CN109859836A - Medical image recognition method and device - Google Patents
Medical image recognition method and device Download PDFInfo
- Publication number
- CN109859836A CN109859836A CN201910036173.1A CN201910036173A CN109859836A CN 109859836 A CN109859836 A CN 109859836A CN 201910036173 A CN201910036173 A CN 201910036173A CN 109859836 A CN109859836 A CN 109859836A
- Authority
- CN
- China
- Prior art keywords
- vector
- medical image
- machine learning
- result
- confidence level
- 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
- 238000000034 method Methods 0.000 title claims abstract description 40
- 239000013598 vector Substances 0.000 claims abstract description 151
- 238000010801 machine learning Methods 0.000 claims abstract description 46
- 201000010099 disease Diseases 0.000 claims abstract description 32
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 32
- 230000002159 abnormal effect Effects 0.000 claims abstract description 22
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 208000010412 Glaucoma Diseases 0.000 description 1
- 208000032843 Hemorrhage Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 208000034158 bleeding Diseases 0.000 description 1
- 230000000740 bleeding effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 208000030533 eye disease Diseases 0.000 description 1
- 201000011066 hemangioma Diseases 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention provides a medical image identification method and equipment, wherein the method comprises the following steps: acquiring a medical image; classifying the medical image by utilizing a first machine learning model to obtain a first vector, wherein the first vector represents a first confidence coefficient that the type of the medical image is healthy or abnormal; classifying the medical image by using a second machine learning model to obtain a second vector, wherein the second vector represents a second confidence coefficient that the type of the medical image is various disease types; obtaining a third vector according to the first vector and the second vector; and obtaining the identification result of the medical image according to the third vector.
Description
Technical field
The present invention relates to medical image process fields, and in particular to a kind of medical image recognition methods and equipment.
Background technique
By machine learning algorithm and model image is carried out identification be it is a kind of it is efficient in the way of, and such as drive automatically
It sails, the Floor layer Technology of the various fields such as intelligent camera, robot.
Medical image can usually reflect a variety of disease types, such as eye fundus image can embody hemangioma, eyeground
A variety of eye diseases such as bleeding, glaucoma.Medical image is identified using machine learning model (such as neural network), first has to make
Model is trained with sample image, the pictures for needing largely to mark when training, it is higher to the degree of dependence of mark personnel.
Country variant, city, doctor define standard difference to some diseases, and mark proficiency is irregular, the image matter marked
Amount difference is bigger, and mistake and not accurate enough marked content can generate data noise.
The existing medical image identifying schemes based on artificial intelligence are obtained using the model that these mark images train
Recognition result, is influenced bigger by data noise, and recognition result accuracy is not high enough, or even will cause the mistake of model training
It loses.
Summary of the invention
In view of this, the present invention provides a kind of medical image recognition methods, comprising:
Obtain medical image；
The medical image is classified using the first machine learning model to obtain primary vector, the primary vector table
The type for showing the medical image is health or the first abnormal confidence level；
The medical image is classified using the second machine learning model to obtain secondary vector, the secondary vector table
The type for showing the medical image is the second confidence level of various disease types；
Third vector is obtained according to the primary vector and the secondary vector；
The recognition result to the medical image is obtained according to the third vector.
It is optionally, described that third vector is obtained according to the primary vector and the secondary vector, comprising:
Splice the primary vector and the secondary vector obtains third vector, the third vector is by first confidence
Degree and second reliability composition.
It is optionally, described that the recognition result to the medical image is obtained according to the third vector, comprising:
The third vector is classified using third machine learning model to obtain classification results, the classification results are
It is normal or abnormal；
The recognition result to the medical image is determined according to the classification results.
Optionally, the recognition result determined according to the classification results to the medical image, comprising:
Judge whether the classification results are normal；
When classification results are normal, the classification results are exported.
Optionally, when classification results are abnormal, the secondary vector is exported.
It is optionally, described that third vector is obtained according to the primary vector and the secondary vector, comprising:
The first numerical value is taken in the primary vector according to default first ratio；
Second value is taken in the secondary vector according to default second ratio；
First numerical value and the second value are summed to obtain third vector.
It is optionally, described that the recognition result to the medical image is obtained according to the third vector, comprising:
The third vector is compared with given threshold；
The recognition result to the medical image is determined according to comparison result.
Optionally, the recognition result determined according to comparison result to the medical image, comprising:
Judge whether the third vector is less than the given threshold；
When the third vector is less than the given threshold, determine recognition result for health.
Optionally, when the third vector is greater than or equal to the given threshold, the secondary vector is exported.
The present invention also provides a kind of medical image identification devices, comprising:
Acquiring unit, for obtaining medical image；
First machine learning model obtains primary vector, the primary vector for being classified to the medical image
The type for indicating the medical image is health or the first abnormal confidence level；
Second machine learning model obtains secondary vector, the secondary vector for being classified to the medical image
The type for indicating the medical image is the second confidence level of various disease types；
Integrated unit, for obtaining third vector according to the primary vector and the secondary vector；
Determination unit, for obtaining the recognition result to the medical image according to the third vector.
The present invention also provides a kind of medical images to identify equipment, comprising: at least one processor；And with it is described at least
The memory of one processor communication connection；Wherein, the memory is stored with the instruction that can be executed by one processor,
Described instruction is executed by least one described processor, so that at least one described processor executes above-mentioned medical image identification side
Method.
The medical image recognition methods provided according to embodiments of the present invention and equipment are distinguished by two machine learning models
Same medical image is identified, two machine learning models purposes therein is different, can divide in training machine model
Safety pin is trained different purposes using different sample datas, to obtain the higher model of accuracy.One of mould
Type carries out two classification to medical image, obtains it and belongs to health or abnormal confidence level, without identifying specific Exception Type；Separately
One model carries out more classification to medical image, is absorbed in identification Exception Type, obtains its confidence level for belonging to various diseases.It is right
The two classification results are merged, and finally obtain the recognition result to medical image according to fusion results, are made an uproar with alleviating data
Interference of the sound to recognition result, it is possible thereby to improve the accuracy of medical image identification.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
It puts, is also possible to obtain other drawings based on these drawings.
Fig. 1 is the flow chart of one of embodiment of the present invention medical image recognition methods；
Fig. 2 is the flow chart of another medical image recognition methods in the embodiment of the present invention；
Fig. 3 is the structural schematic diagram of the medical image identification device in the embodiment of the present invention.
Specific embodiment
Technical solution of the present invention is clearly and completely described below in conjunction with attached drawing, it is clear that described implementation
Example is a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill
Personnel's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
As long as in addition, the non-structure each other of technical characteristic involved in invention described below different embodiments
It can be combined with each other at conflict.
The embodiment of the invention provides a kind of medical image recognition methods, this method can be electric by computer and server etc.
Sub- equipment executes.Machine learning model has been used to identify image in the method, the machine learning model can be multiple types
The neural network of type and structure.This method comprises the following steps as shown in Figure 1:
S1A obtains medical image.The image can be single channel image, such as CT (Computed Tomography, electricity
Sub- computed tomography) image, ultrasound examination image etc.；The image is also possible to multichannel image, e.g. eyeground
Photo etc..In some preferred embodiments, medical image can also be pre-processed, for example, in clip image it is unnecessary
Background, improve the operations such as the contrast of image, convert color spaces and the removal unnecessary region in part.
S2A classifies medical image using the first machine learning model to obtain primary vector, and primary vector indicates doctor
The type for treating image is health or the first abnormal confidence level.
Sample data should be used to be trained it before using the first machine learning model, and to have it certain
Classification capacity.The medical image for being largely known as health specifically can be used and be known as abnormal medical image to initial mould
Type is trained, and training data includes abnormal and normal two classifications, and the model after training can be defeated according to input picture
A vector out, only one numerical value in the vector, the numerical value indicate that the type of medical image is health or abnormal confidence
Degree, the range of the confidence level is between zero and one.
The first machine learning model in this step only needs to carry out medical image block two classification, exports normal confidence
One in degree or confidence level both classification results of exception, the classification results of output may be any one between 0 and 1
A numerical value, rather than deterministic conclusion.
Such as first machine learning model output primary vector be [0.6], 0.6 expression medical image belong to health confidence
Degree is 60%；Or 0.6 expression medical image belong to abnormal confidence level be 60%.It presets and trains in the present embodiment
First machine learning makes it export the confidence level for indicating type exception, therefore output result indicates the medical image category
In abnormal confidence level be 60% (belonging to normal confidence level is 40%).
S3A classifies medical image using the second machine learning model to obtain secondary vector, and secondary vector indicates doctor
The type for treating image is the second confidence level of various disease types.Sample should be used before using the second machine learning model
Data, which are trained it, makes it have certain classification capacity.A large amount of known medical treatment with genius morbi specifically can be used
Image is trained initial model, and may include a variety of genius morbis in a medical image for training.Through
The model crossed after training can export a vector according to input picture, have multiple numerical value in the vector, these numerical value distinguish table
Show that medical image belongs to the confidence level of various disease types, the range of each confidence level is between zero and one.
The second machine learning model in this step needs to carry out medical image block to classify more, in the classification results of output
Each numerical value respectively may be any one numerical value between 0 and 1, rather than deterministic conclusion.
Such as second machine learning model output secondary vector be [0.1,0.5,0.3,0.4,0.5], wherein first number
The confidence level that value 0.1 indicates that medical image belongs to the first disease type is that 10%, second numerical value 0.5 indicates medical image category
It is that 50%, third numerical value 0.3 indicates that medical image belongs to the third disease type in the confidence level of second of disease type
Confidence level is that indicate that medical image belongs to the confidence level of the 4th kind of disease type be 40%, the 5th for 30%, the 4th numerical value 0.4
The confidence level that numerical value 0.5 indicates that medical image belongs to the 5th kind of disease type is 50%.
S4A obtains third vector according to primary vector and secondary vector.There are many ways to merging two vectors, such as
Be added, subtract each other, splice etc..The present embodiment uses a kind of preferred embodiment, by the vector of the output of two models with one
Fixed ratio sums to obtain third vector.
By taking above-mentioned predicted vector as an example, 50% is taken to obtain the first numerical value the primary vector of the first machine learning model output
0.3 (0.6 × 50%=0.3)；10% is taken to obtain multiple second values the secondary vector of the second machine learning model output
0.01 (0.1 × 10%=0.01), 0.05 (0.5 × 10%=0.05), 0.03 (0.3 × 10%=0.03), 0.04 (0.4 ×
10%=0.04), 0.05 (0.5 × 10%=0.05), the first numerical value and multiple second values sum to obtain third vector 0.3+
0.01+0.05+0.03+0.04+0.05=0.48, third vector are [0.48].
In the above citing, it is 50% to the ratio that primary vector carries out value, carries out the ratio of value to secondary vector
It is 10%, this is intended merely to clearly demonstrate example rather than comparative example is defined, according to practical feelings in practical application
Condition sets the two value ratios.
S5A obtains the recognition result to medical image according to third vector.Classifier can be used for example to third vector
Classified to obtain recognition result, or determines that recognition result, final recognition result can be healthy, different using threshold method
Often, belong to a certain or a variety of diseases or belong to the confidence level of various diseases.
The medical image recognition methods provided according to embodiments of the present invention, by two machine learning models respectively to same
Medical image is identified that two machine learning models purposes therein is different, can be directed to respectively in training machine model
Different purposes is trained using different sample datas, to obtain the higher model of accuracy.One of model is to doctor
It treats image and carries out two classification, obtain it and belong to health or abnormal confidence level, without identifying specific Exception Type；Another mould
Type carries out more classification to medical image, is absorbed in identification Exception Type, obtains its confidence level for belonging to various diseases.To the two
Classification results are merged, and finally obtain the recognition result to medical image according to fusion results, to alleviate data noise to knowledge
The interference of other result, it is possible thereby to improve the accuracy of medical image identification.
The present embodiment determines recognition result using a kind of preferred mode.Specifically, judge whether third vector is less than to set
Determine threshold value, determines that recognition result is health if third vector is less than given threshold.Such as given threshold be 0.5, third to
Numerical value 0.48 in amount then finally determines that the medical image is healthy, it is not necessary to judge again secondary vector less than 0.5.
If third vector is greater than or equal to given threshold, can only an output recognition result be it is abnormal, the can also be exported
Two vectors are used to indicate that the medical image belongs to the confidence level of various diseases, or can be further to each in secondary vector
Numerical value is judged, is exported the medical image and is belonged to certain disease qualitative conclusions really.
It is above-mentioned by threshold decision obtain final recognition result in the way of computational efficiency with higher, and it is relatively straight
It sees；For the medical image of exception class, its confidence level for belonging to various diseases is finally exported, provides the reference number of quantization for user
Value, compared to deterministic conclusion is exported, the auxiliaring effect for exporting confidence information is stronger.
The embodiment of the invention provides another medical image recognition methods, this method can be by computer and server etc.
Electronic equipment executes.Machine learning model has been used to identify image in the method, the machine learning model can be a variety of
The neural network of type and structure.This method comprises the following steps as shown in Figure 2:
S1B obtains medical image.It specifically can refer to the step S1A in above-described embodiment, repeated no more in the present embodiment.
S2B classifies medical image using the first machine learning model to obtain primary vector, and primary vector indicates doctor
The type for treating image is health or the first abnormal confidence level.It specifically can refer to the step S2A in above-described embodiment, this implementation
It is repeated no more in example.It presets in the present embodiment and the first machine learning is trained to make its output for indicating that type is normal
Confidence level, such as primary vector is [0.4], then it represents that it is 40% (to belong to exception that the medical image, which belongs to normal confidence level,
Confidence level be 60%).
S3B classifies medical image using the second machine learning model to obtain secondary vector, and secondary vector indicates doctor
The type for treating image is the second confidence level of various disease types.It specifically can refer to the step S3A in above-described embodiment, this implementation
It is repeated no more in example.
Such as second machine learning model output secondary vector be [0.1,0.5,0.3,0.4,0.5], wherein first number
The confidence level that value 0.1 indicates that medical image belongs to the first disease type is that 10%, second numerical value 0.5 indicates medical image category
It is that 50%, third numerical value 0.3 indicates that medical image belongs to the third disease type in the confidence level of second of disease type
Confidence level is that indicate that medical image belongs to the confidence level of the 4th kind of disease type be 40%, the 5th for 30%, the 4th numerical value 0.4
The confidence level that numerical value 0.5 indicates that medical image belongs to the 5th kind of disease type is 50%.
S4B is spliced the vector that two models export to obtain third vector.By taking above-mentioned predicted vector as an example, splicing
Obtain later third vector be [0.4,0.1,0.5,0.3,0.4,0.5], wherein first numerical value 0.4 be in primary vector only
One numerical value, the second to six numerical value are whole numerical value in secondary vector.
In the above citing, by the numerical value of primary vector be located at first place be intended merely to clearly demonstrate example rather than
Connecting method is defined, specific connecting method is set according to actual conditions in practical application, and correspondingly adjusts subsequent calculation
Method.
S5B classifies third vector using third machine learning model to obtain classification results, and classification results are normal
Or it is abnormal.Sample data should be used to be trained it before using third machine learning model, and to have it certain
Classification capacity.The medical image and its corresponding third vector that a large amount of known states specifically can be used carry out initial model
Training, the model after training can differentiate medical image according to input vector and belong to healthy class or exception class.With first
Unlike machine learning model, the output of third machine learning model is deterministic conclusion.Such as third machine learning mould
Type can be a SVM classifier, and the input of the classifier is third vector [0.4,0.1,0.5,0.3,0.4,0.5], output
It is 0 or 1, such as 0 expression health, 1 expression exception.
S6B determines the recognition result to medical image according to classification results.For the two of the output of third machine learning model
Kind classification results, can take different processing modes to obtain final recognition result.Final recognition result is, for example, healthy, different
Often, belong to a certain or a variety of diseases or belong to the confidence level of various diseases.
Two predicted vectors are spliced in medical image recognition methods provided in this embodiment, and utilize third machine learning mould
Type classifies to spliced vector, is had using artificial aptitude manner according to the mode that splicing vector obtains final recognition result
There is higher accuracy.
The present embodiment determines recognition result using a kind of preferred mode.Specifically, judge that third machine learning model is defeated
Whether classification results out are normal, the output category result if classification results are normally.Finally determining the medical image is
Health, it is not necessary to secondary vector be judged again.
If classification results are exception, final recognition result can be exported only as exception, secondary vector use can also be exported
It indicates that the medical image belongs to the confidence level of various diseases, or can further each numerical value in secondary vector be carried out
Judgement, exports the medical image and belongs to certain disease qualitative conclusions really.
For the medical image of exception class, its confidence level for belonging to various diseases is finally exported, provides quantization for user
Referential data, compared to deterministic conclusion is exported, the auxiliaring effect for exporting confidence information is stronger.
One embodiment of the present of invention additionally provides a kind of medical image identification device, which can be set in computer
In the electronic equipments such as server.The device includes: as shown in Figure 3
Acquiring unit 31, for obtaining medical image；
First machine learning model 32, for being classified to obtain primary vector to the medical image, described first to
Amount indicates that the type of the medical image is health or the first abnormal confidence level；
Second machine learning model 33, for being classified to obtain secondary vector to the medical image, described second to
Amount indicates that the type of the medical image is the second confidence level of various disease types；
Integrated unit 34, for obtaining third vector according to the primary vector and the secondary vector；
Determination unit 35, for obtaining the recognition result to the medical image according to the third vector.
As a preferred embodiment, the integrated unit 34 includes:
Vector concatenation unit obtains third vector, the third for splicing the primary vector and the secondary vector
Vector is made of first confidence level and second reliability.
Further, the determination unit 35 includes:
Third machine learning model obtains classification results, the classification results for being classified to the third vector
It is normal or abnormal；
Judging unit, for determining the recognition result to the medical image according to the classification results.
Further, the judgement be applied alone in judge the classification results whether be it is normal, when classification results are normal,
Export the classification results；When classification results are abnormal, the secondary vector is exported.
As another preferred embodiment, the integrated unit 34 includes:
First value unit, for taking the first numerical value in the primary vector according to the first ratio of presetting；
Second value unit, for taking second value in the secondary vector according to the second ratio of presetting；
Summation unit, for summing to obtain third vector to first numerical value and the second value.
Further, the determination unit 35 includes:
Comparing unit, for the third vector to be compared with given threshold；
Judging unit, for determining the recognition result to the medical image according to comparison result.
Further, the judging unit is for judging whether the third vector is less than the given threshold, when described
When third vector is less than the given threshold, determine recognition result for health；When the third vector is greater than or equal to described set
When determining threshold value, the secondary vector is exported.
One embodiment of the present of invention additionally provides a kind of medical image identification equipment, which includes: at least one
Manage device；And the memory being connect at least one processor communication；Wherein, memory, which is stored with, to be executed by a processor
Instruction, instruction executed by least one processor so that at least one processor executes the medical image in above-described embodiment
Recognition methods.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the present invention
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the present invention, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Obviously, the above embodiments are merely examples for clarifying the description, and does not limit the embodiments.It is right
For those of ordinary skill in the art, can also make on the basis of the above description it is other it is various forms of variation or
It changes.There is no necessity and possibility to exhaust all the enbodiments.And it is extended from this it is obvious variation or
It changes still within the protection scope of the invention.
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910036173.1A CN109859836B (en) | 2019-01-15 | 2019-01-15 | Medical image recognition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910036173.1A CN109859836B (en) | 2019-01-15 | 2019-01-15 | Medical image recognition method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109859836A true CN109859836A (en) | 2019-06-07 |
| CN109859836B CN109859836B (en) | 2022-03-22 |
Family
ID=66894868
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910036173.1A Active CN109859836B (en) | 2019-01-15 | 2019-01-15 | Medical image recognition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109859836B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110931125A (en) * | 2019-12-11 | 2020-03-27 | 北京深睿博联科技有限责任公司 | Discrimination signal identification method and device for cerebral apoplexy |
| CN111028219A (en) * | 2019-12-10 | 2020-04-17 | 浙江同花顺智能科技有限公司 | Colon image recognition method and device and related equipment |
| CN111081325A (en) * | 2019-12-27 | 2020-04-28 | 医渡云（北京）技术有限公司 | Medical data processing method and device |
| CN111582235A (en) * | 2020-05-26 | 2020-08-25 | 瑞纳智能设备股份有限公司 | Alarm method, system and equipment for monitoring abnormal events in station in real time |
| CN112598086A (en) * | 2021-03-04 | 2021-04-02 | 四川大学 | Deep neural network-based common colon disease classification method and auxiliary system |
| CN114708539A (en) * | 2022-04-19 | 2022-07-05 | 广州欢聚时代信息科技有限公司 | Image type identification method and device, equipment, medium and product thereof |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108577803A (en) * | 2018-04-26 | 2018-09-28 | 上海鹰瞳医疗科技有限公司 | Fundus image detection method, device and system based on machine learning |
| CN108631727A (en) * | 2018-03-26 | 2018-10-09 | 河北工业大学 | A kind of solar panel defect identification method based on convolutional neural networks |
| CN108665457A (en) * | 2018-05-16 | 2018-10-16 | 腾讯科技（深圳）有限公司 | Image recognition method, device, storage medium and computer equipment |
| US20190005684A1 (en) * | 2017-06-28 | 2019-01-03 | Deepmind Technologies Limited | Generalizable medical image analysis using segmentation and classification neural networks |
-
2019
- 2019-01-15 CN CN201910036173.1A patent/CN109859836B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190005684A1 (en) * | 2017-06-28 | 2019-01-03 | Deepmind Technologies Limited | Generalizable medical image analysis using segmentation and classification neural networks |
| CN108631727A (en) * | 2018-03-26 | 2018-10-09 | 河北工业大学 | A kind of solar panel defect identification method based on convolutional neural networks |
| CN108577803A (en) * | 2018-04-26 | 2018-09-28 | 上海鹰瞳医疗科技有限公司 | Fundus image detection method, device and system based on machine learning |
| CN108665457A (en) * | 2018-05-16 | 2018-10-16 | 腾讯科技（深圳）有限公司 | Image recognition method, device, storage medium and computer equipment |
Non-Patent Citations (1)
| Title |
|---|
| 张骞予: "深度学习在医学图像识别中的研究与应用", 《中国优秀硕士学位论文全文数据库(信息科技辑)》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111028219A (en) * | 2019-12-10 | 2020-04-17 | 浙江同花顺智能科技有限公司 | Colon image recognition method and device and related equipment |
| CN110931125A (en) * | 2019-12-11 | 2020-03-27 | 北京深睿博联科技有限责任公司 | Discrimination signal identification method and device for cerebral apoplexy |
| CN111081325A (en) * | 2019-12-27 | 2020-04-28 | 医渡云（北京）技术有限公司 | Medical data processing method and device |
| CN111081325B (en) * | 2019-12-27 | 2023-12-12 | 医渡云（北京）技术有限公司 | Medical data processing method and device |
| CN111582235A (en) * | 2020-05-26 | 2020-08-25 | 瑞纳智能设备股份有限公司 | Alarm method, system and equipment for monitoring abnormal events in station in real time |
| CN111582235B (en) * | 2020-05-26 | 2023-04-07 | 瑞纳智能设备股份有限公司 | Alarm method, system and equipment for monitoring abnormal events in station in real time |
| CN112598086A (en) * | 2021-03-04 | 2021-04-02 | 四川大学 | Deep neural network-based common colon disease classification method and auxiliary system |
| CN114708539A (en) * | 2022-04-19 | 2022-07-05 | 广州欢聚时代信息科技有限公司 | Image type identification method and device, equipment, medium and product thereof |
| CN114708539B (en) * | 2022-04-19 | 2024-11-08 | 广州欢聚时代信息科技有限公司 | Image type recognition method and its device, equipment, medium, and product |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109859836B (en) | 2022-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109859836A (en) | Medical image recognition method and device | |
| CN109859168A (en) | X-ray chest radiography image quality determination method and device | |
| KR20200005404A (en) | Diagnosis assistance system | |
| CN109697716A (en) | Glaucoma image recognition method, equipment and screening system | |
| CN111553436A (en) | Training data generation method, model training method and equipment | |
| CN113313045A (en) | Man-machine asynchronous recognition method based on multi-task learning and class activation graph feedback | |
| CN110210483A (en) | Medical image lesion region segmentation method, model training method and device | |
| CN116188879B (en) | Image classification and image classification model training method, device, equipment and medium | |
| KR102095731B1 (en) | Mra image learning method and assistance diagnosis method for blood vessel lesion of deep learning based assistance diagnosis system | |
| CN110582783A (en) | Training device, image recognition device, training method and program | |
| JP7398155B2 (en) | A system that estimates the health status of the human body | |
| CN109658385A (en) | Fundus image judgment method and device | |
| CN114450716A (en) | Image processing for stroke characterization | |
| CN110060770A (en) | A kind of information processing method and device of medical treatment detection | |
| KR102655827B1 (en) | Brain tissue segmentation System and method from the brain MRI data using deep-learning | |
| KR102279160B1 (en) | Method and apparatus for identifying anomaly area of image | |
| CN112908479B (en) | Intelligent classification evaluation method and system for chronic wounds | |
| CN112434807B (en) | Deep learning model performance verification method and device based on fundus images | |
| CN109549619A (en) | Method for determining edge width of ocular disc, device and system for diagnosing glaucoma disease | |
| CN109102486A (en) | Detection method of surface flaw and device based on machine learning | |
| CN111767931A (en) | Automatic report generation system based on medical images | |
| CN111369547B (en) | Method and apparatus for evaluating performance of a medical machine learning model based on risk weights | |
| CN112309529B (en) | A method and system for infection control management based on artificial intelligence | |
| CN115205954A (en) | Eye disease identification method, device and equipment | |
| Ho et al. | Constructing the optimal rectilinear Steiner tree derivable from a minimum spanning tree |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230412 Address after: Room 25, 4th Floor, Building 2, Yard A2, West Fourth Ring North Road, Haidian District, Beijing, 100195 Patentee after: Beijing Yingtong Yuanjian Information Technology Co.,Ltd. Address before: 200000 room 01, 8th floor, building 1, No. 180, Yizhou Road, Xuhui District, Shanghai Patentee before: SHANGHAI EAGLEVISION MEDICAL TECHNOLOGY Co.,Ltd. |