CN109767828A - Interactive inline diagnosis cloud platform and its operation method and readable storage medium storing program for executing - Google Patents
Interactive inline diagnosis cloud platform and its operation method and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN109767828A CN109767828A CN201910065957.7A CN201910065957A CN109767828A CN 109767828 A CN109767828 A CN 109767828A CN 201910065957 A CN201910065957 A CN 201910065957A CN 109767828 A CN109767828 A CN 109767828A
- Authority
- CN
- China
- Prior art keywords
- information
- diagnosis
- cloud platform
- diagnosed
- pathological characters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Eye Examination Apparatus (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The present invention discloses a kind of interactive inline diagnosis cloud platform and its operation method and readable storage medium storing program for executing, wherein the operation method includes: the relevant information for receiving object to be diagnosed, and the relevant information to be diagnosed includes the eye fundus image of multispectral imaging device intake；The eye fundus image is sent to primary diagnosis personnel, generates tentative diagnosis information so that the primary diagnosis personnel diagnose；The tentative diagnosis information is obtained, and goes out convolution kernel according to the tentative diagnosis information matches；Image procossing is carried out to the eye fundus image, extracts the pathological characters information in the eye fundus image, described image processing includes the convolutional neural networks algorithm process carried out using the convolution kernel；The pathological characters information is matched in the database of cloud platform, obtains intelligent diagnostics information.Technical solution of the present invention can alleviate the pressure of oculist or promote diagnosis efficiency and accuracy, and the calculation amount for reducing cloud platform reduces and operational efficiency raising.
Description
Technical field
The present invention relates to field of medical technology, in particular to a kind of interactive inline diagnosis cloud platform and its operation method and
Readable storage medium storing program for executing.
Background technique
Cardiovascular and cerebrovascular disease is the principal disease of China's death, and the statistical data of State Statistics Bureau is shown, from 2010
Year to 2014, heart disease and cerebrovascular disease in Died Of Disease reason, occupy second and third position always.Two kinds of diseases it is dead
It dies accounting summation of the population in population of always dying of illness and is always held at 40%.
Diabetes are global great public health problems, and threaten the major chronic illnesses of China's residents ' health.
It is reported that (18 years old or more) diabetes morbidity is 11.6% in China adult, about 1.139 hundred million people.DR(diabetic
Retinopathy, diabetic retinopathy) it is the important complication of diabetes and important blinding eye disease.Research shows that
Disease incidence of the DR in diabetic is 34.5%.The research report of American Diabetes Association claims, and diabetic compares non-saccharide
The blindness probability for urinating patient is 25 times high.It is estimated that the current DR patient in China is about 40,000,000 or more, it is the important blind property in China
Eye disease.The damage of visual performance caused by diabetes is not easy to reverse, and lesion is often in carry out sexual development, but mistake caused by diabetes
It is bright be it is preventable, key is early detection and takes effective treatment.
Hypertension is one of most common chronic disease and the most important risk factor of cardiovascular and cerebrovascular diseases, cerebral apoplexy, the heart
The main complication disability rate such as flesh infarct, heart failure and chronic kidney disease, lethality are high.Chinese residents nutrition in 2002 with
Investigation of health conditions data show, China 18 years old or more adult hypertension illness rate is 18.8%.According to Prevalence of Hypertension
Growth trend, China in 2012 15 years old or more crowd's Prevalence of Hypertension is about 24%.It estimates accordingly, the existing hypertension in the whole nation is suffered from
About 2.7 hundred million people of person (patient not gone to a doctor including illness but) currently, the disease incidence of China's hypertension there is no AUTHORITATIVE DATA, according to
The data of domestic different queue research calculate, the annual morbidity of China 40 years old or more crowd's hypertension is about 3%, annual new hair
Patient at least 18,000,000.Receive about the 1.2 hundred million for the treatment of in hypertensive patient it is estimated that now suffering from, in Basic medical and health institutions
The patient of management is about 86,000,000 people.It is suitble to classification diagnosis and treatment lower management in the hypertensive patient of second-class or above hospitals treatment
Patient account for about 60%.It was verified that hypertension is the disease that can prevent and control.Reduce the blood pressure water of hypertensive patient
It is flat, cardiovascular risk factors integrated management is carried out, cerebral apoplexy and the events of heart attack is can obviously reduce, significantly improves the existence of patient
Country and the Disease Spectrum of patient is effectively reduced in quality.
The interpretation of eye fundus image needs very high ophthalmology profession basis.According to statistics, ophthalmology doctor on record is registered in the whole nation at present
Raw quantity is only 3.2 ten thousand, but group (the eyeground pathological changes patient or potential patient) quantity for needing periodically to receive eyeground screening is high
Up to 92,400,000.Oculist and patient populations' ratio are 1:2888, and the pressure of so current oculist is larger, and because pressure
Power is big or the difference of professional standards, and there is also larger problems for the efficiency and accuracy of diagnosis and treatment.
Therefore it is really necessary to develop a kind of inline diagnosis platform, intelligent diagnostics are carried out to patient according to eye fundus image, to subtract
The pressure of light oculist, however, can exist computationally intensive if carrying out intelligent diagnostics to eye fundus image using artificial intelligence completely
And the problem that operational efficiency is lower.
Summary of the invention
The main object of the present invention is to provide a kind of interactive inline diagnosis cloud platform and its operation method and computer
Readable storage medium storing program for executing, it is intended to be combined by Artificial Diagnosis with intelligent diagnostics, to alleviate the pressure of oculist or promote eye
The efficiency of section doctor and the accuracy of diagnosis, and the calculation amount of interactive inline diagnosis cloud platform is reduced and operational efficiency
It improves.
To achieve the above object, the present invention proposes a kind of operation method of interactive inline diagnosis cloud platform, comprising:
The relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes what multispectral imaging device absorbed
Eye fundus image；
The eye fundus image is sent to primary diagnosis personnel, generates tentative diagnosis so that the primary diagnosis personnel diagnose
Information；
The tentative diagnosis information is obtained, and goes out convolution kernel according to the tentative diagnosis information matches；
Image procossing is carried out to the eye fundus image, extracts the pathological characters information in the eye fundus image, described image
Processing includes the convolutional neural networks algorithm process carried out using the convolution kernel；
The pathological characters information is matched in the database of cloud platform, obtains intelligent diagnostics information.
Preferably, the pathological characters information includes the vascular morphology on eyeground.
Preferably, the tentative diagnosis information includes lesion region information.
Preferably, the eye fundus image is sent to primary diagnosis personnel, is generated so that the primary diagnosis personnel diagnose
The step of tentative diagnosis information, comprising:
According to the relevant information of the object to be diagnosed, inquiry has been the primary diagnosis people that the object to be diagnosed diagnosed
Member；
The eye fundus image is sent to the primary diagnosis personnel diagnosed, for the primary diagnosis diagnosed
Personnel, which diagnose, generates the tentative diagnosis information.
Preferably obtain the tentative diagnosis information, and the step of going out convolution kernel according to the tentative diagnosis information matches
In, the convolution kernel includes the first convolution kernel and the second convolution kernel, wherein second convolution kernel is greater than first convolution
Core；
The step of convolutional neural networks algorithm process carried out using the convolution kernel:
The eye fundus image is checked using the first convolution and carries out the first convolution Processing with Neural Network, to extract the eyeground figure
The first pathological characters information as in；
The eye fundus image is checked using the second convolution and carries out the second convolution Processing with Neural Network, to extract the eyeground figure
The second pathological characters information as in；
Judge whether the similarity of the first pathological characters information and the two pathological characters information is greater than the spy of setting
Levy similarity threshold；
If so, using the second pathological characters information as the pathological characters information of eye fundus image；
If it is not, the eye fundus image is then sent to another primary diagnosis personnel, for another primary diagnosis personnel
Diagnosis generates another tentative diagnosis information, judges that another tentative diagnosis information is set with whether the tentative diagnosis information is greater than
Determine diagnostic message similarity: if so, matching according to another tentative diagnosis information with the intersection of the tentative diagnosis information
Another convolution kernel out, another convolution kernel include third convolution kernel and Volume Four product core, and the Volume Four product core is greater than described
Third convolution kernel, is respectively adopted the third convolution kernel and Volume Four product core carries out convolutional neural networks processing, with correspondence
Third pathological characters information and filatow-Dukes disease reason characteristic information are obtained, when the third pathological characters information and filatow-Dukes disease reason are special
When reference breath is greater than the characteristic similarity threshold value, then using filatow-Dukes disease reason characteristic information as the pathological characters of eye fundus image
Information；If it is not, then feeding back primary diagnosis exception information, and the eye fundus image is sent to expert's grade diagnostic personnel, for institute
It states expert's grade diagnostic personnel diagnosis and generates expert's grade diagnostic message.
Preferably, judge whether the similarity of the first pathological characters information and the two pathological characters information is greater than to set
Before the step of fixed characteristic similarity threshold value, comprising:
According to the tentative diagnosis information, characteristic similarity threshold value is matched from mapping table.
Preferably, the relevant information of the object to be diagnosed further includes Artificial Diagnosis information；
The pathological characters information is matched in the database of cloud platform, obtain intelligent diagnostics information the step of it
Afterwards, comprising:
The Artificial Diagnosis information is matched with the intelligent diagnostics information；
When the matching degree of the Artificial Diagnosis information and the intelligent diagnostics information is less than setting diagnosis matching degree, to doctor
Business end, which is sent, reminds diagnosis exception information.
Preferably, the relevant information of the object to be diagnosed includes the case history of the object to be diagnosed；
The relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes what multispectral imaging device absorbed
After the step of eye fundus image, further includes:
According to the case history of object to be diagnosed, pathological characters information, the Diseases diagnosis information, treatment side of object to be diagnosed are obtained
The identity information of method information and diagnostic personnel；
By the pathological characters information of the object to be diagnosed, Diseases diagnosis information, treatment method information and diagnostic personnel
Identity information, matched in the database of cloud platform, judge the pathological characters information, Diseases diagnosis information, treatment side
Whether the identity information of method information and diagnostic personnel matches；
If mismatching, is sent to medical matters end and remind case history exception information.
The present invention also provides a kind of computer readable storage medium, interaction is stored on the computer readable storage medium
The control program of the operation program of formula inline diagnosis cloud platform, the intelligence ring realizes interactive mode as described above when executing
The step of operation method of inline diagnosis cloud platform.
Another interactive inline diagnosis cloud platform of the present invention, including memory, processor and it is stored in the memory
Operation program that is upper and can running interactive inline diagnosis cloud platform on the processor, the interactive mode inline diagnosis cloud are flat
The operation program of platform is arranged for carrying out the step of operation method of interactive inline diagnosis cloud platform as described in any one of the above embodiments.
Technical solution of the present invention first passes through primary diagnosis personnel and carries out primary diagnosis, to generate primary diagnosis information, into
And go out convolution kernel according to the primary diagnosis information matches, then carried out at convolutional neural networks algorithm using the convolution kernel
Reason so directly matches compatible convolution kernel by primary diagnosis information, rather than uses and continuously attempt to calculate and obtain
Suitable convolution kernel (convolution kernel too intensive is big, and convolution kernel is too small to will lead to Character losing), can so make interactive mode exist
The calculation amount of radiodiagnosis x cloud platform reduces and operational efficiency improves, also, carries out including that convolutional neural networks are calculated to eye fundus image
Image procossing including method processing will then obtain pathological characters information and exist to obtain the pathological characters information in eye fundus image
It is matched in the database of cloud platform, to obtain corresponding intelligent diagnostics information, in this way, in the present invention can be according to eyeground
Image can be diagnosed with on-line intelligence, and to obtain intelligent diagnostics information, which can be sent to the confession of medical care end
With reference to also or can be used as tentative diagnosis as a result, also or the intelligent diagnostics information and Artificial Diagnosis information can be carried out
Matching, to promote the accuracy of Artificial Diagnosis information, in this way, technical solution provided by the invention can also alleviate oculist's
Pressure can promote the efficiency of oculist and the accuracy of diagnosis.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
The structure shown according to these attached drawings obtains other attached drawings.
Fig. 1 is the block schematic illustration of an embodiment of interactive inline diagnosis cloud platform provided by the invention；
Fig. 2 is the process signal of the first embodiment of interactive inline diagnosis cloud platform operation method provided by the invention
Figure；
Fig. 3 is the process signal of the second embodiment of interactive inline diagnosis cloud platform operation method provided by the invention
Figure；
Fig. 4 is the process signal of the 3rd embodiment of interactive inline diagnosis cloud platform operation method provided by the invention
Figure；
Fig. 5 is the process signal of the fourth embodiment of interactive inline diagnosis cloud platform operation method provided by the invention
Figure；
Fig. 6 is the process signal of the 5th embodiment of interactive inline diagnosis cloud platform operation method provided by the invention
Figure.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
His embodiment, shall fall within the protection scope of the present invention.
It is to be appreciated that the directional instruction (such as up, down, left, right, before and after ...) of institute is only used in the embodiment of the present invention
In explaining in relative positional relationship, the motion conditions etc. under a certain particular pose (as shown in the picture) between each component, if should
When particular pose changes, then directionality instruction also correspondingly changes correspondingly.
In addition, the description for being related to " first ", " second " etc. in the present invention is used for description purposes only, and should not be understood as referring to
Show or imply its relative importance or implicitly indicates the quantity of indicated technical characteristic." first ", " are defined as a result,
Two " feature can explicitly or implicitly include at least one of the features.In addition, the technical solution between each embodiment can
It to be combined with each other, but must be based on can be realized by those of ordinary skill in the art, when the combination of technical solution occurs
Conflicting or cannot achieve when, will be understood that the combination of this technical solution is not present, also not the present invention claims protection model
Within enclosing.
As shown in Figure 1, the interactive inline diagnosis cloud that Fig. 1 is the hardware running environment that the embodiment of the present invention is related to is put down
Platform structural schematic diagram.
As shown in Figure 1, the interactive mode inline diagnosis cloud platform includes: processor 1001, such as CPU, communication bus 1002,
Data-interface 1003 and memory 1004.Wherein, communication bus 1002 is for realizing the connection communication between these components.Data
Interface 1003 can also include wireline interface (such as USB interface or I/O interface), the wireless interface (such as WI-FI interface) of standard.
Memory 1004 can be high speed RAM memory, be also possible to stable memory (non-volatile memory), such as
Magnetic disk storage.Memory optionally can also be the storage device independently of aforementioned processor.
It will be understood by those skilled in the art that interactive mode inline diagnosis cloud platform structure shown in Fig. 1 is not constituted pair
The restriction of interactive inline diagnosis cloud platform may include components more more or fewer than diagram, or combine certain components, or
The different component layout of person.
As shown in Figure 1, as may include that operating system, data connect in a kind of memory 1004 of computer storage medium
Cause for gossip shows the operation program of program and interactive inline diagnosis cloud platform.
In interactive inline diagnosis cloud platform shown in Fig. 1, processor can be in interactive inline diagnosis cloud platform
Chip is controlled, which can be used for calling the operation program of the interactive inline diagnosis cloud platform stored in memory, and
Execute following operation:
The relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes what multispectral imaging device absorbed
Eye fundus image；
The eye fundus image is sent to primary diagnosis personnel, generates tentative diagnosis so that the primary diagnosis personnel diagnose
Information；
The tentative diagnosis information is obtained, and goes out convolution kernel according to the tentative diagnosis information matches；
Image procossing is carried out to the eye fundus image, extracts the pathological characters information in the eye fundus image, described image
Processing includes the convolutional neural networks algorithm process carried out using the convolution kernel；
The pathological characters information is matched in the database of cloud platform, obtains intelligent diagnostics information.
Wherein, the pathological characters information includes the vascular morphology on eyeground.
Wherein, the tentative diagnosis information includes lesion region information.
Further, processor 1001 can call the network operation stored in memory 1004 to control application program, also
Execute following operation:
The eye fundus image is sent to primary diagnosis personnel, generates tentative diagnosis so that the primary diagnosis personnel diagnose
The step of information, comprising:
According to the relevant information of the object to be diagnosed, inquiry has been the primary diagnosis people that the object to be diagnosed diagnosed
Member；
The eye fundus image is sent to the primary diagnosis personnel diagnosed, for the primary diagnosis diagnosed
Personnel, which diagnose, generates the tentative diagnosis information.
Further, processor 1001 can call the network operation stored in memory 1004 to control application program, also
Execute following operation:
The tentative diagnosis information is obtained, and in the step of going out convolution kernel according to the tentative diagnosis information matches, it is described
Convolution kernel includes the first convolution kernel and the second convolution kernel, wherein second convolution kernel is greater than first convolution kernel；
The step of convolutional neural networks algorithm process carried out using the convolution kernel:
The eye fundus image is checked using the first convolution and carries out the first convolution Processing with Neural Network, to extract the eyeground figure
The first pathological characters information as in；
The eye fundus image is checked using the second convolution and carries out the second convolution Processing with Neural Network, to extract the eyeground figure
The second pathological characters information as in；
Judge whether the similarity of the first pathological characters information and the two pathological characters information is greater than the spy of setting
Levy similarity threshold；
If so, using the second pathological characters information as the pathological characters information of eye fundus image；
If it is not, if it is not, the eye fundus image is then sent to another primary diagnosis personnel, for another primary diagnosis
Personnel, which diagnose, generates another tentative diagnosis information, judges whether another tentative diagnosis information and the tentative diagnosis information are big
In setting diagnostic message similarity: if so, according to the intersection of another tentative diagnosis information and the tentative diagnosis information
Another convolution kernel is matched, another convolution kernel includes third convolution kernel and Volume Four product core, and the Volume Four product core is greater than
The third convolution kernel, is respectively adopted the third convolution kernel and Volume Four product core carries out convolutional neural networks processing, with
It is corresponding to obtain third pathological characters information and filatow-Dukes disease reason characteristic information, when the third pathological characters information and the filatow-Dukes disease
When managing characteristic information greater than the characteristic similarity threshold value, then using filatow-Dukes disease reason characteristic information as the pathology of eye fundus image
Characteristic information；If it is not, then feeding back primary diagnosis exception information, and the eye fundus image is sent to expert's grade diagnostic personnel, with
Expert's grade diagnostic message is generated for expert's grade diagnostic personnel diagnosis.
Further, processor 1001 can call the network operation stored in memory 1004 to control application program, also
Execute following operation:
Judge whether the similarity of the first pathological characters information and the two pathological characters information is greater than the spy of setting
Before the step of levying similarity threshold, comprising:
According to the tentative diagnosis information, characteristic similarity threshold value is matched from mapping table.
Further, the relevant information of the object to be diagnosed further includes Artificial Diagnosis information, and processor 1001 can be adjusted
Application program is controlled with the network operation stored in memory 1004, also executes following operation:
The pathological characters information is matched in the database of cloud platform, obtain intelligent diagnostics information the step of it
Afterwards, comprising:
The Artificial Diagnosis information is matched with the intelligent diagnostics information；
When the matching degree of the Artificial Diagnosis information and the intelligent diagnostics information is less than setting diagnosis matching degree, to doctor
Business end, which is sent, reminds diagnosis exception information.
Further, processor 1001 can call the network operation stored in memory 1004 to control application program, also
Execute following operation:
The relevant information of the object to be diagnosed includes the case history of the object to be diagnosed；
The relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes what multispectral imaging device absorbed
After the step of eye fundus image, further includes:
According to the case history of object to be diagnosed, pathological characters information, the Diseases diagnosis information, treatment side of object to be diagnosed are obtained
The identity information of method information and diagnostic personnel；
By the pathological characters information of the object to be diagnosed, Diseases diagnosis information, treatment method information and diagnostic personnel
Identity information, matched in the database of cloud platform, judge the pathological characters information, Diseases diagnosis information, treatment side
Whether the identity information of method information and diagnostic personnel matches；
If mismatching, is sent to medical matters end and remind case history exception information.
The present invention proposes that a kind of operation method of interactive inline diagnosis cloud platform, Fig. 2 to Fig. 6 are friendship provided by the invention
The embodiment of the operation method of mutual formula inline diagnosis cloud platform.
Referring to Fig. 2, proposing a kind of first embodiment of the operation method of interactive inline diagnosis cloud platform in the present invention
In, the operation method of the interactive mode inline diagnosis cloud platform includes:
Step S10, the relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes multispectral imaging dress
Set the eye fundus image of intake；
Multispectral fundus imaging technology (Multiple Spectrum Imaging, MSI), can obtain than traditional eye-ground photography
Obtain deeper, wider, more accurate fundus tissue Pathological Information.Compared with traditional fundus imaging Technology application visible light is light source,
The technology is the tissue for projecting eyeground different depth (comprising RPE layers and choroid) respectively using multiple monochromatic LED light, using not
With the difference of tissue absorbance spectrum, the image of eyeground different depth is acquired, being formed includes retinal hemorrhage, glass-film
Wart, oxygen content, lipofuscin, pigment change etc. can reflect cardiovascular and cerebrovascular disease and systemic microcirculqtory system disease injury degree list
Color spectrum image.
Step S20, the eye fundus image is sent to primary diagnosis personnel, is generated so that the primary diagnosis personnel diagnose
Tentative diagnosis information；
Primary diagnosis personnel carry out primary diagnosis to the eye fundus image, are not related to specific disease and diagnostic method, only
It is to be related to the region of eyeground pathological changes, alternatively, the degree of eyeground pathological changes or stage etc..
Step S30, the tentative diagnosis information is obtained, and goes out convolution kernel according to the tentative diagnosis information matches；
Interactive inline diagnosis cloud platform storage has tentative diagnosis information mapping table corresponding with convolution kernel, for this purpose,
Once obtaining the tentative diagnosis information, then convolution kernel can be matched by the mapping table.
Step S40, image procossing is carried out to the eye fundus image, extracts the pathological characters information in the eye fundus image,
Described image processing includes the convolutional neural networks algorithm process carried out using the convolution kernel；
Convolutional neural networks algorithm is a kind of mature and continuous evolution intelligent algorithm, mainly to be included in image
Processing etc., by convolution to extract the characteristic information in image, and then can be according to characteristic information to be identified, herein
Specific introduction is not done.
Step S50, the pathological characters information is matched in the database of cloud platform, obtains intelligent diagnostics letter
Breath.
Technical solution of the present invention first passes through primary diagnosis personnel and carries out primary diagnosis, to generate primary diagnosis information, into
And go out convolution kernel according to the primary diagnosis information matches, then carried out at convolutional neural networks algorithm using the convolution kernel
Reason so directly matches compatible convolution kernel by primary diagnosis information, rather than uses and continuously attempt to calculate and obtain
Suitable convolution kernel (convolution kernel too intensive is big, and convolution kernel is too small to will lead to Character losing), can so make interactive mode exist
The calculation amount of radiodiagnosis x cloud platform reduces and operational efficiency improves, also, carries out including that convolutional neural networks are calculated to eye fundus image
Image procossing including method processing will then obtain pathological characters information and exist to obtain the pathological characters information in eye fundus image
It is matched in the database of cloud platform, to obtain corresponding intelligent diagnostics information, in this way, in the present invention can be according to eyeground
Image can be diagnosed with on-line intelligence, and to obtain intelligent diagnostics information, which can be sent to the confession of medical care end
With reference to also or can be used as tentative diagnosis as a result, also or the intelligent diagnostics information and Artificial Diagnosis information can be carried out
Matching, to promote the accuracy of Artificial Diagnosis information, in this way, technical solution provided by the invention can also alleviate oculist's
Pressure can promote the efficiency of oculist and the accuracy of diagnosis.
Intuitive observation eyeground form, which is undoubtedly, finds cardiovascular disease most straightforward approach.Academia compares high praise now
Be observing eye bottom retinal vessel form, retinal vessel can uniquely observe the device of parteriole and vein as whole body
Official, its lesion are able to reflect the lesion of system vascular, predict the transmission of cardiocerebrovasculaevents events.For this purpose, in the present embodiment, institute
State the vascular morphology that pathological characters information includes eyeground.Obvious the design is without being limited thereto, and the pathological characters information can also be
The texture of retina, tortuous degree, spatial relationship etc..
Referring to Fig. 3, in the second embodiment of the operation method of interactive inline diagnosis cloud platform provided by the invention,
Step S20 includes:
Step S21, according to the relevant information of the object to be diagnosed, it is first that inquiry be that the object to be diagnosed diagnosed
Grade diagnostic personnel；
Step S22, the eye fundus image is sent to the primary diagnosis personnel diagnosed, was diagnosed for described
Primary diagnosis personnel, which diagnose, generates the tentative diagnosis information.
It for example can be identity information by the relevant information of the diagnosis object, be also possible to medical record number information etc.
Deng the primary diagnosis personnel diagnosed for the object to be diagnosed can be inquired, and the primary diagnosis personnel diagnosed are opposite
It is known more about in the state of an illness of other primary diagnosis personnel to the object to be diagnosed, it can be quickly and accurately to the object to be diagnosed
Carry out tentative diagnosis.
Referring to Fig. 4, in the 3rd embodiment of the operation method of interactive inline diagnosis cloud platform provided by the invention,
In step s 30, the convolution kernel includes the first convolution kernel and the second convolution kernel, wherein second convolution kernel is greater than described
First convolution kernel；
In step S40, described image processing includes the steps that convolutional neural networks algorithm process:
Step S41, the eye fundus image is checked using the first convolution and carries out the first convolution Processing with Neural Network, to extract
State the first pathological characters information in eye fundus image；
Step S42, the eye fundus image is checked using the second convolution and carries out the second convolution Processing with Neural Network, to extract
State the second pathological characters information in eye fundus image；
Step S43, judge whether the first pathological characters information and the similarity of the two pathological characters information are greater than
The characteristic similarity threshold value of setting；
Step S44, if so, using the second pathological characters information as the pathological characters information of eye fundus image；
Step S45, if it is not, the eye fundus image is then sent to another primary diagnosis personnel, for another primary
Diagnostic personnel diagnosis generates another tentative diagnosis information, judges that another tentative diagnosis information and the tentative diagnosis information are
It is no to be greater than setting diagnostic message similarity: if so, according to another tentative diagnosis information and the tentative diagnosis information
Intersection matches another convolution kernel, and another convolution kernel includes third convolution kernel and Volume Four product core, the Volume Four product core
Greater than the third convolution kernel, the third convolution kernel is respectively adopted and Volume Four product core carries out at convolutional neural networks
Reason obtains third pathological characters information and filatow-Dukes disease reason characteristic information with corresponding, when the third pathological characters information and described
When filatow-Dukes disease manages characteristic information greater than the characteristic similarity threshold value, then using filatow-Dukes disease reason characteristic information as eye fundus image
Pathological characters information；If it is not, then feeding back primary diagnosis exception information, and the eye fundus image is sent to expert grade diagnosis people
Member, so that expert's grade diagnostic personnel diagnosis generates expert's grade diagnostic message.
When eye fundus image is carried out convolutional neural networks algorithm process, the more big corresponding accuracy of information of convolution kernel is more
Height, but its calculation amount is also bigger simultaneously, the smaller calculation amount of convolution kernel is smaller, but the problem of exist simultaneously the omission of characteristic information,
So in the present embodiment, obtaining pathological characters information (the i.e. first disease according to the convolutional neural networks algorithm process of front and back 2 times
Reason characteristic information and the second pathological characters information) between similarity (diversity factor in other words) come carry out determine use convolution it is big
It is small whether suitable, when similarity reaches characteristic similarity threshold value (such as 98%), that is, indicate at convolutional neural networks algorithm twice
It is little that reason obtains pathological characters information difference, namely indicates to carry out the acquisition of convolutional neural networks algorithm process after increasing convolution kernel again
Pathological characters information has little significance, i.e. the convolution kernel size of expression front is suitable, when similarity does not reach characteristic similarity
Threshold value, that is, indicate twice convolutional neural networks algorithm process obtain that pathological characters information difference is larger namely the first convolution kernel and
Second convolution kernel is improper, the primary diagnosis inaccuracy of primary diagnosis personnel is likely due at this time, for this purpose, needing to lead at this time
It crosses another primary diagnosis personnel and carries out primary diagnosis, rather than directly to expert's grade diagnostic personnel (because of expert's grade diagnostic personnel
It is more rare), it is compared by another primary diagnosis personnel with the primary diagnosis information of primary diagnosis personnel before, when two
When person's similarity is higher, illustrate that there is no problem for diagnosis, and then the intersection of the two is subjected to matching convolution kernel, because of the intersection of the two
Relative to any, more comprehensively, the convolution kernel accuracy matched is higher for information, and then subsequent progress convolutional Neural net twice
Network algorithm process, with improve examine diagnostic accuracy, when the two similarity is lower, illustrate to diagnose it is problematic, then at this time directly general
The eye fundus image is sent to expert's grade diagnostic personnel, so that expert's grade diagnostic personnel diagnosis generates expert's grade diagnosis letter
Breath so reduces the pressure to expert's grade diagnostic personnel.
Certainly, in an embodiment of the present invention, when the third pathological characters information and the filatow-Dukes disease manage characteristic information
When less than the characteristic similarity threshold value, then primary diagnosis exception information can also be fed back, and the eye fundus image is sent to specially
Family's grade diagnostic personnel, so that expert's grade diagnostic personnel diagnosis generates expert's grade diagnostic message.
It should be noted that " characteristic similarity threshold value " be not fixed value, particular situation can be regarded and different, such as can be with
And difference different according to the classification of the pathological characters information of object to be diagnosed, can also be according to specifically object to be diagnosed
Depending on, it is also or different according to the specifying information amount of pathological characters information, it can also be according to the history case history of diagnosis object
Data are determined, mainly the stage according to locating for the disease before diagnosis object, for certain stage diseases generation in other words
Lesion is obvious, at this point it is possible to using lesser characteristic similarity threshold value such as 80%, and be not for certain stage lesions
It is obvious that being needed at this time using biggish characteristic similarity such as 95%, namely it must assure that the accuracy of feature extraction.At this
In the embodiment of invention, before step S43 further include: according to the tentative diagnosis information, spy is matched from mapping table
Levy similarity threshold.Namely interactive inline diagnosis cloud platform is also stored with tentative diagnosis information and the mapping of similarity threshold is closed
It is table, the similarity threshold can be gone out by tentative diagnosis information matches.
Referring to Fig. 5, in the fourth embodiment of the operation method of interactive inline diagnosis cloud platform provided by the invention,
In the present embodiment, the relevant information of the object to be diagnosed further includes Artificial Diagnosis information；
After step S50, comprising:
Step S60, the Artificial Diagnosis information is matched with the intelligent diagnostics information；
Step S70, it is matched when the Artificial Diagnosis information is less than setting diagnosis with the matching degree of the intelligent diagnostics information
When spending, is sent to medical matters end and remind diagnosis exception information.
In the present embodiment, it is compared by the intelligent diagnostics of interactive inline diagnosis cloud platform with Artificial Diagnosis, when
When the two differs greatly, the effect for reminding diagnosis exception information that medical worker can be improved is sent to medical matters end and reduces error
Rate.
It should be noted that the diagnostic message includes disease type, disease duration and the treatment method of diagnosis, as long as its
In any difference it is larger, line intelligent diagnostics cloud platform all can to medical matters end send diagnosis exception information.
Referring to Fig. 6, in the 5th embodiment of the operation method of interactive inline diagnosis cloud platform provided by the invention,
In the present embodiment, the relevant information of the object to be diagnosed includes the case history of the object to be diagnosed；
After step S10, further includes:
Step S11, according to the case history of object to be diagnosed, pathological characters information, the Diseases diagnosis letter of object to be diagnosed are obtained
The identity information of breath, treatment method information and diagnostic personnel；
Step S12, by the pathological characters information of the object to be diagnosed, Diseases diagnosis information, treatment method information and
The identity information of diagnostic personnel is matched in the database of cloud platform, judges the pathological characters information, Diseases diagnosis letter
Whether the identity information of breath, treatment method information and diagnostic personnel matches；
If step S13, mismatching, is sent to medical matters end and remind case history exception information.
In the present embodiment, in the database of interactive inline diagnosis cloud platform to pathological characters information and with it is corresponding
Mapping is established between the diagnostic personnel of the corresponding disease type of pathological characters information, methods for the diagnosis of diseases and corresponding professional ability
Relationship, to avoid the erroneous judgement that Artificial Diagnosis is sent, when there is case history input, the case history by obtaining object to be diagnosed (including is gone through
History case history and case history currently entered) in pathological characters information, Diseases diagnosis information, treatment method information and diagnostic personnel
Identity information, by the pathological characters information of the object to be diagnosed, Diseases diagnosis information, treatment method information and diagnostic personnel
Identity information, matched in the database of cloud platform, judge the pathological characters information, Diseases diagnosis information, treatment side
Whether the identity information of method information and diagnostic personnel matches, and when mismatching, illustrates the case where being likely that there are mistaken diagnosis, in turn
Case history exception information is reminded to diagnostic personnel.
The present invention also proposes a kind of computer readable storage medium, is stored with intelligence on the computer readable storage medium
The control program of the control program of ring, the intelligence ring realizes the interaction such as above-mentioned any embodiment when being executed by processor
The step of operation method of formula inline diagnosis cloud platform.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the system that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or system institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or system.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in a storage as above
In medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, calculate
Machine, server, television set or network equipment etc.) method that executes each embodiment of the present invention.
The above description is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all at this
Under the inventive concept of invention, using equivalent structure transformation made by description of the invention and accompanying drawing content, or directly/use indirectly
It is included in other related technical areas in scope of patent protection of the invention.
Claims (10)
1. a kind of operation method of interactive mode inline diagnosis cloud platform characterized by comprising
The relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes the eyeground of multispectral imaging device intake
Image；
The eye fundus image is sent to primary diagnosis personnel, generates tentative diagnosis letter so that the primary diagnosis personnel diagnose
Breath；
The tentative diagnosis information is obtained, and goes out convolution kernel according to the tentative diagnosis information matches；
Image procossing is carried out to the eye fundus image, extracts the pathological characters information in the eye fundus image, described image processing
Including the convolutional neural networks algorithm process carried out using the convolution kernel；
The pathological characters information is matched in the database of cloud platform, obtains intelligent diagnostics information.
2. the operation method of interactive mode inline diagnosis cloud platform as described in claim 1, which is characterized in that the pathological characters
Information includes the vascular morphology on eyeground.
3. the operation method of interactive mode inline diagnosis cloud platform as claimed in claim 2, which is characterized in that the tentative diagnosis
Information includes lesion region information.
4. the operation method of interactive mode inline diagnosis cloud platform as described in claim 1, which is characterized in that by the eyeground figure
As being sent to primary diagnosis personnel, so that the primary diagnosis personnel diagnose the step of generating tentative diagnosis information, comprising:
According to the relevant information of the object to be diagnosed, inquiry has been the primary diagnosis personnel that the object to be diagnosed diagnosed；
The eye fundus image is sent to the primary diagnosis personnel diagnosed, for the primary diagnosis personnel diagnosed
Diagnosis generates the tentative diagnosis information.
5. the operation method of the interactive inline diagnosis cloud platform as described in Claims 1-4 any one, which is characterized in that
The tentative diagnosis information is obtained, and in the step of going out convolution kernel according to the tentative diagnosis information matches, the convolution kernel packet
Include the first convolution kernel and the second convolution kernel, wherein second convolution kernel is greater than first convolution kernel；
The step of convolutional neural networks algorithm process carried out using the convolution kernel:
The eye fundus image is checked using the first convolution and carries out the first convolution Processing with Neural Network, to extract in the eye fundus image
The first pathological characters information；
The eye fundus image is checked using the second convolution and carries out the second convolution Processing with Neural Network, to extract in the eye fundus image
The second pathological characters information；
Judge whether the similarity of the first pathological characters information and the two pathological characters information is greater than the feature phase of setting
Like degree threshold value；
If so, using the second pathological characters information as the pathological characters information of eye fundus image；
If it is not, the eye fundus image is then sent to another primary diagnosis personnel, so that another primary diagnosis personnel diagnose
Another tentative diagnosis information is generated, judges that another tentative diagnosis information is examined with whether the tentative diagnosis information is greater than setting
Disconnected information similarity: if so, being matched separately according to the intersection of another tentative diagnosis information and the tentative diagnosis information
One convolution kernel, another convolution kernel include third convolution kernel and Volume Four product core, and the Volume Four product core is greater than the third
Convolution kernel, is respectively adopted the third convolution kernel and Volume Four product core carries out convolutional neural networks processing, is obtained with correspondence
Third pathological characters information and filatow-Dukes disease manage characteristic information, when the third pathological characters information and filatow-Dukes disease reason feature letter
When breath is greater than the characteristic similarity threshold value, then believe using filatow-Dukes disease reason characteristic information as the pathological characters of eye fundus image
Breath；If it is not, then feeding back primary diagnosis exception information, and the eye fundus image is sent to expert's grade diagnostic personnel, for described
Expert's grade diagnostic personnel diagnosis generates expert's grade diagnostic message.
6. the operation method of interactive mode inline diagnosis cloud platform as claimed in claim 5, which is characterized in that judge described first
The step of whether similarity of pathological characters information and the two pathological characters information is greater than the characteristic similarity threshold value of setting it
Before, comprising:
According to the tentative diagnosis information, characteristic similarity threshold value is matched from mapping table.
7. the operation method of interactive mode inline diagnosis cloud platform as described in claim 1, which is characterized in that described wait diagnose pair
The relevant information of elephant further includes Artificial Diagnosis information；
After the step of pathological characters information is matched in the database of cloud platform, obtains intelligent diagnostics information,
Include:
The Artificial Diagnosis information is matched with the intelligent diagnostics information；
When the matching degree of the Artificial Diagnosis information and the intelligent diagnostics information is less than setting diagnosis matching degree, to medical matters end
It sends and reminds diagnosis exception information.
8. the operation method of interactive mode inline diagnosis cloud platform as described in claim 1, which is characterized in that described wait diagnose pair
The relevant information of elephant includes the case history of the object to be diagnosed；
The relevant information of object to be diagnosed is received, the relevant information to be diagnosed includes the eyeground of multispectral imaging device intake
After the step of image, further includes:
According to the case history of object to be diagnosed, pathological characters information, Diseases diagnosis information, the treatment method letter of object to be diagnosed are obtained
The identity information of breath and diagnostic personnel；
By the body of the pathological characters information of the object to be diagnosed, Diseases diagnosis information, treatment method information and diagnostic personnel
Part information, is matched in the database of cloud platform, judges the pathological characters information, Diseases diagnosis information, treatment method letter
Whether breath and the identity information of diagnostic personnel match；
If mismatching, is sent to medical matters end and remind case history exception information.
9. a kind of computer readable storage medium, which is characterized in that be stored with interactive mode on the computer readable storage medium
The control program of the operation program of inline diagnosis cloud platform, the intelligence ring is realized when executing as any in claim 1 to 8
The step of operation method of interactive inline diagnosis cloud platform described in.
10. a kind of interactive mode inline diagnosis cloud platform, which is characterized in that including memory, processor and be stored in the storage
The operation program of interactive inline diagnosis cloud platform, the interactive mode inline diagnosis cloud can be run on device and on the processor
The operation program of platform is arranged for carrying out the fortune such as interactive inline diagnosis cloud platform described in any item of the claim 1 to 8
The step of row method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910065957.7A CN109767828B (en) | 2019-01-23 | 2019-01-23 | Interactive online diagnosis cloud platform, operation method thereof and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910065957.7A CN109767828B (en) | 2019-01-23 | 2019-01-23 | Interactive online diagnosis cloud platform, operation method thereof and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109767828A true CN109767828A (en) | 2019-05-17 |
| CN109767828B CN109767828B (en) | 2023-09-12 |
Family
ID=66454290
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910065957.7A Active CN109767828B (en) | 2019-01-23 | 2019-01-23 | Interactive online diagnosis cloud platform, operation method thereof and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109767828B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110660488A (en) * | 2019-10-31 | 2020-01-07 | 深圳盛达同泽科技有限公司 | Sugar network screening method and platform based on multispectral fundus imaging and storage medium |
| CN111309950A (en) * | 2020-01-18 | 2020-06-19 | 青柠知享科技（深圳）有限公司 | E-commerce transaction information interaction method, e-commerce platform and readable storage medium |
| CN111477309A (en) * | 2020-03-02 | 2020-07-31 | 深圳市新产业眼科新技术有限公司 | Private cloud access method based on big data analysis, public cloud and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106725292A (en) * | 2016-12-30 | 2017-05-31 | 深圳市新产业眼科新技术有限公司 | Multispectral fundus imaging remote diagnosis system and its operation method |
| CN107423571A (en) * | 2017-05-04 | 2017-12-01 | 深圳硅基仿生科技有限公司 | Diabetic retinopathy identifying system based on eye fundus image |
| CN108345818A (en) * | 2017-01-23 | 2018-07-31 | 北京中科奥森数据科技有限公司 | A kind of human face in-vivo detection method and device |
-
2019
- 2019-01-23 CN CN201910065957.7A patent/CN109767828B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106725292A (en) * | 2016-12-30 | 2017-05-31 | 深圳市新产业眼科新技术有限公司 | Multispectral fundus imaging remote diagnosis system and its operation method |
| CN108345818A (en) * | 2017-01-23 | 2018-07-31 | 北京中科奥森数据科技有限公司 | A kind of human face in-vivo detection method and device |
| CN107423571A (en) * | 2017-05-04 | 2017-12-01 | 深圳硅基仿生科技有限公司 | Diabetic retinopathy identifying system based on eye fundus image |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110660488A (en) * | 2019-10-31 | 2020-01-07 | 深圳盛达同泽科技有限公司 | Sugar network screening method and platform based on multispectral fundus imaging and storage medium |
| CN111309950A (en) * | 2020-01-18 | 2020-06-19 | 青柠知享科技（深圳）有限公司 | E-commerce transaction information interaction method, e-commerce platform and readable storage medium |
| CN111477309A (en) * | 2020-03-02 | 2020-07-31 | 深圳市新产业眼科新技术有限公司 | Private cloud access method based on big data analysis, public cloud and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109767828B (en) | 2023-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107423571B (en) | Diabetic retinopathy identifying system based on eye fundus image | |
| KR101977645B1 (en) | Eye image analysis method | |
| CN110101361A (en) | Based on big data on-line intelligence diagnostic platform and its operation method and storage medium | |
| CN107358605B (en) | The deep neural network apparatus and system of diabetic retinopathy for identification | |
| CN106682389B (en) | A kind of Eye disease for monitoring hypertension initiation is health management system arranged | |
| CN109686440A (en) | A kind of on-line intelligence diagnosis cloud platform and its operation method and readable storage medium storing program for executing | |
| US20210391056A1 (en) | Health big data service method and system based on remote fundus screening | |
| CN109767828A (en) | Interactive inline diagnosis cloud platform and its operation method and readable storage medium storing program for executing | |
| Liu et al. | Computerized macular pathology diagnosis in spectral domain optical coherence tomography scans based on multiscale texture and shape features | |
| CN109411086A (en) | The screening system of eye fundus image lesion | |
| CN116703915B (en) | Diabetes retinopathy consultation method and system based on multiple image fusion | |
| CN112233087A (en) | Artificial intelligence-based ophthalmic ultrasonic disease diagnosis method and system | |
| JP2022084511A (en) | Image processing system and image processing method | |
| CN111292286B (en) | Analysis method and system for change of characteristic data of sugar mesh bottom and storage device | |
| CN111402184B (en) | Method and system for realizing remote fundus screening and health service | |
| CN111383775A (en) | Member user anonymous communication method based on fundus screening and storage device | |
| Tan et al. | Is routine pupil dilation safe among asian patients with diabetes? | |
| CN111435612B (en) | Method and system for personalized health service of mobile medical treatment | |
| CN208969732U (en) | The screening system of eye fundus image lesion based on artificial intelligence | |
| Lim et al. | A review of machine learning techniques for applied eye fundus and tongue digital image processing with diabetes management system | |
| Biswas et al. | Investigation of bilateral similarity in central retinal blood vessels | |
| CN111325631A (en) | Health insurance fee control method and system based on fundus images | |
| CN117936079B (en) | Manifold learning-based diabetic retinopathy identification method, medium and system | |
| Biswas et al. | A study of bilateral symmetry in color fundus photographs | |
| Ashwin et al. | Using Fundus Photography for Early Glaucoma Diagnosis by VGG-16 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |