CN110265140A - Foot deformity detection model, foot deformity detection system and foot deformity detection method - Google Patents
Foot deformity detection model, foot deformity detection system and foot deformity detection method Download PDFInfo
- Publication number
- CN110265140A CN110265140A CN201910127177.0A CN201910127177A CN110265140A CN 110265140 A CN110265140 A CN 110265140A CN 201910127177 A CN201910127177 A CN 201910127177A CN 110265140 A CN110265140 A CN 110265140A
- Authority
- CN
- China
- Prior art keywords
- foot
- image information
- image
- foot deformity
- deformity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 206010061159 Foot deformity Diseases 0.000 title claims abstract description 237
- 238000001514 detection method Methods 0.000 title claims abstract description 148
- 238000012545 processing Methods 0.000 claims abstract description 24
- 230000001052 transient effect Effects 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 36
- 238000013527 convolutional neural network Methods 0.000 claims description 34
- 208000004067 Flatfoot Diseases 0.000 claims description 24
- 238000004458 analytical method Methods 0.000 claims description 20
- 238000002203 pretreatment Methods 0.000 claims description 12
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 208000029386 Apodia Diseases 0.000 claims description 5
- 241000208340 Araliaceae Species 0.000 claims description 4
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 4
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 4
- 235000008434 ginseng Nutrition 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 abstract description 4
- 210000002683 foot Anatomy 0.000 description 171
- 238000000034 method Methods 0.000 description 23
- 238000010586 diagram Methods 0.000 description 6
- 238000002595 magnetic resonance imaging Methods 0.000 description 6
- 230000003287 optical effect Effects 0.000 description 6
- 238000012360 testing method Methods 0.000 description 5
- 238000003745 diagnosis Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000010191 image analysis Methods 0.000 description 3
- 230000001537 neural effect Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 230000001737 promoting effect Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 210000000459 calcaneus Anatomy 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 229940079593 drug Drugs 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 208000032170 Congenital Abnormalities Diseases 0.000 description 1
- 206010061619 Deformity Diseases 0.000 description 1
- 208000005392 Spasm Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 210000000988 bone and bone Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013434 data augmentation Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000002224 dissection Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000003871 fifth metatarsal bone Anatomy 0.000 description 1
- 210000003108 foot joint Anatomy 0.000 description 1
- 210000001872 metatarsal bone Anatomy 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 210000004003 subcutaneous fat Anatomy 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
- A61B6/50—Clinical applications
- A61B6/505—Clinical applications involving diagnosis of bone
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Abstract
The present invention provides a kind of foot deformity detection system, includes an image acquisition unit and a non-transient machine-readable medium.Target the inner side of the foot position X-ray image information of the image acquisition unit to obtain a subject.Non-transient machine-readable medium includes a storage element and a processing unit.Storage element is to store foot deformity detection program.When foot deformity detects the foot deformity type and foot deformity occurrence degree when program is executed by processing unit to judge subject.Whereby, foot deformity detection system of the invention can effectively promote accuracy and the susceptibility of foot deformity detection, and can provide instant and effective foot deformity check and evaluation result.
Description
Technical field
The invention relates to a kind of medical information analysis model, system and method, especially a kind of foot deformity detection
Model, foot deformity detection system and foot deformity detection method.
Background technique
Until today, the quickest and the most direct method whether foot deformity occurs that detects is by image analysing computer
Mode assesses the dissection geometrical construction of foot, wherein being again that detection foot deformity is most widely used with X-ray image analysis
Image analysing computer detection method.It still needs to whether however, being occurred with X-ray image analysis mode detection foot deformity through analyst couple
Skeletal image is analyzed, and is to be easy in subsequent interpretation X optical image because the different of different analysts compare habit and produce
Raw different diagnostic result causes the existing known accuracy by the progress foot deformity judgement of X-ray image analysis mode more not
It is good.
Due to the development of Medical Imaging Technology, Magnetic resonance imaging (Magnetic Resonance Imaging, MRI) is opened
Begin to be applied in detection relevant to foot deformity, since Magnetic resonance imaging can effectively obtain complete foot three-dimensional structure shadow
Picture makes it have efficient detection precision.Not only need consuming higher however, carrying out foot deformity detection with Magnetic resonance imaging
Cost, the time of detection is also longer, be with using Magnetic resonance imaging carry out foot deformity diagnosis in current practice application side
Face is less universal.
Therefore, a kind of foot deformity detection system with pin-point accuracy rate and quickly detected how is developed, actually a tool
Commercially valuable technical task.
Summary of the invention
The purpose of the invention is to provide the detections of a kind of foot deformity detection model, foot deformity detection system and foot deformity
Method can provide instant (real by the determination module automated under the basis of a large amount of X-ray image information
Time) and accurate foot deformity check and evaluation is as a result, using the foundation as the monitoring of the subsequent state of an illness or curative effect evaluation.
An aspect of the invention is to be to provide a kind of foot deformity detection model, and it includes following set-up steps.Obtain one
Referring to information bank, wherein above-mentioned include multiple reference the inner side of the foot position X-ray image information referring to information bank.Before carrying out an image
Processing step is that each image size referring to the inner side of the foot position X-ray image information is adjusted using an image information editor module
And an image black white contrast, to obtain multiple standardization the inner side of the foot positions X-ray image information.A Feature Selection step is carried out,
It is special to obtain at least one image after analyzing standardization the inner side of the foot position above-mentioned X-ray image information using a characteristic selecting module
Value indicative.A training step is carried out, is that at least one image feature value above-mentioned is passed through into a convolution neural network learning point
Class device is trained and reaches convergence, to obtain foot deformity detection model above-mentioned, wherein foot deformity detection model above-mentioned is to use
To judge the foot deformity type and a foot deformity occurrence degree of a subject.
According to foot deformity detection model above-mentioned, wherein convolutional neural networks Study strategies and methods above-mentioned can be
Inception-ResNet-v2 convolutional neural networks.
According to foot deformity detection model above-mentioned, wherein image pre-treatment step above-mentioned can be more to each referring to the inner side of the foot
Position X-ray image information carries out image chroma extension processing.
According to foot deformity detection model above-mentioned, wherein the image format above-mentioned referring to the inner side of the foot position X-ray image information
Can for digital medical image storage standard agreement (Digital Imaging and Communications in Medicine,
DICOM image format).
According to foot deformity detection model above-mentioned, wherein foot deformity type above-mentioned can be flexibility flat foot or structure
Property flat foot.
According to foot deformity detection model above-mentioned, wherein foot deformity occurrence degree above-mentioned can be apodia, slight foot
Deformity, moderate foot deformity or severe foot deformity.
According to foot deformity detection model above-mentioned, wherein may include respectively in a left foot referring to the inner side of the foot position X-ray image information
Side position X-ray image information and a right the inner side of the foot position X-ray image information.
Another aspect of the invention is to be to provide a kind of foot deformity detection system, and the foot to judge subject is abnormal
Shape type and a foot deformity occurrence degree, and foot deformity detection system includes an image acquisition unit and a non-transient machine
Device readable media.Target the inner side of the foot position X-ray image information of the image acquisition unit to obtain subject.Non-transient machine
Readable media signal connects image acquisition unit above-mentioned, and non-transient machine-readable medium includes at a storage element and one
Unit is managed, wherein storage element detects program to store target the inner side of the foot position X-ray image information above-mentioned and a foot deformity,
And processing unit is to execute foot deformity detection program above-mentioned.Wherein, foot deformity detection program above-mentioned includes that a foot is abnormal
Shape detection model establishes subprogram and foot deformity detection subprogram.Foot deformity detection model establishes subroutine pack containing a reference
Information bank obtains module, one referring to image information editor module, a characteristic selecting module and a training module.Referring to information bank
Obtaining module is to obtain one referring to information bank, wherein above-mentioned include multiple reference the inner side of the foot position X-ray shadow referring to information bank
As information.It is each image size and a shadow referring to the inner side of the foot position X-ray image information of adjustment referring to image information editor module
As black white contrast, to obtain multiple standardization the inner side of the foot positions X-ray image information.Characteristic selecting module is aforementioned to analyze
Standardization the inner side of the foot position X-ray image information after at least one referring to image feature value.Training module is to will be aforementioned
At least one be trained referring to image feature value by a convolution neural network learning classifier and reach convergence, with
One foot deformity detection model.Foot deformity detects subroutine pack and chooses mould containing a target image information editing module, a target signature
Block and a comparison module.Target image information editing's module is a shadow of adjustment target the inner side of the foot position X-ray image information above-mentioned
As size and an image black white contrast, to obtain a standardization target the inner side of the foot position X-ray image information.Target signature is chosen
Module be to analyze after the X-ray image information of standardization target the inner side of the foot above-mentioned position at least one target image feature
Value.Comparison module is to be divided at least one target image characteristic value above-mentioned with foot deformity detection model above-mentioned
Analysis to obtain a target image characteristic value weighted data, and analyzes target image characteristic value weighted data above-mentioned, tested to obtain
The foot deformity type and foot deformity occurrence degree of person.
According to foot deformity detection system above-mentioned, wherein convolutional neural networks Study strategies and methods above-mentioned can be
Inception-ResNet-v2 convolutional neural networks.
According to foot deformity detection system above-mentioned, wherein above-mentioned can be more to each reference referring to image information editor module
The inner side of the foot position X-ray image information carries out image chroma extension processing, and target image information editing module above-mentioned can be more to mesh
It marks the inner side of the foot position X-ray image information and carries out image chroma extension processing.
According to foot deformity detection system above-mentioned, wherein the image format of target the inner side of the foot above-mentioned position X-ray image information
It can be the image format of digital medical image storage standard agreement, the image lattice above-mentioned referring to the inner side of the foot position X-ray image information
Formula can be the image format of digital medical image storage standard agreement.
According to foot deformity detection system above-mentioned, wherein foot deformity type above-mentioned can be flexibility flat foot or structure
Property flat foot.
According to foot deformity detection system above-mentioned, wherein foot deformity occurrence degree above-mentioned can be apodia, slight foot
Deformity, moderate foot deformity or severe foot deformity.
According to foot deformity detection system above-mentioned, wherein may include respectively in a left foot referring to the inner side of the foot position X-ray image information
Side position X-ray image information and a right the inner side of the foot position X-ray image information, target the inner side of the foot position above-mentioned X-ray image information may include
Position X-ray image information and a right the inner side of the foot position X-ray image information on the inside of one left foot.
Another aspect of the invention is to be to provide a kind of foot deformity detection method, and it includes following step.There is provided just like
Foot deformity detection model described in leading portion.The target the inner side of the foot position X-ray image information of one subject is provided.To mesh above-mentioned
It marks the inner side of the foot position X-ray image information and carries out pre-treatment, be to adjust target the inner side of the foot using image information editor module above-mentioned
An image size and an image black white contrast for position X-ray image information, to obtain a standardization target the inner side of the foot position X-ray shadow
As information.Using after the X-ray image information of characteristic selecting module analytical standard target the inner side of the foot above-mentioned position at least one
Image feature value.At least one image feature value above-mentioned is analyzed using foot deformity detection model above-mentioned, it is tested to judge
The foot deformity type and a foot deformity occurrence degree of person.
According to foot deformity detection method above-mentioned, wherein image information editor module above-mentioned can be more to target the inner side of the foot
Position X-ray image information carries out image chroma extension processing.
According to foot deformity detection method above-mentioned, wherein the image format of target the inner side of the foot above-mentioned position X-ray image information
It can be the image format of digital medical image storage standard agreement.
According to foot deformity detection method above-mentioned, wherein target the inner side of the foot above-mentioned position X-ray image information may include a left side
The inner side of the foot position X-ray image information and a right the inner side of the foot position X-ray image information.
Whereby, foot deformity detection model of the invention, foot deformity detection system and foot deformity detection method are by will be in foot
Side position X-ray image information and target the inner side of the foot position X-ray image information carry out image and standardize pre-treatment, and utilize Feature Selection
Module analysis and after at least one image feature value, then with convolutional neural networks Study strategies and methods image feature value is carried out
Training, whether quickly and accurately to judge that the foot deformity of subject occurs, foot deformity type and foot deformity occurrence degree,
And the detection reproducibility with height.Furthermore by the inclusion of the foot deformity detection model energy of convolutional neural networks Study strategies and methods
Accuracy and the susceptibility for effectively promoting foot deformity detection, make foot deformity detection model of the invention, foot deformity detection system
And foot deformity detection method has excellent clinical application potentiality.
Detailed description of the invention
Fig. 1 is the establishment step flow chart for being painted the foot deformity detection model of an embodiment of the present invention.
Fig. 2A is the configuration diagram for being painted the foot deformity detection system of another embodiment of the present invention.
Fig. 2 B is the configuration diagram of the foot deformity detection program for the foot deformity detection system for being painted Fig. 2A.
Fig. 3 is the step flow chart for being painted the foot deformity detection method of another embodiment of the present invention.
Fig. 4 is the configuration diagram for being painted the convolutional neural networks Study strategies and methods of foot deformity detection model of the invention.
[main element symbol description]
100: foot deformity detection model
110: obtaining one referring to information bank
120: carrying out an image pre-treatment step
130: carrying out a Feature Selection step
140: carrying out a training step
200: foot deformity detection system
300: image acquisition unit
400: non-transient machine-readable medium
411: target the inner side of the foot position X-ray image information
410: storage element
420: processing unit
430: foot deformity detects program
440: foot deformity detection model establishes subprogram
441: obtaining module referring to information bank
442: referring to image information editor module
443: characteristic selecting module
444: training module
450: foot deformity detects subprogram
451: target image information editing's module
452: target signature chooses module
453: comparison module
500: foot deformity detection method
510: foot deformity detection model is provided
520: the target the inner side of the foot position X-ray image information of subject is provided
530: pre-treatment is carried out to target the inner side of the foot position X-ray image information
540: using characteristic selecting module analytical standard target the inner side of the foot position X-ray image information after at least one
Image feature value
550: analyzing at least one image feature value using foot deformity detection model
600: convolutional neural networks Study strategies and methods
601: target image characteristic value weighted data
Specific embodiment
It is following that each embodiment of the present invention will be discussed in greater detail.However, this embodiment can answering for various concept of the invention
With can specifically be carried out in a variety of different particular ranges.Specific embodiment be only for the purpose of description, and not by
It is limited to the range disclosed.
Fig. 1 is please referred to, is the establishment step process for being painted the foot deformity detection model 100 of an embodiment of the present invention
Figure.The establishment step of foot deformity detection model 100 includes step 110, step 120, step 130 and step 140.
Step 110 is to obtain one referring to information bank, wherein including multiple referring to the inner side of the foot position X-ray image referring to information bank
Information.Preferably, the image format above-mentioned referring to the inner side of the foot position X-ray image information can be digital medical image storage standard
The image format of agreement (Digital Imaging and Communications in Medicine, DICOM), by each ginseng
It is stored according to essential informations such as the inner side of the foot position physiological age information of X-ray image information, gender information referring to the inner side of the foot position X-ray
In the shelves head (header) of image information, in favor of subsequent analysis.Preferably, can be wrapped referring to the inner side of the foot position X-ray image information
Containing position X-ray image information on the inside of a left foot and a right the inner side of the foot position X-ray image information, to establish left foot and right sufficient foot respectively
Deformity is conducive to carry out more accurately foot deformity detection respectively to two foot of left and right referring to information bank.
Step 120 is to carry out an image pre-treatment step, is that each reference foot is adjusted using an image information editor module
An image size and an image black white contrast for inside position X-ray image information, to obtain multiple standardization the inner side of the foot positions X-ray
Image information.Specifically, image information editor module can be respectively by the shadow of different reference the inner side of the foot position X-ray image information
After being adjusted to 1024 pixels (pixel) × 1024 pixel as size, and adjust the gray-scale distribution that its black white contrast is 0 to 255
Image intensity, to reduce the different black and white coloration differences referring between the X-ray image information of the inner side of the foot position and increase image
Clarity, in favor of subsequent analysis.In addition, in the step 120, image information editor module can be further to each referring to foot
Inside position X-ray image information carries out the processing of image chroma extension.Specifically, image information editor module can calculate each reference
The image gray scale degree of the inner side of the foot position X-ray image information, and according to calculated result above-mentioned and sequentially to each referring to the inner side of the foot position
The image pixel row, column of X-ray image information fills up color automatically, and each referring to the inner side of the foot position X-ray shadow of grayscale tone will be presented
As information is converted to color shades, to increase each information strength referring to the inner side of the foot position X-ray image information, and then promoted subsequent
The accuracy of analysis, but the present invention is not limited with the content that preceding description is disclosed with schema.
Step 130 is to carry out a Feature Selection step, is that standardization foot above-mentioned is analyzed using a characteristic selecting module
To obtain at least one image feature value after the X-ray image information of inside position.Specifically, foot deformity detection model 100 of the invention
Automatically the image information of standardization the inner side of the foot position X-ray image information is analyzed using characteristic selecting module, and automatic
Corresponding image feature value is extracted, so as to promoting the foot deformity diagnosis efficiency of foot deformity detection model 100 of the invention.
Step 140 is to carry out a training step, is that at least one image feature value above-mentioned is passed through a convolutional Neural
E-learning classifier is trained and reaches convergence, with foot deformity detection model 100, wherein foot deformity detection model 100
It is the foot deformity type and a foot deformity occurrence degree to judge a subject.Preferably, convolutional Neural net above-mentioned
Network Study strategies and methods can be Inception-ResNet-v2 convolutional neural networks.Inception-ResNet-v2 convolutional Neural
Network is a kind of extensive visual recognition (Large Scale Visual based on ImageNet visible data information library
Recognition) convolutional neural networks can effectively expand in such a way that residual error connects (Residual connections)
Open up convolutional neural networks training depth, and then make Inception-ResNet-v2 convolutional neural networks in image classification with distinguish
Knowing aspect has quite high accuracy rate.
Preferably, foot deformity type above-mentioned can be flexibility flat foot or structural flat foot, and foot deformity above-mentioned
Occurrence degree then can be apodia, slight foot deformity, moderate foot deformity or severe foot deformity.Specifically, the foot of normal person
A natural arch is presented in bottom, is the arch of foot of sole, when human body is in walking or running, arch of foot can be according to various landform
The elastic force and torsion of appropriateness are provided, to achieve the purpose that shock-absorbing and balance.However, the arch of foot of flat foot patient it is very unobvious or
There are few arch of foots, stand or put down when it and step on when ground, and the arch of foot of medial plantar will sink and make vola that flat form be presented,
It is even smooth in ground, and then influence normal walking, the wherein subcutaneous fat more Zhao Yin of flexibility flat foot thicker in plantar bottom
Or more loose foot joint, and structural flat foot is then because of foot caused by joint exception or Surrounding muscles spasm
The hard phenomenon in portion.In the X-ray image information of the sufficient side of standing appearance position, its calcaneum of the splayfooted patient of hbm structure
(calcaneus) the angle arch of foot angle (arch angle) of most lower edge line and fifth metatarsal bone (5th metatarsal) are big
Flat foot is regarded as when 168 degree, and foot deformity detection model 100 of the invention then can be automatically according to standardization the inner side of the foot position X
Optical image information and judge whether subject suffers from flat foot disease and its foot deformity occurrence degree, avoid because of different analysts
It is different compare habit and to measure mode different and generate different diagnostic results, and then foot deformity of the invention is made to detect mould
The accuracy in detection of type 100 is substantially improved.
A and Fig. 2 B referring to figure 2., Fig. 2A are the framves for being painted the foot deformity detection system 200 of another embodiment of the present invention
Structure schematic diagram, Fig. 2 B are the configuration diagrams of the foot deformity detection program 430 for the foot deformity detection system 200 for being painted Fig. 2A.Foot
A foot deformity type and a foot deformity occurrence degree of the lopsided detection system 200 to judge a subject, and foot deformity is examined
Examining system 200 includes an image acquisition unit 300 and a non-transient machine-readable medium 400.
Target the inner side of the foot position X-ray image information 411 of the image acquisition unit 300 to obtain subject.In detail and
Speech, image acquisition unit 300 can be an X-ray check instrument, and the foot of subject is irradiated using the x-ray of low dosage, with
Obtain resolution target the inner side of the foot position X-ray image information 411 appropriate.Preferably, target the inner side of the foot position X-ray image information 411
Image format can be digital medical image storage standard agreement image format, by target the inner side of the foot position X-ray image information
The essential informations such as 411 physiological age information, gender information are stored in the shelves head of target the inner side of the foot position X-ray image information 411
In, in favor of subsequent analysis.Preferably, target the inner side of the foot position X-ray image information 411 may include position X-ray on the inside of a left foot
Image information and a right the inner side of the foot position X-ray image information, in favor of carrying out more accurately foot deformity inspection respectively to two foot of left and right
It surveys.
Non-transient 400 signal of machine-readable medium connects image acquisition unit 300, and non-transient machine-readable medium 400
Comprising a storage element 410 and a processing unit 420, wherein storage element 410 is to store target the inner side of the foot position X-ray image
Information 411 and a foot deformity detect program 430, and processing unit 420 is to execute foot deformity detection program 430.Foot deformity detection
Program 430 includes that a foot deformity detection model establishes subprogram 440 and foot deformity detection subprogram 450.
It includes one referring to 441, one reference image letter of information bank acquirement module that foot deformity detection model, which establishes subprogram 440,
Cease editor module 442, a characteristic selecting module 443 and a training module 444.
Obtaining module 441 referring to information bank is to obtain one referring to information bank, wherein above-mentioned include referring to information bank
Multiple reference the inner side of the foot position X-ray image information.Preferably, the image format above-mentioned referring to the inner side of the foot position X optical image information can
For the image format of digital medical image storage standard agreement, by each physiological age referring to the inner side of the foot position X-ray image information
The essential informations such as information, gender information are stored in the shelves head referring to the inner side of the foot position X-ray image information, in favor of subsequent point
Analysis.Preferably, may include position X-ray image information and a right the inner side of the foot position on the inside of a left foot referring to the inner side of the foot position X-ray image information
X-ray image information, to establish left foot and right sufficient foot deformity respectively referring to information bank, in favor of carrying out respectively to two foot of left and right
More accurately foot deformity detects.
Referring to image information editor module 442 be adjustment it is each referring to the inner side of the foot position X-ray image information an image size and
One image black white contrast, to obtain multiple standardization the inner side of the foot positions X-ray image information.Specifically, it is compiled referring to image information
1024 pixel × 1024 can be adjusted to for the different image sizes referring to the inner side of the foot position X-ray image information respectively by collecting module 442
After pixel, and the image intensity for the gray-scale distribution that its black white contrast is 0 to 255 is adjusted, it is different referring to the inner side of the foot position to reduce
The clarity of black and white coloration difference and increase image between X-ray image information, in favor of subsequent analysis.In addition, reference
Image information editor module 442 further can carry out the processing of image chroma extension referring to the inner side of the foot position X-ray image information to each.
Specifically, image information editor module can calculate it is each referring to the inner side of the foot position X-ray image information image gray scale degree, and according to
Color is sequentially filled up automatically to each image pixel row, column referring to the inner side of the foot position X-ray image information according to calculated result above-mentioned
Coloured silk will be presented each of grayscale tone and be converted to color shades referring to the inner side of the foot position X-ray image information, each referring to foot to increase
The information strength of inside position X-ray image information, and then promote the accuracy of subsequent analysis.
Characteristic selecting module 443 be to analytical standard the inner side of the foot position X-ray image information after at least one reference
Image feature value.Specifically, the image that characteristic selecting module 443 can automatically to standardization the inner side of the foot position X-ray image information
Information is analyzed, and automatically extracts corresponding image feature value.
Training module 444 is at least one to be classified referring to image feature value by a convolution neural network learning
Device is trained and reaches convergence, to obtain a foot deformity detection model.Preferably, convolutional neural networks learning classification above-mentioned
Device can be Inception-ResNet-v2 convolutional neural networks, effectively to extend the training depth of convolutional neural networks, in turn
The image classification and identification capability of training for promotion module 444.
It includes a target image information editing module 451, target signature selection module that foot deformity, which detects subprogram 450,
452 and a comparison module 453.
Target image information editing module 451 is to adjust an image size of target the inner side of the foot position X-ray image information 411
And an image black white contrast, to obtain a standardization target the inner side of the foot position X-ray image information.Specifically, target image is believed
Breath editor module 451 is that the image size of target the inner side of the foot position X-ray image information 411 is adjusted to 1024 pixels × 1024 pictures
After element, and the image intensity for the gray-scale distribution that its black white contrast is 0 to 255 is adjusted, and then obtains standardization target above-mentioned
The inner side of the foot position X-ray image information.Preferably, target image information editing module 451 can be further to target the inner side of the foot position X-ray
Image information 411 carries out the processing of image chroma extension, is the image ash for calculating target the inner side of the foot position X-ray image information 411
Rank degree, and according to calculated result above-mentioned and sequentially to the image pixel row, column of target the inner side of the foot position X-ray image information 411
Automatically color is filled up, the target the inner side of the foot that grayscale tone is presented position X-ray image information 411 is converted into color shades, to increase
Add the information strength of target the inner side of the foot position X-ray image information, and the shadow with each reference the inner side of the foot position above-mentioned X-ray image information
As format is consistent, and then promote the accuracy of subsequent analysis.
Target signature choose module 452 be to analytical standard target the inner side of the foot position X-ray image information after at least
One target image characteristic value.Specifically, target signature selection module 452 can be automatically to standardization target the inner side of the foot position X
The image information of optical image information is analyzed, and automatically extracts corresponding image feature value.
Comparison module 453 be at least one target image characteristic value to be analyzed with foot deformity detection model, with
A target image characteristic value weighted data is obtained, and analyzes target image characteristic value weighted data, to obtain the foot deformity class of subject
Type and foot deformity occurrence degree.
Preferably, foot deformity type above-mentioned can be flexibility flat foot or structural flat foot, and foot deformity above-mentioned
Occurrence degree then can be apodia, slight foot deformity, moderate foot deformity or severe foot deformity, but the present invention not as
Limit.
It referring to figure 3., is the step flow chart for being painted the foot deformity detection method 500 of another embodiment of the present invention.
Foot deformity detection method 500 includes step 510, step 520, step 530, step 540 and step 550.
Step 510 is to provide foot deformity detection model, and foot deformity detection model is via abovementioned steps 110 to step
140 are established.
Step 520 is to provide the target the inner side of the foot position X-ray image information of subject.Preferably, target the inner side of the foot position X
The image format of optical image information can be the image format of digital medical image storage standard agreement, by target the inner side of the foot position X
The essential informations such as the physiological age information of optical image information, gender information are stored in the shelves of target the inner side of the foot position X-ray image information
In head, in favor of subsequent analysis.Preferably, target the inner side of the foot position X-ray image information may include position X-ray shadow on the inside of a left foot
As information and a right the inner side of the foot position X-ray image information, in favor of carrying out more accurately foot deformity detection respectively to two foot of left and right.
Step 530 is to carry out pre-treatment to target the inner side of the foot position X-ray image information, is to utilize image information editor mould
Block adjusts an image size and an image black white contrast for target the inner side of the foot position X-ray image information, to obtain a standardization mesh
Mark the inner side of the foot position X-ray image information.Specifically, target image information editing module is by target the inner side of the foot position X-ray image letter
After the image size of breath is adjusted to 1024 pixels × 1024 pixels, and adjust the gray-scale distribution that its black white contrast is 0 to 255
Image intensity, and then obtain standardization target the inner side of the foot above-mentioned X-ray image information.Preferably, target image information is compiled
Collecting module further can carry out the processing of image chroma extension to target the inner side of the foot position X-ray image information, be to calculate target foot
The image gray scale degree of inside position X-ray image information, and according to calculated result above-mentioned and sequentially to target the inner side of the foot position X-ray
The image pixel row, column of image information fills up color automatically, and the target the inner side of the foot position X-ray image letter of grayscale tone will be presented
Breath is converted to color shades, to increase the information strength of target the inner side of the foot position X-ray image information, and then promotes subsequent analysis
Accuracy.
Step 540 be using characteristic selecting module analytical standard target the inner side of the foot position X-ray image information after at least
One image feature value.Specifically, foot deformity detection method 500 can be using characteristic selecting module automatically to standardization mesh
The image information of mark the inner side of the foot position X-ray image information is analyzed, and automatically extracts corresponding image feature value, so as to promoting
The foot deformity diagnosis efficiency of foot deformity detection method 500 of the invention.
Step 550 is to analyze at least one image feature value using foot deformity detection model, to judge the foot of subject
Lopsided type and a foot deformity occurrence degree.
According to above embodiment, specific test example set forth below simultaneously cooperates schema to be described in detail.
<test example>
One, referring to information bank
Reference information bank used in the present invention is TaiWan, China hospital general (Taichung Armed Forces
General Hospital) backtracking collected go connectionization subject clinical the inner side of the foot position X-ray image information, be
Through China Medical University and set up the clinical test meter that region examination board checks and approves in the research ethics committee, hospital
It draws, number are as follows: CRREC-107-079.Reference the inner side of the foot position X-ray above-mentioned referring to information bank comprising 6000 subjects
Image information, and the image format above-mentioned referring to the inner side of the foot position X-ray image information is all digital medical image storage standard association
Fixed image format, by related letters such as the physiological age information of each subject, gender information, case history number, tested numbers
Breath is stored in the shelves head of image information, in favor of subsequent analysis.
In addition, referring to foot when it shoots referring to the inner side of the foot position X-ray image information of every subject in information bank
Placement position and placing direction are all consistent, and each the inner side of the foot position X-ray image information includes position X-ray image information on the inside of a left foot
And a right the inner side of the foot position X-ray image information, to exclude to invert (Random Horizontal using Random Level
Inversion), the modes such as the information augments such as random translation (Random Translation), rotation, shearing processing are to left and right
Two foots carry out the analysis of foot deformity, and then improve foot deformity detection model of the invention, foot deformity detection system and foot deformity inspection
The analysis accuracy rate of survey method.
Two, foot deformity detection model of the invention
For foot deformity detection model of the invention after obtaining referring to information bank, each reference the inner side of the foot position X-ray image information will
It is adjusted using an image information editor module, each image size referring to the inner side of the foot position X-ray image information is adjusted to
1024 pixels × 1024 pixels, and the image intensity for the gray-scale distribution that its black white contrast is 0 to 255 is adjusted, to reduce difference
Referring to the clarity of black and white coloration difference and increase image between the X-ray image information of the inner side of the foot position, and further to each ginseng
Color is filled up automatically according to the image pixel row, column of the inner side of the foot position X-ray image information, and each referring to foot of grayscale tone will be presented
Inside position X-ray image information is converted to color shades, and obtains multiple standardization the inner side of the foot positions X-ray image information.
Specifically, since current deep neural network model operationally needs a large amount of training information
(Training Data, i.e., foot deformity detection model of the invention referring to the inner side of the foot position X-ray image information) reaches stabilization
The classification accuracy of convergence and height is intended if the number of training information not enough abundance will be such that deep neural network generated
Close phenomenon (Overfitting) and cause the error amount of judging result excessively high, cause the confidence level of deep neural network model compared with
It is low.In order to solve foregoing problems, foot deformity detection model of the invention is by image pre-treatment step to each referring to the inner side of the foot position
X-ray image information carries out information augments (Data Augmentation), with each referring to the inner side of the foot position X-ray image by changing
The geometric size of information, image intensity are distributed and are added the mode of noise to reduce the Probability of over-fitting, make it
Each raw information referring to the inner side of the foot position X-ray image information can also be retained while increasing image information data.
Then, respectively standardization the inner side of the foot position X-ray image information will be analyzed with characteristic selecting module, to obtain at least one
A image feature value.Specifically, characteristic selecting module can be distinguished further in the X-ray image information of each standardization the inner side of the foot position
Bony areas and background area, and the X-ray image information in analyzing bone region and obtain each standardization the inner side of the foot position X-ray
Image feature value in image information.
Then, image feature value above-mentioned will be trained by a convolution neural network learning classifier reaches receipts
It holds back, to obtain foot deformity detection model of the invention.In this test example, foot deformity detection model will be applied to judge foot deformity
Flexibility flat foot or structural flat foot in type, and the flat foot of subject is further exported based on the analysis results
Foot deformity occurrence degree judging result.
It referring to figure 4., is the convolutional neural networks Study strategies and methods 600 for being painted foot deformity detection model of the invention
Configuration diagram.In the test example of Fig. 4, convolutional neural networks Study strategies and methods 600 are Inception-ResNet-v2
Convolutional neural networks, it includes multiple convolutional layers (Convolution), multiple maximum pond layers (MaxPool), multiple average
Pond layer (AvgPool) and multiple cascading layers (Concat), image feature value is trained and be analyzed.
Specifically, Inception-ResNet-v2 convolutional neural networks are based on ImageNet visible data information
The extensive visual recognition convolutional neural networks in library, and the image information inside ImageNet visible data information library is all
Two-dimensional color image, therefore existing known GoogLeNet convolutional neural networks model has in its first convolutional layer
The filter of RGB triple channel.However, each raw video archives referring to the inner side of the foot position X-ray image information are all three-dimensional grayscale
Image is with foot deformity detection model of the invention further by the GoogLeNet convolution of the filter comprising RGB triple channel
Neural network model is converted to single channel by arithmetic mean method, and by stochastic gradient descent method (Stochastic
Gradient Descent, SGD) it is applied in the pre-training Model Neural of foot deformity detection model of the invention, with excellent
Change its training process, frequency of training can be the gradient descent method of 100 phases (Epochs) and 96 Mini-Batch Size of use,
And by changing initial learning rate (Learning Rates) to carry out modulation, wherein learning rate is instructed to neural network
The important parameter of control weight (weight) and deviation (bias) variation when practicing, is logical with foot deformity detection model of the invention
The numerical value for crossing adjustment learning rate can be further assured that loss function (Loss Function) up to stable convergence.
During foot deformity detection model of the invention is trained image feature value, the inner side of the foot is respectively standardized
The image feature value of position X-ray image information carries out two layers of convolutional layer and one layer of maximum pond layer (MaxPool) is handled, by institute
The image feature value of extraction carries out maximum output, and repeats two layers of convolutional layer above-mentioned and one layer of maximum pond layer output again
Afterwards, parallel tower (parallel towers) training is carried out using multiple convolutional layers, to complete the elementary training of image feature value
(Inception)。
After completing elementary training above-mentioned, each image feature value for standardizing the inner side of the foot position X-ray image information will be carried out
The residual error of 10 times (10 ×), 20 times (20 ×) and the different depth of 10 times (10 ×), different estate and different aspects
(Residual) module training is trained and is reached with the image feature value to each standardization the inner side of the foot position X-ray image information
Convergence.Specifically, since Inception-ResNet convolutional neural networks are after the weight operation by multiple stratum, because
Is carried out by different operation and is sentenced for the image feature value of each standardization the inner side of the foot position X-ray image information for each residual error module
It is disconnected, cause error accumulation, therefore the training of Inception-ResNet convolutional neural networks will transport the node of specific stratum
The input terminal that calculation value is withdrawn into the stratum carries out operation again, to prevent convolutional neural networks Study strategies and methods 600 to above-mentioned
The degradation phenomena that gradient disappears occurs after training for the weight operation that image feature value carries out multilayer, and error accumulation is avoided to lead
It causes information to lose, and can effectively promote the training effectiveness of convolutional neural networks Study strategies and methods 600.
It, will be sequentially with one layer of convolutional layer, average a pond layer, one after completing deep layer and the training of duplicate residual error module
Replace global average pond layer (Global Average Pooling 2D, GloAvePool2D) and a line rectification unit
Training layer (Rectified Linear Unit, ReLU) carries out final training and processing to convergent image feature value, so as to
Judge the flat foot occurrence type and flat foot occurrence degree of subject.Wherein, average pond layer can be first to completion residual error mould
The image feature value of block training is calculated, and in the hope of the average value of each image feature value, global average pond layer is replaced then may be used
Regularization (Regularization) processing is carried out to the overall network framework of convolutional neural networks Study strategies and methods 600, is prevented
Under the training mode for pursuing low error over-fitting occurs for convolutional neural networks Study strategies and methods 600, and judgement is caused to be tied
The error amount of fruit is excessively high, finally, line rectification module training layer then further swashs the image feature value completed after training
It is living, and a target image characteristic value weighted data 601 is exported, to carry out subsequent comparison and analysis.Line rectification above-mentioned
Module training layer can avoid foot deformity detection model output target image characteristic value weighted data 601 level off to zero or approach
In infinity, in favor of the subsequent progress for comparing step, and then the accuracy of judgement of foot deformity detection model of the invention is promoted
Rate.
Then, the flat foot occurrence type of aforementioned subject and flat foot occurrence degree judging result will be further whole
Together in referring in information bank, to optimize to foot deformity detection model of the invention, and then detect foot deformity of the invention
The training effect and accuracy of judgement degree of model are further promoted.
By above content it is found that foot deformity detection model, foot deformity detection system and foot deformity detection method pass through by
The inner side of the foot position X-ray image information and target the inner side of the foot position X-ray image information carry out image and standardize pre-treatment, and utilize feature
After choosing module analysis and obtaining at least one image feature value, then with convolutional neural networks Study strategies and methods to image feature value
Be trained, whether quickly and accurately to judge that the foot deformity of subject occurs, foot deformity type and foot deformity journey occurs
Degree, and the detection reproducibility with height.Furthermore mould is detected by the inclusion of the foot deformity of convolutional neural networks Study strategies and methods
Type can effectively promote accuracy and the susceptibility of foot deformity detection, make foot deformity detection model of the invention, foot deformity detection
System and foot deformity detection method can provide instant and accurate foot deformity detection in the case of a large amount of X-ray image information
Assessment result makes it have excellent clinical application potentiality using the foundation as the monitoring of the subsequent state of an illness or curative effect evaluation.
Although the present invention is disclosed above with embodiment, however, it is not to limit the invention, any to be familiar with this skill
Person, without departing from the spirit and scope of the present invention, when can be used for a variety of modifications and variations, therefore protection scope of the present invention
Subject to view claim institute defender.
Claims (18)
1. a kind of foot deformity detection model, which is characterized in that the foot deformity detection model is with the foundation of following establishment steps:
One is obtained referring to information bank, wherein the reference information bank includes multiple referring to the inner side of the foot position X-ray image information；
An image pre-treatment step is carried out, is to adjust respectively reference the inner side of the foot position X-ray shadow using an image information editor module
As the image size and an image black white contrast of information, to obtain multiple standardization the inner side of the foot positions X-ray image information；
A Feature Selection step is carried out, is to analyze standardization the inner side of the foot position X-ray image using a characteristic selecting module to believe
To obtain at least one image feature value after breath；And
A training step is carried out, is to carry out at least one image feature value by a convolution neural network learning classifier
It trains and reaches convergence, to obtain the foot deformity detection model, wherein the foot deformity detection model is to judge a subject
One foot deformity type and a foot deformity occurrence degree.
2. foot deformity detection model according to claim 1, it is characterised in that: the wherein convolutional neural networks learning classification
Device is Inception-ResNet-v2 convolutional neural networks.
3. foot deformity detection model according to claim 1, it is characterised in that: wherein the image pre-treatment step is more to each
Reference the inner side of the foot position X-ray image information carries out image chroma extension processing.
4. foot deformity detection model according to claim 1, it is characterised in that: wherein described referring to the inner side of the foot position X-ray shadow
As the image format that the image format of information is digital medical image storage standard agreement.
5. foot deformity detection model according to claim 1, it is characterised in that: wherein the foot deformity type is that flexibility is flat
Flatfoot or structural flat foot.
6. foot deformity detection model according to claim 1, it is characterised in that: wherein the foot deformity occurrence degree is apodia
Lopsided, slight foot deformity, moderate foot deformity or severe foot deformity.
7. foot deformity detection model according to claim 1, it is characterised in that: wherein respectively the reference the inner side of the foot position X-ray shadow
As information includes position X-ray image information and a right the inner side of the foot position X-ray image information on the inside of a left foot.
8. a kind of foot deformity detection system, to judge the foot deformity type and a foot deformity occurrence degree of a subject,
It is characterized in that, the foot deformity detection system includes:
One image acquisition unit, to obtain the target the inner side of the foot position X-ray image information of the subject；And
One non-transient machine-readable medium, signal connects the image acquisition unit, and the non-transient machine-readable medium includes one
Storage element and a processing unit, wherein the storage element is abnormal to store target the inner side of the foot position X-ray image information and a foot
Shape detects program, and the processing unit is to execute foot deformity detection program；
Wherein, foot deformity detection program includes:
One foot deformity detection model establishes subprogram, includes:
One obtains module referring to information bank, is to obtain one referring to information bank, wherein the reference information bank includes multiple ginsengs
According to the inner side of the foot position X-ray image information；
One referring to image information editor module, be adjust a respectively image size of reference the inner side of the foot X-ray image information and
One image black white contrast, to obtain multiple standardization the inner side of the foot positions X-ray image information；
One characteristic selecting module, be to analyze after the X-ray image information of standardization the inner side of the foot position at least one ginseng
According to image feature value；And
One training module is to pass through a convolution neural network learning classifier at least one reference image feature value by this
It is trained and reaches convergence, to obtain a foot deformity detection model；And
One foot deformity detects subprogram, includes:
One target image information editing's module is the image size and one for adjusting target the inner side of the foot position X-ray image information
Image black white contrast, to obtain a standardization target the inner side of the foot position X-ray image information；
One target signature choose module, be to analyze the standardization target the inner side of the foot position X-ray image information after at least
One target image characteristic value；And
One comparison module, be at least one target image characteristic value to be analyzed with the foot deformity detection model,
To obtain a target image characteristic value weighted data, and the target image characteristic value weighted data is analyzed, to obtain being somebody's turn to do for the subject
Foot deformity type and the foot deformity occurrence degree.
9. foot deformity detection system according to claim 8, it is characterised in that: the wherein convolutional neural networks learning classification
Device is Inception-ResNet-v2 convolutional neural networks.
10. foot deformity detection system according to claim 8, it is characterised in that: wherein reference image information editor's mould
Block more carries out image chroma extension processing, target image information editing's mould to respectively reference the inner side of the foot position X-ray image information
Block more carries out image chroma extension processing to target the inner side of the foot position X-ray image information.
11. foot deformity detection system according to claim 8, it is characterised in that: wherein target the inner side of the foot position X-ray image
The image format of information is the image format of digital medical image storage standard agreement, described referring to the inner side of the foot position X-ray image letter
The image format of breath is the image format of digital medical image storage standard agreement.
12. foot deformity detection system according to claim 8, it is characterised in that: wherein the foot deformity type is flexibility
Flat foot or structural flat foot.
13. foot deformity detection system according to claim 8, it is characterised in that: wherein the foot deformity occurrence degree is nothing
Foot deformity, slight foot deformity, moderate foot deformity or severe foot deformity.
14. foot deformity detection system according to claim 8, it is characterised in that: wherein respectively the reference the inner side of the foot position X-ray shadow
As information includes position X-ray image information and a right the inner side of the foot position X-ray image information, target the inner side of the foot position X-ray on the inside of a left foot
Image information includes position X-ray image information and a right the inner side of the foot position X-ray image information on the inside of a left foot.
15. a kind of foot deformity detection method, characterized by comprising:
One foot deformity detection model as claimed in claim 1 is provided；
The target the inner side of the foot position X-ray image information of one subject is provided；
Pre-treatment is carried out to target the inner side of the foot position X-ray image information, is to adjust the mesh using the image information editor module
An image size and an image black white contrast for the inner side of the foot position X-ray image information is marked, to obtain a standardization target the inner side of the foot
Position X-ray image information；
Using this feature choose after the X-ray image information of module analysis standardization target the inner side of the foot position at least one image is special
Value indicative；And
At least one image feature value is analyzed using the foot deformity detection model, to judge the foot deformity type of the subject
An and foot deformity occurrence degree.
16. foot deformity detection method according to claim 15, it is characterised in that: wherein the image information editor module is more
The processing of one image chroma extension is carried out to target the inner side of the foot position X-ray image information.
17. foot deformity detection method according to claim 15, it is characterised in that: wherein target the inner side of the foot position X-ray shadow
As the image format that the image format of information is digital medical image storage standard agreement.
18. foot deformity detection method according to claim 15, it is characterised in that: wherein target the inner side of the foot position X-ray shadow
As information includes position X-ray image information and a right the inner side of the foot position X-ray image information on the inside of a left foot.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW108101866 | 2019-01-17 | ||
| TW108101866A TWI663960B (en) | 2019-01-17 | 2019-01-17 | Foot deformity detection model, foot deformity detection system and foot deformity detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110265140A true CN110265140A (en) | 2019-09-20 |
Family
ID=67911747
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910127177.0A Pending CN110265140A (en) | 2019-01-17 | 2019-02-20 | Foot deformity detection model, foot deformity detection system and foot deformity detection method |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110265140A (en) |
| TW (1) | TWI663960B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210106403A1 (en) * | 2019-10-15 | 2021-04-15 | Dommar LLC | Apparatus and methods for orthodontic treatment planning |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2022139796A (en) * | 2021-03-12 | 2022-09-26 | オムロン株式会社 | Support device and method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106295124A (en) * | 2016-07-27 | 2017-01-04 | 广州麦仑信息科技有限公司 | Utilize the method that multiple image detecting technique comprehensively analyzes gene polyadenylation signal figure likelihood probability amount |
| WO2017055412A1 (en) * | 2015-09-30 | 2017-04-06 | Siemens Healthcare Gmbh | Method and system for classification of endoscopic images using deep decision networks |
| CN107133496A (en) * | 2017-05-19 | 2017-09-05 | 浙江工业大学 | Gene expression characteristicses extracting method based on manifold learning Yu closed loop depth convolution dual network model |
| CN107543788A (en) * | 2017-08-15 | 2018-01-05 | 焦作市人民医院 | A kind of urine erythrocyte abnormal rate detection method and system |
| CN107767928A (en) * | 2017-09-15 | 2018-03-06 | 深圳市前海安测信息技术有限公司 | Medical image report preparing system and method based on artificial intelligence |
| CN107767376A (en) * | 2017-11-02 | 2018-03-06 | 西安邮电大学 | X-ray film stone age Forecasting Methodology and system based on deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201683886U (en) * | 2009-12-09 | 2010-12-29 | 佳木斯大学 | Intelligent wirelessly-transmitted medical clinical thermometer |
| TWI394556B (en) * | 2011-01-07 | 2013-05-01 | Global Action Inc | A method for determining the arch type |
-
2019
- 2019-01-17 TW TW108101866A patent/TWI663960B/en not_active IP Right Cessation
- 2019-02-20 CN CN201910127177.0A patent/CN110265140A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017055412A1 (en) * | 2015-09-30 | 2017-04-06 | Siemens Healthcare Gmbh | Method and system for classification of endoscopic images using deep decision networks |
| CN106295124A (en) * | 2016-07-27 | 2017-01-04 | 广州麦仑信息科技有限公司 | Utilize the method that multiple image detecting technique comprehensively analyzes gene polyadenylation signal figure likelihood probability amount |
| CN107133496A (en) * | 2017-05-19 | 2017-09-05 | 浙江工业大学 | Gene expression characteristicses extracting method based on manifold learning Yu closed loop depth convolution dual network model |
| CN107543788A (en) * | 2017-08-15 | 2018-01-05 | 焦作市人民医院 | A kind of urine erythrocyte abnormal rate detection method and system |
| CN107767928A (en) * | 2017-09-15 | 2018-03-06 | 深圳市前海安测信息技术有限公司 | Medical image report preparing system and method based on artificial intelligence |
| CN107767376A (en) * | 2017-11-02 | 2018-03-06 | 西安邮电大学 | X-ray film stone age Forecasting Methodology and system based on deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210106403A1 (en) * | 2019-10-15 | 2021-04-15 | Dommar LLC | Apparatus and methods for orthodontic treatment planning |
| US11793605B2 (en) * | 2019-10-15 | 2023-10-24 | 3D Smile Usa, Inc. | Apparatus and methods for orthodontic treatment planning |
Also Published As
| Publication number | Publication date |
|---|---|
| TW202027680A (en) | 2020-08-01 |
| TWI663960B (en) | 2019-07-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11776119B2 (en) | Confidence-based method and system for analyzing images of a retina | |
| US20210142477A1 (en) | Bone Age Assessment And Height Prediction Model, System Thereof And Prediction Method Thereof | |
| CN106682616A (en) | Newborn-painful-expression recognition method based on dual-channel-characteristic deep learning | |
| CN110265119A (en) | Bone age assessment and prediction of height model, its system and its prediction technique | |
| CN109770932A (en) | The processing method of multi-modal brain neuroblastoma image feature | |
| CN110415234A (en) | Brain tumor dividing method based on multi-parameter magnetic resonance imaging | |
| US20190122363A1 (en) | Tomographic Data Analysis | |
| US11501441B2 (en) | Biomarker determination using optical flows | |
| CN101667289A (en) | Retinal image segmentation method based on NSCT feature extraction and supervised classification | |
| CN109124660A (en) | The postoperative risk checking method of gastrointestinal stromal tumor and system based on deep learning | |
| Huang et al. | Gastroesophageal reflux disease diagnosis using hierarchical heterogeneous descriptor fusion support vector machine | |
| CN113743463B (en) | Tumor benign and malignant recognition method and system based on image data and deep learning | |
| CN107292103A (en) | A kind of prognostic chart picture generation method and device | |
| CN113850753A (en) | Medical image information calculation method and device, edge calculation equipment and storage medium | |
| CN110265140A (en) | Foot deformity detection model, foot deformity detection system and foot deformity detection method | |
| CN115564741A (en) | Device for quantifying esophageal and gastric fundus varices based on CT (computed tomography) images | |
| CN110427987A (en) | A kind of the plantar pressure characteristic recognition method and system of arthritic | |
| Bhat et al. | Identification of intracranial hemorrhage using ResNeXt model | |
| Dusa et al. | Smart framework for black fungus detection using VGG 19 deep learning approach | |
| Kamaruzaman et al. | Systematic literature review: application of deep learning processing technique for fig fruit detection and counting | |
| CN112508884A (en) | Comprehensive detection device and method for cancerous region | |
| CN114240934B (en) | Image data analysis method and system based on acromegaly | |
| CN113129286B (en) | Method, device, equipment and storage medium for detecting osteoporosis | |
| CN109480842B (en) | System and apparatus for diagnosing functional dyspepsia | |
| CN113379724A (en) | Age inference system and method based on curved surface fault slice and deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190920 |