CN109272002B - Bone age tablet classification method and device - Google Patents
Bone age tablet classification method and device Download PDFInfo
- Publication number
- CN109272002B CN109272002B CN201811161328.6A CN201811161328A CN109272002B CN 109272002 B CN109272002 B CN 109272002B CN 201811161328 A CN201811161328 A CN 201811161328A CN 109272002 B CN109272002 B CN 109272002B
- Authority
- CN
- China
- Prior art keywords
- bone age
- bone
- age
- slices
- feature extraction
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30008—Bone
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the application provides a method and a device for classifying bone age tablets, which relate to the technical field of machine learning, and the method comprises the following steps: the method comprises the steps of obtaining bone age tablets of a user to be detected, adjusting each bone in the bone age tablets to a reference position, inputting the adjusted bone age tablets into a classification model, and determining the type of the bone age tablets, wherein the classification model is determined after a depth residual error network is trained by using a plurality of bone age tablets which are marked with the types of the bone age tablets in advance as training samples, and the types of the bone age tablets comprise a left hand, a right hand and a non-hand. Because a plurality of bone age tablets with bone age tablet types marked in advance are used as training samples, a classification model is determined after a deep residual error network is trained, then the classification model is adopted to determine the type of the bone age tablet of a user to be detected, and the left-hand type bone age tablet is used for detecting the bone age of the user to be detected, the problem of low detection precision caused by non-standard bone age tablets is solved.
Description
Technical Field
The embodiment of the invention relates to the technical field of machine learning, in particular to a method and a device for classifying bone age tablets.
Background
The bone age is short for bone age, is the development age obtained by comparing the bone development level of teenagers and children with the bone development standard, can reflect the maturity of the body more accurately than the age, the height and the weight, and can reflect the growth development level and the maturity of an individual more accurately.
The biological age of the children is judged and read clinically through detecting the bone age, the development condition of the children is evaluated through the difference between the biological age and the calendar age, the sexual maturity trend of the children is known, the adult height of the children is predicted, the biological age-based pediatric endocrine disease diagnosis method is widely used for treatment and monitoring of diseases affecting growth and development of the children, and the diagnosis of some pediatric endocrine diseases is greatly facilitated.
In the prior art, the bone age is estimated mainly by shooting bone age tablets and manually checking the bone age tablets. In order to ensure the consistency of bone age assessment, the bone age tablets are generally left-handed orthostatic bone age tablets. However, sometimes the bone age tablets are not left-handed correct bone age tablets due to irregular shooting, so that the estimated bone age is greatly different from the actual bone age.
Disclosure of Invention
In the prior art, when shooting is not standard, the bone age tablets are not left-handed bone age tablets, so that the problem that the difference between the estimated bone age and the actual bone age is large is caused.
In a first aspect, an embodiment of the present application provides a method for classifying bone age tablets, including:
acquiring a bone age sheet of a user to be detected;
adjusting each bone in the bone age tablets to a reference position;
inputting the adjusted bone age tablets into a classification model, and determining the types of the bone age tablets, wherein the classification model is determined after training a depth residual error network by taking a plurality of bone age tablets with bone age tablet types marked in advance as training samples, and the types of the bone age tablets comprise a left hand, a right hand and a non-hand.
Optionally, the adjusting each bone in the bone age slices to a reference position comprises:
acquiring coordinates of a preset reference point;
determining coordinates of key points in the bone age slices by adopting an adjusting model, wherein the adjusting model is determined after a depth residual error network is trained by taking a plurality of bone age slices marked with the coordinates of the key points in advance as training samples, and the coordinates of the key points and the coordinates of the preset reference points are coordinates in the same coordinate system;
determining the corresponding relation between the current position of each bone of the bone age tablet and the reference position according to the coordinates of the preset reference point and the coordinates of the key points;
and adjusting each bone in the bone age slices to a reference position according to the corresponding relation.
Optionally, before adjusting each bone in the bone age plate to a reference position, the method further includes:
carrying out Gaussian filtering and binarization processing on the bone age tablets;
dividing the bone age tablets subjected to binarization processing into a plurality of area blocks by adopting a water diffusion method;
segmenting a bone image from the region block with the largest area;
and pasting the segmented bone image to a preset background image.
Optionally, the method further comprises:
when the type of the bone age piece is determined to be a left hand, determining the position of each target epiphysis in the bone age piece;
for each target epiphysis, determining an age score of the target epiphysis according to the characteristics of the target epiphysis and the position of the target epiphysis;
and determining the bone age of the user to be detected according to the bone age scores of the target epiphyses.
In a second aspect, an embodiment of the present application provides a bone age plate sorting device, including:
the acquisition module is used for acquiring a bone age sheet of a user to be detected;
the adjusting module is used for adjusting each bone in the bone age slices to a reference position;
and the classification module is used for inputting the adjusted bone age tablets into a classification model and determining the types of the bone age tablets, the classification model is determined after a plurality of bone age tablets of which the types are marked in advance are used as training samples and a depth residual error network is trained, and the types of the bone age tablets comprise a left hand, a right hand and a non-hand.
Optionally, the adjusting module is specifically configured to:
acquiring coordinates of a preset reference point;
determining coordinates of key points in the bone age slices by adopting an adjusting model, wherein the adjusting model is determined after a depth residual error network is trained by taking a plurality of bone age slices marked with the coordinates of the key points in advance as training samples, and the coordinates of the key points and the coordinates of the preset reference points are coordinates in the same coordinate system;
determining the corresponding relation between the current position of each bone of the bone age tablet and the reference position according to the coordinates of the preset reference point and the coordinates of the key points;
and adjusting each bone in the bone age slices to a reference position according to the corresponding relation.
Optionally, the adjusting module is further configured to:
before adjusting each bone in the bone age tablets to a reference position, carrying out Gaussian filtering and binarization processing on the bone age tablets;
dividing the bone age tablets subjected to binarization processing into a plurality of area blocks by adopting a water diffusion method;
segmenting a bone image from the region block with the largest area;
and pasting the segmented bone image to a preset background image.
Optionally, the system further comprises a detection module;
the detection module is specifically configured to:
when the type of the bone age piece is determined to be a left hand, determining the position of each target epiphysis in the bone age piece;
for each target epiphysis, determining an age score of the target epiphysis according to the characteristics of the target epiphysis and the position of the target epiphysis;
and determining the bone age of the user to be detected according to the bone age scores of the target epiphyses.
In a third aspect, an embodiment of the present application provides a computer device, including at least one processor and at least one memory, where the storage unit stores a computer program, and when the program is executed by the processor, the processor is caused to execute the steps of the method in the first aspect.
In a fourth aspect, embodiments of the present application provide a computer-readable medium storing a computer program executable by a computer device, the program, when executed on the computer device, causing the computer device to perform the steps of the method of the first aspect.
In the embodiment of the application, a plurality of bone age tablets with bone age tablet types marked in advance are used as training samples, a classification model is determined after a deep residual error network is trained, then the classification model is used for determining the type of the bone age tablet of a user to be detected, the left-hand type bone age tablet is used for detecting the bone age of the user to be detected, and the problem of low detection precision caused by non-standard bone age tablets is solved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
Fig. 1 is a schematic flow chart illustrating a method for classifying bone age tablets according to an embodiment of the present disclosure;
FIG. 2 is a schematic illustration of a bone age tablet according to an embodiment of the present disclosure;
fig. 3 is a schematic flow chart of a preprocessing method according to an embodiment of the present disclosure;
FIG. 4 is a schematic flow chart illustrating a method for adjusting bone age slices according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of a deep residual error network according to an embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of a convolution feature extraction block according to an embodiment of the present disclosure;
FIG. 7 is a schematic flow chart illustrating a method for detecting bone age according to an embodiment of the present disclosure;
fig. 8 is a schematic structural diagram of a bone age plate sorting device according to an embodiment of the present application;
fig. 9 is a schematic structural diagram of a computer device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more clearly apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
For convenience of understanding, terms referred to in the embodiments of the present invention are explained below.
Age of bone: the short term bone age needs to be determined by means of a specific image of the bone in the X-ray image. Usually, a left-hand wrist of a person is photographed by an X-ray film, and a doctor observes the development degree of ossification centers at the lower ends of the metacarpal phalanges, the wrist bones and the radioulnar bones of the left hand through the X-ray film to determine the bone age.
The method for classifying the bone age tablets in the embodiment of the application is suitable for identifying the bone age tablets of a user to be detected before detecting the bone age and judging whether the bone age tablets meet the standard of detecting the bone age. On the other hand, the method is also suitable for classifying the training samples of the training deep residual error network.
Fig. 1 is a schematic flow chart illustrating a method for classifying bone age tablets according to an embodiment of the present application, where the flow chart may be executed by a bone age tablet classification device, and specifically includes the following steps:
and step S101, obtaining a bone age tablet of a user to be detected.
Bone age slices refer to specific images taken using X-rays, typically taking an X-ray of the left hand of a person as a bone age slice, as shown, for example, in fig. 2. Sometimes the bone age plate is an X-ray film of the right hand or an X-ray film of non-hands such as feet.
Step S102, adjusting each bone in the bone age slices to a reference position.
The reference position is preset, and specifically, the reference position may be a preset position of one or more bones in the bone age plate.
Optionally, before adjusting each bone in the bone age slices to the reference position, the bone age slices are preprocessed, and the preprocessing process includes the following steps, as shown in fig. 3:
step S301, Gaussian filtering and binarization processing are carried out on the bone age tablets.
The threshold value of the binaryzation is obtained by solving the maximum class interval of the image gray level histogram.
And step S302, dividing the bone age tablets after binarization processing into a plurality of area blocks by adopting a water diffusion method.
In step S303, a bone image is segmented from the region block having the largest area.
Step S304, pasting the segmented bone image to a preset background image.
The preset background image may be a solid black image corresponding to the length and width of the image of the hand bones.
Further, the image normalization processing comprises the following steps: the bone age tablet is an image in a dicom format, a window width and window level is selected according to dicom information, and the bone age tablet is converted into an image in a png format. The length-width ratio of the bone age tablet image is adjusted to 1:1 by adding black edges on the upper side or two sides of the bone age tablet image, and finally the bone age tablet image is zoomed to 512 x 512.
Optionally, adjusting each bone in the bone age plate to the reference position specifically includes the following steps, as shown in fig. 4:
in step S401, coordinates of a preset reference point are acquired.
The coordinates of the preset reference point may be coordinates of a preset part bone for representing a reference position of the part bone, for example, the coordinates of the preset reference point may be preset coordinates of a point associated with a middle finger for representing a reference position of the middle finger, and the coordinates of the preset reference point may also be preset coordinates of a point associated with a little finger for representing a reference position of the little finger.
Step S402, determining the coordinates of key points in bone age slices by adopting an adjustment model, wherein the adjustment model is determined after training a depth residual error network by taking a plurality of bone age slices marked with the coordinates of the key points in advance as training samples.
The coordinates of the key point and the coordinates of the preset reference point are coordinates in the same coordinate system.
Illustratively, a plurality of bone age slices are obtained, then key points around a little finger in the bone age slices are marked manually, then the bone age slices marked with the key points are input into a depth residual error network for training, and when a target function of the depth residual error network meets a preset condition, an adjustment model is determined. When the bone age sheet of the user to be detected is obtained, the bone age sheet is input into the adjustment model, and key points around the little finger in the bone age sheet are determined.
And step S403, determining the corresponding relation between the current position of each bone of the bone age tablet and the reference position according to the coordinates of the preset reference point and the coordinates of the key point.
And S404, adjusting each bone in the bone age slices to a reference position according to the corresponding relation.
When the preset reference point and the key point are all points related to the little finger, the corresponding relation between the current position of the little finger in the bone age tablet and the reference position of the little finger in the bone age tablet can be determined according to the coordinates of the preset reference point and the coordinates of the key point, and further the corresponding relation between the current positions of other bones in the bone age tablet and the reference positions can also be obtained, wherein the corresponding relation comprises a translation relation and a rotation relation. Then, each bone in the bone age tablets is adjusted to the reference position according to the corresponding relation. Since the bone age slices of the left hand and the right hand are similar, but when the bones in the bone age slices of the left hand and the right hand are adjusted to the reference positions, the bone age slices of the left hand and the bone age slices of the right hand are much different, the bones in the bone age slices are adjusted to the reference positions before the bone age slices are classified by adopting a classification model, and therefore the accuracy of distinguishing the bone age slices of the left hand from the bone age slices of the right hand is improved.
And step S103, inputting the adjusted bone age tablets into a classification model, and determining the types of the bone age tablets.
The classification model is determined after a depth residual error network is trained by taking a plurality of bone age slices which are marked with bone age slice types in advance as training samples, wherein the bone age slices types comprise a left hand, a right hand and a non-hand.
The training process of the classification model is described in detail as follows: multiple bone age tablets were obtained as training samples. For each bone age tablet, preprocessing the bone age tablet, and then adjusting each bone in the bone age tablet to a reference position, wherein the processes of adjusting the position of the bone age tablet and preprocessing the bone age tablet are described in the foregoing, and are not described herein again. And marking the type of each bone age tablet by a marking person, wherein the types of the bone age tablets comprise a left hand, a right hand and a non-hand. Then, data enhancement is carried out on the training samples, and the data enhancement method comprises but is not limited to:
1. and randomly rotating for a certain angle.
2. And randomly shifting 0-30 pixels up, down, left and right.
3. And randomly scaling by 0.85-1.15 times.
4. The image contrast and brightness are dithered by a small amount.
5. And more abnormal data are obtained by cutting, partially rotating and pasting the normal data.
And then inputting the training sample into a deep residual error network for training. During training, a loss function is calculated according to the type of the labeled bone age slices and the type of the bone age slices predicted by the network, training is carried out through a back propagation method, and the trained optimization algorithm uses an sgd algorithm with momentum and step attenuation.
Optionally, the structure of the depth residual error network is shown in fig. 5, and includes N consecutive convolution feature extraction blocks and a full connection layer, where N is greater than 0. And for any two continuous first convolution feature extraction blocks and second convolution feature extraction blocks in the N convolution feature extraction blocks, adding second image features output by the second convolution feature extraction blocks and first image features output by the first convolution feature extraction blocks to serve as input of a third convolution feature extraction block or output of the N continuous convolution feature extraction blocks. The third convolution feature extraction block is a convolution feature extraction block located after the second convolution feature extraction block and continuous with the second convolution feature extraction block. The convolution feature extraction block includes L convolution modules, where L is greater than 0, and any one convolution module includes a convolution layer, a BN layer, and an excitation layer, as shown in fig. 6.
After the classification model is trained, the bone age tablets are processed by N continuous convolution feature extraction blocks to obtain image features of the bone age tablets, the image features of the bone age tablets are input into the full-connection layer, and the types of the bone age tablets are output.
Because a plurality of bone age tablets with bone age tablet types marked in advance are used as training samples, a classification model is determined after a deep residual error network is trained, then the classification model is adopted to determine the type of the bone age tablet of a user to be detected, and the left-hand type bone age tablet is used for detecting the bone age of the user to be detected, the problem of low detection precision caused by non-standard bone age tablets is solved, and the bone age detection precision is improved.
After the step S103, when it is determined that the type of the bone age tablet is the left hand, the method for detecting the bone age of the user to be detected by using the bone age tablet with the type of the left hand specifically includes the following steps, as shown in fig. 7:
step S701, determining the position of each target epiphysis in the age block.
And determining the position of each target epiphysis in the bone age slices by adopting a positioning model, wherein the positioning model comprises a coarse positioning module and a fine positioning module, the coarse positioning module and the fine positioning module are both depth residual error networks, and the coarse positioning module is determined after training the depth residual error networks by taking a plurality of bone age slices marked with key points in advance as training samples. The fine positioning module takes bone age slices of a plurality of coarse segmentation areas marked with key points in advance as training samples, and determines the bone age slices after training the depth residual error network, and the number of the fine positioning modules is determined according to the number of the coarse segmentation areas output by the coarse positioning module.
The method comprises the steps of firstly inputting an age block into a rough positioning module, determining the coordinates of first key points corresponding to target epiphysis in the age block, and then determining one or more rough segmentation areas in the age block according to the coordinates of the first key points corresponding to the target epiphysis, wherein the rough segmentation areas comprise one or more target epiphysis. And inputting the rough segmentation area into a corresponding fine positioning module aiming at each rough segmentation area, determining the coordinates of second key points corresponding to each target epiphysis in the rough segmentation area, and finally determining the position of each target epiphysis according to the coordinates of the second key points corresponding to each target epiphysis in the rough segmentation area.
Step S702, aiming at each target epiphysis, determining the age score of the target epiphysis according to the characteristics of the target epiphysis and the position of the target epiphysis.
The characteristics of the target epiphysis are identified by adopting an identification model, and the identification model is determined after a depth residual error network is trained by taking a plurality of epiphysis images which are marked with the characteristic types of the target epiphysis in advance as training samples. Because the characteristics of each target epiphysis are different, a recognition model can be trained aiming at each target epiphysis, so that the accuracy of recognizing the characteristics of the target epiphysis is improved.
And inquiring an age scoring standard according to the characteristics of the target epiphysis and the position of the target epiphysis to determine the age score. Different locations and different characteristic epiphyses correspond to different age scores. The bone age scores determined by the different bone age scoring criteria will also vary when scoring, including but not limited to TW3 bone age scoring criteria, G-P (atlas).
And step S703, determining the bone age of the user to be detected according to the bone age scores of the target epiphyses.
Because different forms of the epiphysis represent different age stages, and forms of the epiphysis at different positions can also have certain differences, the embodiment of the application adopts a positioning model to automatically determine the position of the target epiphysis, then determines the age score by combining the characteristics of the target epiphysis and the position of the target epiphysis, determines the age of the user to be detected based on the age score, and does not need to artificially and subjectively judge the age according to the age slice, thereby improving the accuracy of detecting the age on one hand and the efficiency of detecting the age on the other hand.
Based on the same technical concept, the present application provides a bone age plate sorting device, as shown in fig. 8, the device 800 includes:
an obtaining module 801, configured to obtain a bone age slice of a user to be detected;
an adjusting module 802, configured to adjust each bone in the bone age slices to a reference position;
and the classification module 803 is configured to input the adjusted bone age slices into a classification model, and determine the types of the bone age slices, where the classification model is determined after training a depth residual error network by using a plurality of bone age slices which are labeled with bone age slice types in advance as training samples, and the types of the bone age slices include a left hand, a right hand, and a non-hand.
Optionally, the adjusting module 802 is specifically configured to:
acquiring coordinates of a preset reference point;
determining coordinates of key points in the bone age slices by adopting an adjusting model, wherein the adjusting model is determined after a depth residual error network is trained by taking a plurality of bone age slices marked with the coordinates of the key points in advance as training samples, and the coordinates of the key points and the coordinates of the preset reference points are coordinates in the same coordinate system;
determining the corresponding relation between the current position of each bone of the bone age tablet and the reference position according to the coordinates of the preset reference point and the coordinates of the key points;
and adjusting each bone in the bone age slices to a reference position according to the corresponding relation.
Optionally, the adjusting module 802 is further configured to:
before adjusting each bone in the bone age tablets to a reference position, carrying out Gaussian filtering and binarization processing on the bone age tablets;
dividing the bone age tablets subjected to binarization processing into a plurality of area blocks by adopting a water diffusion method;
segmenting a bone image from the region block with the largest area;
and pasting the segmented bone image to a preset background image.
Optionally, a detection module 804 is further included;
the detection module is specifically configured to:
when the type of the bone age piece is determined to be a left hand, determining the position of each target epiphysis in the bone age piece;
for each target epiphysis, determining an age score of the target epiphysis according to the characteristics of the target epiphysis and the position of the target epiphysis;
and determining the bone age of the user to be detected according to the bone age scores of the target epiphyses.
Based on the same technical concept, the embodiment of the present application provides a computer device, as shown in fig. 9, including at least one processor 901 and a memory 902 connected to the at least one processor, where a specific connection medium between the processor 901 and the memory 902 is not limited in this embodiment of the present application, and the processor 901 and the memory 902 are connected through a bus in fig. 9 as an example. The bus may be divided into an address bus, a data bus, a control bus, etc.
In the embodiment of the present application, the memory 902 stores instructions executable by the at least one processor 901, and the at least one processor 901 can execute the steps included in the foregoing method for classifying bone age slices by executing the instructions stored in the memory 902.
The processor 901 is a control center of the computer device, and can connect various parts of the computer device by using various interfaces and lines, and implement bone age classification by executing or executing instructions stored in the memory 902 and calling data stored in the memory 902. Optionally, the processor 901 may include one or more processing units, and the processor 901 may integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, an application program, and the like, and the modem processor mainly processes wireless communication. It will be appreciated that the modem processor described above may not be integrated into the processor 901. In some embodiments, the processor 901 and the memory 902 may be implemented on the same chip, or in some embodiments, they may be implemented separately on separate chips.
The processor 901 may be a general-purpose processor, such as a Central Processing Unit (CPU), a digital signal processor, an Application Specific Integrated Circuit (ASIC), a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof, and may implement or perform the methods, steps, and logic blocks disclosed in the embodiments of the present Application. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of a method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware processor, or may be implemented by a combination of hardware and software modules in a processor.
Based on the same inventive concept, the present application also provides a computer readable medium storing a computer program executable by a computer device, which when running on the computer device causes the computer device to perform the steps of the method for bone age classification.
It should be apparent to those skilled in the art that embodiments of the present invention may be provided as a method, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A method for classifying bone age tablets, comprising:
acquiring a bone age sheet of a user to be detected;
adjusting each bone in the bone age tablets to a reference position;
inputting the adjusted bone age slices into N continuous convolution feature extraction blocks in a classification model to obtain image features of the bone age slices, inputting the image features of the bone age slices into a full connection layer in the classification model to determine the type of the bone age slices, wherein N is greater than 0, the classification model is determined by taking a plurality of bone age slices marked with the types of the bone age slices in advance as training samples after training a depth residual error network, aiming at any two continuous first convolution feature extraction blocks and second convolution feature extraction blocks in the N continuous convolution feature extraction blocks, a second image feature output by the second convolution feature extraction block is added with a first image feature output by the first convolution feature extraction block and then is used as an input of a third convolution feature extraction block or an output of the N continuous convolution feature extraction blocks, and the third convolution feature extraction block is located after the second convolution feature extraction block and is connected with the second convolution feature extraction block And extracting continuous convolution feature extraction blocks of the blocks, wherein the types of the bone age slices comprise a left hand, a right hand and a non-hand.
2. The method of claim 1, wherein said adjusting each bone in said bone age slices to a reference position comprises:
acquiring coordinates of a preset reference point;
determining coordinates of key points in the bone age slices by adopting an adjusting model, wherein the adjusting model is determined after a depth residual error network is trained by taking a plurality of bone age slices marked with the coordinates of the key points in advance as training samples, and the coordinates of the key points and the coordinates of the preset reference points are coordinates in the same coordinate system;
determining the corresponding relation between the current position of each bone of the bone age tablet and the reference position according to the coordinates of the preset reference point and the coordinates of the key points;
and adjusting each bone in the bone age slices to a reference position according to the corresponding relation.
3. The method of claim 1, wherein prior to adjusting each bone in the bone age slices to a reference position, further comprising:
carrying out Gaussian filtering and binarization processing on the bone age tablets;
dividing the bone age tablets subjected to binarization processing into a plurality of area blocks by adopting a water diffusion method;
segmenting a bone image from the region block with the largest area;
and pasting the segmented bone image to a preset background image.
4. The method of any of claims 1 to 3, further comprising:
when the type of the bone age piece is determined to be a left hand, determining the position of each target epiphysis in the bone age piece;
for each target epiphysis, determining an age score of the target epiphysis according to the characteristics of the target epiphysis and the position of the target epiphysis;
and determining the bone age of the user to be detected according to the bone age scores of the target epiphyses.
5. A bone age plate sorting device, comprising:
the acquisition module is used for acquiring a bone age sheet of a user to be detected;
the adjusting module is used for adjusting each bone in the bone age slices to a reference position;
a classification module, configured to input the adjusted bone age slices into N consecutive convolution feature extraction blocks in a classification model, obtain image features of the bone age slices, input the image features of the bone age slices into a full connection layer in the classification model, determine the type of the bone age slices, where N is greater than 0, the classification model is determined after training a depth residual error network with multiple bone age slices in which the bone age slice type is labeled in advance as a training sample, and for any two consecutive first convolution feature extraction blocks and second convolution feature extraction blocks in the N consecutive convolution feature extraction blocks, a second image feature output by the second convolution feature extraction block is added to a first image feature output by the first convolution feature extraction block and then used as an input of a third convolution feature extraction block or an output of the N consecutive convolution feature extraction blocks, the third convolution feature extraction block is a convolution feature extraction block which is located behind the second convolution feature extraction block and is continuous with the second convolution feature extraction block, and the types of the bone age slices comprise a left hand, a right hand and a non-hand.
6. The apparatus of claim 5, wherein the adjustment module is specifically configured to:
acquiring coordinates of a preset reference point;
determining coordinates of key points in the bone age slices by adopting an adjusting model, wherein the adjusting model is determined after a depth residual error network is trained by taking a plurality of bone age slices marked with the coordinates of the key points in advance as training samples, and the coordinates of the key points and the coordinates of the preset reference points are coordinates in the same coordinate system;
determining the corresponding relation between the current position of each bone of the bone age tablet and the reference position according to the coordinates of the preset reference point and the coordinates of the key points;
and adjusting each bone in the bone age slices to a reference position according to the corresponding relation.
7. The apparatus of claim 5, wherein the adjustment module is further to:
before adjusting each bone in the bone age tablets to a reference position, carrying out Gaussian filtering and binarization processing on the bone age tablets;
dividing the bone age tablets subjected to binarization processing into a plurality of area blocks by adopting a water diffusion method;
segmenting a bone image from the region block with the largest area;
and pasting the segmented bone image to a preset background image.
8. The apparatus of any of claims 5 to 7, further comprising a detection module;
the detection module is specifically configured to:
when the type of the bone age piece is determined to be a left hand, determining the position of each target epiphysis in the bone age piece;
for each target epiphysis, determining an age score of the target epiphysis according to the characteristics of the target epiphysis and the position of the target epiphysis;
and determining the bone age of the user to be detected according to the bone age scores of the target epiphyses.
9. A computer arrangement comprising at least one processor and at least one memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to carry out the steps of the method according to any one of claims 1 to 4.
10. A computer-readable medium, in which a computer program for execution by a computer device is stored, which program, when run on the computer device, causes the computer device to carry out the steps of the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811161328.6A CN109272002B (en) | 2018-09-30 | 2018-09-30 | Bone age tablet classification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811161328.6A CN109272002B (en) | 2018-09-30 | 2018-09-30 | Bone age tablet classification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109272002A CN109272002A (en) | 2019-01-25 |
| CN109272002B true CN109272002B (en) | 2020-11-24 |
Family
ID=65195047
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811161328.6A Active CN109272002B (en) | 2018-09-30 | 2018-09-30 | Bone age tablet classification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109272002B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109994172A (en) * | 2019-03-06 | 2019-07-09 | 杭州津禾生物科技有限公司 | Stone age digitizes measuring and calculating and height data base management system online |
| CN109961099B (en) * | 2019-03-26 | 2022-03-04 | 浙江康体汇科技有限公司 | Multi-classifier wrist bone grade identification method based on height and age |
| CN109977864B (en) * | 2019-03-26 | 2022-03-04 | 浙江康体汇科技有限公司 | Wrist bone grade identification method based on secondary classification |
| CN110516715B (en) * | 2019-08-05 | 2022-02-11 | 杭州依图医疗技术有限公司 | Hand bone classification method and device |
| CN110598030B (en) * | 2019-09-26 | 2022-05-17 | 西南大学 | Oracle bone rubbing classification method based on local CNN framework |
| CN111882517B (en) * | 2020-06-08 | 2024-06-21 | 杭州深睿博联科技有限公司 | Bone age evaluation method, system, terminal and storage medium based on graph convolution neural network |
| CN114141367A (en) * | 2020-09-03 | 2022-03-04 | 大理大学 | Forensic teenager bone age identification and calculation method and system |
| CN112785555B (en) * | 2020-12-30 | 2022-08-02 | 深兰智能科技(上海)有限公司 | Bone detection method, bone detection device, electronic equipment and storage medium |
| CN113570618B (en) * | 2021-06-28 | 2023-08-08 | 内蒙古大学 | Weighted bone age assessment method and system based on deep learning |
| CN113487656B (en) * | 2021-07-26 | 2022-11-22 | 推想医疗科技股份有限公司 | Image registration method and device, training method and device, control method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100715763B1 (en) * | 2006-03-15 | 2007-05-08 | 재단법인서울대학교산학협력재단 | Method for evaluation of skeletal malformation using atlas matching method |
| KR100838339B1 (en) * | 2007-08-06 | 2008-06-13 | 주식회사 오스테오시스 | Method for bone age judgment and prediction of adult height and the computer readable recording medium thereof |
| CN102945545A (en) * | 2012-10-18 | 2013-02-27 | 重庆医科大学 | Method for preprocessing robust skeletal age evaluation image and positioning skeletal key point |
| CN107591200A (en) * | 2017-08-25 | 2018-01-16 | 卫宁健康科技集团股份有限公司 | Stone age marker recognition appraisal procedure and system based on deep learning and image group |
| CN107895367A (en) * | 2017-11-14 | 2018-04-10 | 中国科学院深圳先进技术研究院 | A kind of stone age recognition methods, system and electronic equipment |
| CN108056786A (en) * | 2017-12-08 | 2018-05-22 | 浙江大学医学院附属儿童医院 | A kind of stone age detection method and device based on deep learning |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103300872B (en) * | 2013-05-29 | 2015-09-09 | 罗伟 | A kind of method and checkout gear analyzing the stone age |
| US20160300146A1 (en) * | 2015-04-12 | 2016-10-13 | Behzad Nejat | Method of Design for Identifying Fixed and Removable Medical Prosthetics Using a Dynamic Anatomic Database |
| CN106327495A (en) * | 2016-08-26 | 2017-01-11 | 穆达文 | Biological bone recognition method, device and system |
| CN107767376B (en) * | 2017-11-02 | 2021-03-26 | 西安邮电大学 | X-ray bone age prediction method and system based on deep learning |
| CN108334899A (en) * | 2018-01-28 | 2018-07-27 | 浙江大学 | Quantify the bone age assessment method of information integration based on hand bone X-ray bone and joint |
-
2018
- 2018-09-30 CN CN201811161328.6A patent/CN109272002B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100715763B1 (en) * | 2006-03-15 | 2007-05-08 | 재단법인서울대학교산학협력재단 | Method for evaluation of skeletal malformation using atlas matching method |
| KR100838339B1 (en) * | 2007-08-06 | 2008-06-13 | 주식회사 오스테오시스 | Method for bone age judgment and prediction of adult height and the computer readable recording medium thereof |
| CN102945545A (en) * | 2012-10-18 | 2013-02-27 | 重庆医科大学 | Method for preprocessing robust skeletal age evaluation image and positioning skeletal key point |
| CN107591200A (en) * | 2017-08-25 | 2018-01-16 | 卫宁健康科技集团股份有限公司 | Stone age marker recognition appraisal procedure and system based on deep learning and image group |
| CN107895367A (en) * | 2017-11-14 | 2018-04-10 | 中国科学院深圳先进技术研究院 | A kind of stone age recognition methods, system and electronic equipment |
| CN108056786A (en) * | 2017-12-08 | 2018-05-22 | 浙江大学医学院附属儿童医院 | A kind of stone age detection method and device based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| A study of bone age evaluation based on hand knuckles radiogram;Chih-Yen Chen et al.;《2014 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) Proceedings》;IEEE;20140721;第68-71页 * |
| 基于X光图像的骨龄评估系统设计与实现;董娜 等;《计算技术与自动化》;20100315;第29卷(第1期);第67-71页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109272002A (en) | 2019-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272002B (en) | Bone age tablet classification method and device | |
| CN109377484B (en) | Method and device for detecting bone age | |
| CN109146879B (en) | Method and device for detecting bone age | |
| CN109741309B (en) | Bone age prediction method and device based on deep regression network | |
| WO2020164282A1 (en) | Yolo-based image target recognition method and apparatus, electronic device, and storage medium | |
| CN110866893B (en) | Pathological image-based TMB classification method and system and TMB analysis device | |
| CN110060237B (en) | Fault detection method, device, equipment and system | |
| CN109255786B (en) | Method and device for detecting bone age | |
| CN108108739B (en) | Method and device for detecting image target area, X-ray system and storage medium | |
| CN111986183B (en) | Chromosome scattered image automatic segmentation and identification system and device | |
| CN111340130A (en) | Urinary calculus detection and classification method based on deep learning and imaging omics | |
| CN109509177B (en) | Method and device for recognizing brain image | |
| CN114972922B (en) | Coal gangue sorting and identifying method, device and equipment based on machine learning | |
| CN112907576B (en) | Vehicle damage grade detection method and device, computer equipment and storage medium | |
| CN114511523B (en) | Gastric cancer molecular subtype classification method and device based on self-supervision learning | |
| CN111369530A (en) | CT image pulmonary nodule rapid screening method based on deep learning | |
| CN112819797A (en) | Diabetic retinopathy analysis method, device, system and storage medium | |
| CN112613471B (en) | Face living body detection method, device and computer readable storage medium | |
| CN114757950A (en) | Ultrasonic image processing method, device and computer readable storage medium | |
| CN104239843B (en) | Positioning method and device for face feature points | |
| CN110517234B (en) | Method and device for detecting characteristic bone abnormality | |
| CN110516715B (en) | Hand bone classification method and device | |
| CN116843940A (en) | Labeling and evaluating method, device, equipment and medium | |
| CN110458024B (en) | Living body detection method and device and electronic equipment | |
| CN110390671B (en) | Method and device for detecting mammary gland calcification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information |
Inventor after: Fu Junfen Inventor after: Wei Zikun Inventor after: Hua Yiwei Inventor after: Wang Qi Inventor before: Wei Zikun Inventor before: Hua Yiwei Inventor before: Wang Qi |
|
| CB03 | Change of inventor or designer information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |