CN109242867A - A kind of hand bone automatic division method based on template - Google Patents
A kind of hand bone automatic division method based on template Download PDFInfo
- Publication number
- CN109242867A CN109242867A CN201810902656.0A CN201810902656A CN109242867A CN 109242867 A CN109242867 A CN 109242867A CN 201810902656 A CN201810902656 A CN 201810902656A CN 109242867 A CN109242867 A CN 109242867A
- Authority
- CN
- China
- Prior art keywords
- template
- bone
- samples
- hand bone
- training
- 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
- 210000002411 hand bone Anatomy 0.000 title claims abstract description 31
- 238000000034 method Methods 0.000 title claims abstract description 31
- 210000000988 bone and bone Anatomy 0.000 claims abstract description 50
- 238000012549 training Methods 0.000 claims abstract description 27
- 230000011218 segmentation Effects 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims description 16
- 239000013598 vector Substances 0.000 claims description 16
- 238000000513 principal component analysis Methods 0.000 claims description 10
- 238000001514 detection method Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 241000228740 Procrustes Species 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000013519 translation Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 230000014461 bone development Effects 0.000 claims description 4
- 210000000623 ulna Anatomy 0.000 claims description 4
- 210000000236 metacarpal bone Anatomy 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims description 2
- 238000000354 decomposition reaction Methods 0.000 claims 1
- 238000011161 development Methods 0.000 abstract description 3
- 230000018109 developmental process Effects 0.000 abstract description 3
- 210000002745 epiphysis Anatomy 0.000 abstract description 3
- 241000208340 Araliaceae Species 0.000 abstract 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 abstract 1
- 235000003140 Panax quinquefolius Nutrition 0.000 abstract 1
- 239000000284 extract Substances 0.000 abstract 1
- 235000008434 ginseng Nutrition 0.000 abstract 1
- 239000004575 stone Substances 0.000 abstract 1
- 210000000707 wrist Anatomy 0.000 abstract 1
- 238000012545 processing Methods 0.000 description 7
- 238000011156 evaluation Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000003010 carpal bone Anatomy 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 210000003275 diaphysis Anatomy 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013077 scoring method Methods 0.000 description 1
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
- G06T7/12—Edge-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
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20116—Active contour; Active surface; Snakes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a kind of hand bone automatic division method based on template.The present invention initially sets up stone age sample database.Then training sample and template is established, needs to establish template respectively to each section bone block in contemporaneity.The finally hand bone segmentation based on template.The present invention extracts the complete skeletal shape feature including epiphysis, includes richer skeleton development information;It avoids simultaneously and falls into undying tune ginseng;And the workload manually marked is effectively reduced by the method that training set updates.The present invention solves hand bone segmentation problem, to provide strong tool followed by Age Assessment ofHuman Wrist Bones.
Description
Technical Field
The invention relates to the field of medical image processing, in particular to a template-based hand bone automatic segmentation method.
Background
The bone age is the biological age reflecting the development of the human body, and the maturity of the human body can be accurately evaluated.
The evaluation of bone age by hand bone X-ray film is a bone age detection method generally adopted in the industry at present. According to the difference of the bone development degree of different races, the evaluation standards of the 'CHN method' and 'Chinese-05 method' and the like currently used in China are slightly different from those of the TW-2, TW-3, G-P method and the like at foreign countries in details, but are consistent in principle.
At present, the interpretation of the hand bones X-ray film in China is mainly performed by medical practitioners, and the bone age is artificially interpreted due to the facts that the shooting sources are few, the subjectivity is high, the scoring method is complicated, the error of the atlas method is large and the like, and the bone age is ambiguous and interfered.
Carrying out bone age detection by means of a method in the field of image processing in early foreign countries and combining an SVM (support vector machine) technology; then, a classifier method of machine learning is introduced to search for a characteristic region; bone age identification is now mainly performed by means of a deep learning framework. However, since the detection accuracy is not high or the classifier is complicated to tune, it is still not widely used.
The invention provides a hand bone segmentation method based on a template aiming at a bone age X-ray plain film image. The method is used as an important component of a bone age evaluation system, has the characteristics of high accuracy and convenient realization, is not possessed by other methods, and lays a good foundation for subsequent bone maturity evaluation.
Disclosure of Invention
The invention aims to solve the problems of complicated hand bone X-ray image preprocessing, inaccurate key point positioning, larger palm contour error, fuzzy finger bone boundary and the like in the general situation.
The invention provides a template-based method, which can realize effective segmentation and extraction of hand bones in data sets at home and abroad and provides powerful support for a next bone maturity evaluation method.
In order to achieve the purpose, the invention adopts a template-based hand bone automatic segmentation method, which comprises the following steps:
step 1, establishing a bone age sample library
A public hand bone X-ray surview dataset is collected which should include valid annotations such as gender, age, bone age.
According to the staged stage of hand bone development, samples can be divided into several stages by age, denoted G0, G1, G2, G3, … ….
Samples of different periods are contained in the same sample library and used for establishing different templates. Wherein the age samples of each period are required to be distributed approximately uniformly, but allow for partial period sample concentration.
Step 2, training samples and establishing templates
The invention automatically cuts the bone blocks to be detected by establishing the outline template of the hand bones, so that templates are required to be respectively established for all the bone blocks in the same period.
The sample training and template establishment for a single bone piece follows the following steps:
a. selecting n samples of a training set, and manually recording the position coordinates of the jth key feature point on the ith bone blockThen, it is formed into a one-dimensional shape vector, which is recorded as αi(i=1,2,…,n)。
b. The obtained shape vectors are normalized. The adopted normalization method is Procrustes transformation, and the parameter needing to be calculated has a rotation angle thetaiHorizontal direction translation amount DeltaXiTranslation amount Δ Y in the vertical directioniPair αiMaking a Procrustes transformation and recording
c. The shape vectors after alignment are subjected to Principal Component Analysis (PCA).
First, an average shape vector is calculated
Then, a covariance matrix is calculated
The covariance matrix S is then matrix decomposed. P eigenvalues lambda of the samei(i ═ 1,2, …, q) are ranked from large to small, and the first t eigenvalues are chosen to satisfy:
wherein f is a proportionality coefficient which is generally 95-98%. The matrix formed by corresponding eigenvectors is denoted as Pt。
d. And establishing local gray scale characteristics of the key points. Firstly, the gray values of all training samples are normalized, then pixel points near key points are sampled and PCA is carried out, and finally the result is stored as a characteristic parameter. Wherein the key point gray scale characteristic is g1,g2,…gkFor representing, average gradation characteristicsAnd (4) showing.
Through the steps a-d, a template can be initially established.
Since the number of initial samples in the training set is far from the requirement of sufficient training, the template needs to be used for search matching of the verification set, and images successfully matched are added into the training set to serve as training samples.
If the matching of key points of the bone blocks fails, the correction is carried out by a manual marking method, and then a training set is added. Until all the samples of the verification set can find the coordinates of the key points, and the coordinates are combined into the training set.
Thus, the characteristic point information of all samples except the test sample is obtained; further generating a template for testing.
Step 3, hand bone segmentation based on template
For the hand bone segmentation based on the template, the method comprises the following two steps:
(1) and (6) detecting. The previously established template comprises an average shapeFirst t eigenvalues after PCA and corresponding eigen matrix PtThe parameter b reflects the change of the bone block posture:
when the template is applied to the search of new images, the average gray feature of each feature point trained before is usedNewly calculated gray level feature of j-th key pointAndthe similarity measurement between the two is represented by the Mahalanobis distance and is used as the basis for updating the parameters.
(2) And (6) dividing. The shape characteristics and the local characteristics of each hand bone are obtained through a template search mode, and then the phalanges, the metacarpal bones and the ulna and radius bones can be sequentially segmented.
The invention has the beneficial effects that: extracting complete bone shape characteristics including epiphyses and richer bone development information; meanwhile, endless parameter adjustment is avoided; and the workload of manual labeling is effectively reduced by a training set updating method. The invention solves the difficult problem of hand bone segmentation and provides a powerful tool for the subsequent bone age assessment.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
fig. 2 is an effect diagram of the embodiment of the present invention.
Detailed Description
Aiming at the defects of the image preprocessing stage in the current bone age identification method, the invention realizes the automatic detection and segmentation of the hand bones by a new mode. Details of the implementation of the scheme are set forth below in conjunction with FIG. 1, and the following steps:
step 1, establishing a bone age sample library
It is first necessary to collect bone age samples containing various ages, and divide them into three periods (defined as G0, G1, G2) according to actual conditions. The number of samples per period is 600 for example (half for both male and female).
For the Gi (i ═ 0,1,2) period: the samples of the period are extracted from the sample library and divided into 3 parts, one part is used as a training set (Trainingset), one part is used as a verification set (Validationset), and the rest is used as a test set (Testset).
The starting size of these three datasets is Validationset > Transingset > Testset; wherein the number of samples of the first two data sets gradually changes, while the number of samples of the test set is fixed.
The protocol initially requires numbering of the left-hand bones. According to the standard of the Chinese-05 method, 13 bone blocks including the 1 st, 3 rd and 5 th metacarpal bones, the 1 st, 3 rd and 5 th proximal phalanges, the 3 rd and 5 th middle phalanges, the 1 st, 3 th and 5 th distal phalanges and ulna and radius bones are included in the bone blocks to be detected (the invention does not include the treatment of carpal bones).
Step 2, training samples and establishing templates
a. A shape vector is constructed. For each bone block to be measured, k key point positions are determined firstly, and a one-dimensional shape vector is formed by the key point positions as follows:
wherein,and the coordinates of the jth characteristic point on the ith training sample are shown, and n is the number of the training samples.
The number of feature points is kept consistent for bone pieces of the same number. Meanwhile, the number of the key points is required to satisfy the requirements of not only completely representing the shape characteristics of the bone block, but also ensuring that the workload of manual marking is as small as possible.
b. And carrying out shape normalization. Because the bone blocks of different samples have differences of distance, angle, posture and the like, normalization processing needs to be carried out on all shape vectors. The n shape vectors are aligned to the average shape by appropriate rotation, translation operations.
To achieve normalization of all shape vectors, a diagonal matrix W needs to be defined first:
wherein,ikindicating the distance from the kth point to the 1 st point on the ith bone piece,representing all sample bone blocks between ikThe variance of (c).
By usingRepresents Procrustes transformation; the alignment process, which is essentially a process of minimizing the distance between all sample feature points, is defined as
Wherein
c. And carrying out PCA processing on the normalized result. ComputingAnd αiThe covariance matrix S of (2) is obtained by dividing its p eigenvalue lambdai(i ═ 1,2, …, q) are ranked from large to small, and the first t eigenvectors are selected to satisfy:
wherein f is 98 percent.
d. Local gray scale features are established. The position information of the feature points is very important in the template search process. In order to move towards the target position in each iteration, a set of gray matrices needs to be defined for each keypoint as characteristic parameters.
For a group of feature points, firstly, sampling a gray value in an 8 x 8 area taking the feature point as a center, and then convolving the gray value with a defined kernel to obtain a parameter matrix; the matrix is then normalized by dividing each element by the sum of the absolute values of all elements.
The PCA processing is also performed on the gray matrix of the feature points in a similar manner to the processing of the shape vectors.
Step 3, hand bone segmentation based on template
Through the above processing, now the shape of any bone piece to be measured can be represented as:
whereinIs an average shape, PtB is a weight vector (the change of b needs to be in a reasonable range) for a feature matrix obtained after PCA;
b=(b1,b2,…,bt)T
for the hand bone image to be detected, the approximate position of the bone block to be matched in the picture is determined firstly. In order to converge the template to the target contour as soon as possible, a region of the initial search position is provided. The starting search position is determined by the position of the average shape.
Then, in each iteration process, all characteristic points are sampled nearby and the characteristic parameter g 'is calculated'iIt is compared with the key point information recorded in the templateAnd comparing, obtaining a new position of the current feature point by calculating the Mahalanobis distance, and then finding new positions for all the feature points, thereby completing one iteration. After N iterations, if the template finally converges to the target position, the contour of the bone block is considered to be successfully found.
Because the outline of the bone block contains the key information of growth and development, the shapes of metaphysis, epiphysis and diaphysis can be used as judgment bases when the bone age is evaluated.
In a specific embodiment, the number of update iterations of the template has an effect on the detection result, as a saturation point exists, after which the result is affected little.
Except for removing partial bone age X-ray films which seriously affect identification or are incomplete, the technical scheme can effectively and automatically segment phalanges, metacarpals, ulna and radius bones in the same picture according to different age groups, as shown in figure 2.
The above description of the embodiments of the present invention is not intended to limit the scope of the claims of the present invention.
Claims (3)
1. A template-based hand bone automatic segmentation method is characterized by comprising the following steps:
step 1, establishing a bone age sample library
Collecting a public hand bone X-ray plain film data set, wherein the data set comprises effective marks;
dividing the sample into a plurality of periods according to the age according to the staged stage of hand bone development;
samples in different periods are contained in the same sample library and used for establishing different templates;
step 2, training samples and establishing templates
Automatically segmenting bone blocks to be detected by establishing a contour template of a hand bone, and respectively establishing templates for all the bone blocks in the same period;
the sample training and template establishment for a single bone piece follows the following steps:
a. selecting n samples of a training set, and manually recording the position coordinates of the jth key feature point on the ith bone blockThen, it is formed into a one-dimensional shape vector, which is recorded as αi,i=1,2,…,n;
b. Normalizing the obtained shape vector; the normalization method is Procrustes transformation, and the parameters needing to be calculated comprise the rotation angle thetaiHorizontal direction translation amount DeltaXiTranslation amount Δ Y in the vertical directioniFor one-dimensional shape vector αiMaking a Procrustes transformation and recording
c. Performing principal component analysis on the aligned shape vectors;
first, an average shape vector is calculated
Then, a covariance matrix is calculated
Then carrying out matrix decomposition on the covariance matrix S; p eigenvalues lambda of the sameiAnd (3) arranging from large to small, selecting the first t characteristic values to satisfy:
wherein f is a scaling factor and f is a constant,the matrix formed by corresponding eigenvectors is denoted as Pt;
d. Establishing local gray scale characteristics of the key points;
firstly, normalizing the gray values of all training samples, then sampling pixel points near key points and performing principal component analysis, and finally storing the result as a characteristic parameter; wherein the key point gray scale characteristic is g1,g2,…gkFor representing, average gradation characteristicsRepresents;
a template is initially established through the steps a to d;
because the number of initial samples in the training set far cannot meet the requirement of full training, the template needs to be used for searching and matching of the verification set, and images successfully matched are added into the training set to serve as training samples;
if the matching of key points of the bone blocks fails, correcting by using a manual marking method, and then adding a training set; until all samples of the verification set can find the coordinates of the key points, and the coordinates are combined to a training set;
thus, the characteristic point information of all samples except the test sample is obtained; further generating a template for testing;
step 3, hand bone segmentation based on template
For the hand bone segmentation based on the template, the method comprises the following two steps:
(1) and (3) detection: the previously established template comprises an average shapeThe first t eigenvalues and corresponding eigen matrix P after principal component analysistThe parameter b reflects the change of the bone block posture:
when applying templates to a search for new images, an average of each feature point previously trained is requiredGrayscale featuresNewly calculated gray level feature of j-th key pointAndthe similarity measurement between the two is represented by the Mahalanobis distance and is used as the basis for updating the parameters;
(2) and (3) dividing: the shape characteristics and the local characteristics of each hand bone are obtained through a template searching mode, and then the phalanges, the metacarpal bones and the ulna and radius bones are sequentially segmented.
2. The automatic template-based hand bone segmentation method according to claim 1, wherein: the labels in the data set in step 1 that are valid include gender, age, and bone age.
3. The automatic template-based hand bone segmentation method according to claim 1, wherein: step 1 age samples for each epoch are distributed approximately evenly but allow for partial epoch samples to be concentrated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810902656.0A CN109242867A (en) | 2018-08-09 | 2018-08-09 | A kind of hand bone automatic division method based on template |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810902656.0A CN109242867A (en) | 2018-08-09 | 2018-08-09 | A kind of hand bone automatic division method based on template |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109242867A true CN109242867A (en) | 2019-01-18 |
Family
ID=65071415
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810902656.0A Pending CN109242867A (en) | 2018-08-09 | 2018-08-09 | A kind of hand bone automatic division method based on template |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109242867A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110211674A (en) * | 2019-04-23 | 2019-09-06 | 平安科技(深圳)有限公司 | Stone age test method and relevant device based on machine learning model |
CN111325734A (en) * | 2020-02-25 | 2020-06-23 | 杭州电子科技大学 | Bone age prediction method and device based on visual model |
Citations (2)
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 |
CN103824281A (en) * | 2014-01-07 | 2014-05-28 | 沈阳航空航天大学 | Bone inhibition method in chest X-ray image |
-
2018
- 2018-08-09 CN CN201810902656.0A patent/CN109242867A/en active Pending
Patent Citations (2)
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 |
CN103824281A (en) * | 2014-01-07 | 2014-05-28 | 沈阳航空航天大学 | Bone inhibition method in chest X-ray image |
Non-Patent Citations (2)
Title |
---|
胡永利: ""基于ASM模型的骨龄评价系统研究"", 《中国图象图形学报》 * |
黄飞等: ""ASM的手骨提取方法研究"", 《计算机工程与应用》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110211674A (en) * | 2019-04-23 | 2019-09-06 | 平安科技(深圳)有限公司 | Stone age test method and relevant device based on machine learning model |
CN110211674B (en) * | 2019-04-23 | 2023-06-27 | 平安科技(深圳)有限公司 | Bone age testing method and related equipment based on machine learning model |
CN111325734A (en) * | 2020-02-25 | 2020-06-23 | 杭州电子科技大学 | Bone age prediction method and device based on visual model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019174376A1 (en) | Lung texture recognition method for extracting appearance and geometrical feature based on deep neural network | |
Hill et al. | Model-based interpretation of 3d medical images. | |
CN106485695B (en) | Medical image Graph Cut dividing method based on statistical shape model | |
CN108921057B (en) | Convolutional neural network-based prawn form measuring method, medium, terminal equipment and device | |
CN111671448A (en) | Bone age detection method and device based on deep learning | |
CN107230203A (en) | Casting defect recognition methods based on human eye vision attention mechanism | |
CN102831614B (en) | Sequential medical image quick segmentation method based on interactive dictionary migration | |
CN112784879A (en) | Medical image segmentation or classification method based on small sample domain self-adaption | |
CN106340000A (en) | Bone age assessment method | |
CN108629762B (en) | Image preprocessing method and system for reducing interference characteristics of bone age evaluation model | |
CN108446613A (en) | A kind of pedestrian's recognition methods again based on distance centerization and projection vector study | |
CN109567839A (en) | Hip joint x-ray image automatic analysis method | |
CN109242867A (en) | A kind of hand bone automatic division method based on template | |
CN112419452A (en) | Rapid merging system and method for PD-L1 digital pathological section images of stomach cancer | |
CN103839050A (en) | ASM positioning algorithm based on feature point expansion and PCA feature extraction | |
CN109993116B (en) | Pedestrian re-identification method based on mutual learning of human bones | |
CN109523484B (en) | Fractal feature-based finger vein network repair method | |
CN106971176A (en) | Tracking infrared human body target method based on rarefaction representation | |
CN114863129A (en) | Instrument numerical analysis method, device, equipment and storage medium | |
CN103745473A (en) | Brain tissue extraction method | |
CN113948190A (en) | Method and equipment for automatically identifying X-ray skull positive position film cephalogram measurement mark points | |
CN113158981A (en) | Riding posture analysis method based on cascade convolution neural network | |
Li et al. | Extraction of key regions of beef cattle based on bidirectional tomographic slice features from point cloud data | |
CN109886320B (en) | Human femoral X-ray intelligent recognition method and system | |
CN109300150B (en) | Hand bone X-ray image texture feature extraction method for bone age assessment |
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 |
Application publication date: 20190118 |
|
WD01 | Invention patent application deemed withdrawn after publication |