring model training device according to claim 6 based on tongue phase partitioning technique, which is characterized in that described to obtain
Modulus block obtains tongue phase images sample set especially by following manner:
It obtains each patient and stretches out the patient image acquired when tongue；
Tongue position in each patient image is marked, and face is carried out to each patient image using facefinder
Detection, obtains multiple face key feature points；
Tongue image is intercepted from each patient image according to the multiple face key feature points detected；
Each tongue image of interception is input in tongue parted pattern trained in advance, obtains the tongue of each tongue image
Tab area；
By the pixel of the pixel value of each pixel of each tongue image and each pixel of corresponding tongue tab area
Value carries out product calculation, to obtain tongue phase images sample set.
8. the monitoring model training device according to claim 7 based on tongue phase partitioning technique, which is characterized in that described to obtain
Modulus block intercepts tongue image especially by following manner from each patient image:
Lower jaw cusp and upper lip midpoint are obtained from the multiple face key feature points；
Corresponding rectangular area is determined from each patient image according to the lower jaw cusp and upper lip midpoint；Wherein, described
The abscissa of the central point of rectangular area is the sum of the abscissa of the lower jaw cusp and the abscissa at the upper lip midpoint
Half, ordinate are the half of the sum of the ordinate of the lower jaw cusp and the ordinate at the upper lip midpoint, are highly institute
State 2 times of the difference of the ordinate at upper lip midpoint and the ordinate of the lower jaw cusp, width be the vertical of the upper lip midpoint
1.5 times of the difference of the ordinate of coordinate and the lower jaw cusp；
The rectangular area in each patient image is intercepted to intercept tongue image from each patient image.
9. the monitoring model training device according to claim 6 based on tongue phase partitioning technique, which is characterized in that described to mention
Training module is taken to extract the fat in each first tongue phase images sample and the second tongue phase images sample especially by following manner
Liver associated region:
For each first tongue phase images sample and the second tongue phase images sample, by the tongue phase images sample in tongue phase short transverse
It is equally divided into five first areas, is equally divided into four second areas in tongue phase width direction, it is decent to obtain subregion tongue phasor
This；
Extract the first area of centre three in five first areas in the subregion tongue phase images sample and four second areas
In be located at fatty liver associated region of the overlapping region as the tongue phase images sample between two second areas at edge,
The fatty liver associated region in each first tongue phase images sample and the second tongue phase images sample is obtained to extract.
10. the monitoring model training device based on tongue phase partitioning technique according to any one of claim 6-9, special
Sign is, described device further include:
Generation module, the fatty liver for generating the patient to be measured specifically for the confidence level to suffer from fatty liver according to the patient to be measured monitor
As a result, specifically including:
Whether the confidence level for judging that the patient to be measured suffers from fatty liver is greater than setting confidence level；
If the confidence level that the patient to be measured suffers from fatty liver is greater than setting confidence level, the rouge that the patient to be measured suffers from fatty liver is generated
Fat liver prediction result；
If the confidence level that the patient to be measured suffers from fatty liver generates the patient to be measured and does not suffer from fatty liver no more than setting confidence level
Fatty liver prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910075133.8A CN109785346A (en) | 2019-01-25 | 2019-01-25 | Monitoring model training method and device based on tongue phase partitioning technique |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910075133.8A CN109785346A (en) | 2019-01-25 | 2019-01-25 | Monitoring model training method and device based on tongue phase partitioning technique |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109785346A true CN109785346A (en) | 2019-05-21 |
Family
ID=66502464
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910075133.8A Pending CN109785346A (en) | 2019-01-25 | 2019-01-25 | Monitoring model training method and device based on tongue phase partitioning technique |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109785346A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111444971A (en) * | 2020-03-31 | 2020-07-24 | 联想(北京)有限公司 | Information processing method, information processing device, electronic equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1830380A (en) * | 2006-04-20 | 2006-09-13 | 上海交通大学 | Tongue picture anulgis and diagnosis system of traditional Chinese medicine |
| US20080139966A1 (en) * | 2006-12-07 | 2008-06-12 | The Hong Kong Polytechnic University | Automatic tongue diagnosis based on chromatic and textural features classification using bayesian belief networks |
| CN104463865A (en) * | 2014-12-05 | 2015-03-25 | 浙江大学 | Human image segmenting method |
| CN106295139A (en) * | 2016-07-29 | 2017-01-04 | 姹ゅ钩 | A kind of tongue body autodiagnosis health cloud service system based on degree of depth convolutional neural networks |
| CN107563123A (en) * | 2017-09-27 | 2018-01-09 | 百度在线网络技术（北京）有限公司 | Method and apparatus for marking medical image |
| CN108553081A (en) * | 2018-01-03 | 2018-09-21 | 京东方科技集团股份有限公司 | A kind of diagnostic system based on tongue fur image |
| CN109190535A (en) * | 2018-08-23 | 2019-01-11 | 南京邮电大学 | A kind of face blee analysis method and system based on deep learning |
-
2019
- 2019-01-25 CN CN201910075133.8A patent/CN109785346A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1830380A (en) * | 2006-04-20 | 2006-09-13 | 上海交通大学 | Tongue picture anulgis and diagnosis system of traditional Chinese medicine |
| US20080139966A1 (en) * | 2006-12-07 | 2008-06-12 | The Hong Kong Polytechnic University | Automatic tongue diagnosis based on chromatic and textural features classification using bayesian belief networks |
| CN104463865A (en) * | 2014-12-05 | 2015-03-25 | 浙江大学 | Human image segmenting method |
| CN106295139A (en) * | 2016-07-29 | 2017-01-04 | 姹ゅ钩 | A kind of tongue body autodiagnosis health cloud service system based on degree of depth convolutional neural networks |
| CN107563123A (en) * | 2017-09-27 | 2018-01-09 | 百度在线网络技术（北京）有限公司 | Method and apparatus for marking medical image |
| CN108553081A (en) * | 2018-01-03 | 2018-09-21 | 京东方科技集团股份有限公司 | A kind of diagnostic system based on tongue fur image |
| CN109190535A (en) * | 2018-08-23 | 2019-01-11 | 南京邮电大学 | A kind of face blee analysis method and system based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| 刘龙飞: "中医舌诊辅助系统", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 * |
| 李刚: "基于光谱法舌诊的脂肪肝快速诊断", 《光谱学与光谱分析》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111444971A (en) * | 2020-03-31 | 2020-07-24 | 联想(北京)有限公司 | Information processing method, information processing device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lei et al. | RETRACTED: 5G-oriented IoT coverage enhancement and physical education resource management | |
| US11633119B2 (en) | Medical imaging and efficient sharing of medical imaging information | |
| US20230290460A1 (en) | Medical imaging, efficient sharing and secure handling of medical imaging information | |
| US20210391040A1 (en) | Systems and methods for tracking, accessing and merging protected health information | |
| US9662040B2 (en) | Computer-aided diagnosis apparatus and method | |
| US20210012883A1 (en) | Systems and methods for longitudinally tracking fully de-identified medical studies | |
| CN108198191A (en) | Image processing method and device | |
| CN102542242B (en) | The biological characteristic area positioning method and device of contactless collection image | |
| US11357435B2 (en) | Automatic extraction of disease-specific features from doppler images | |
| TW202125415A (en) | Training method, equipment and storage medium of 3d target detection and model | |
| CN109146993A (en) | A kind of Method of Medical Image Fusion and system | |
| Rasouli D et al. | A novel depth image analysis for sleep posture estimation | |
| CN109390040A (en) | A kind of intelligent electronic medical records system | |
| CN109740611A (en) | Tongue image analysis method and device | |
| CN109785346A (en) | Monitoring model training method and device based on tongue phase partitioning technique | |
| CN109035212A (en) | A kind of labeling method of lung ct image particular tissues | |
| US11132793B2 (en) | Case-adaptive medical image quality assessment | |
| Zhang et al. | Computing group cardinality constraint solutions for logistic regression problems | |
| Huang et al. | Online reflective learning for robust medical image segmentation | |
| JPWO2019146358A1 (en) | Learning systems, methods and programs | |
| CN111260651A (en) | Stomach low-quality MRI image segmentation method based on deep migration learning | |
| WO2022145988A1 (en) | Apparatus and method for facial fracture reading using artificial intelligence | |
| JP7265805B2 (en) | Image analysis method, image analysis device, image analysis system, control program, recording medium | |
| WO2021124293A1 (en) | Deep learning for optical coherence tomography segmentation | |
| Sartori et al. | An infrastructure for wearable environments acquisition and representation |
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 | ||
| CB03 | Change of inventor or designer information |
Inventor after: Chao Dai Inventor after: He Fan Inventor after: Liu Li Inventor after: Zhou Zhen Inventor after: Chen Jin Inventor after: Yao Shukun Inventor after: Duan Shaojie Inventor after: Chen Jialiang Inventor before: Chao Dai Inventor before: He Fan Inventor before: Liu Li Inventor before: Zhou Zhen Inventor before: Chen Jin |
|
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210302 Address after: 610200 Sichuan Chengdu Shuangliu District Dongsheng Street Chengdu core Valley Industrial Park concentrated area Applicant after: CHINA POWER HEALTH CLOUD TECHNOLOGY Co.,Ltd. Applicant after: China Japan Friendship Hospital (China Japan Friendship Institute of clinical medicine) Address before: 610200 Sichuan Chengdu Shuangliu District Dongsheng Street Chengdu core Valley Industrial Park concentrated area Applicant before: CHINA POWER HEALTH CLOUD TECHNOLOGY Co.,Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190521 |