CN108681731A - A kind of thyroid cancer ultrasound picture automatic marking method and system - Google Patents
A kind of thyroid cancer ultrasound picture automatic marking method and system Download PDFInfo
- Publication number
- CN108681731A CN108681731A CN201810298494.4A CN201810298494A CN108681731A CN 108681731 A CN108681731 A CN 108681731A CN 201810298494 A CN201810298494 A CN 201810298494A CN 108681731 A CN108681731 A CN 108681731A
- Authority
- CN
- China
- Prior art keywords
- cancer
- cluster
- picture
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of thyroid cancer ultrasound picture automatic marking method and systems, by being pre-processed to pending cancer image data collection, extract the ROI sub-graph data collection of every cancer picture, then after using VGG16 deep learnings network model to carry out feature extraction to ROI sub-graph data collection, the feature obtained to extraction using K means++ algorithms is clustered, and then after the result that cluster obtains is compared with the benchmark cluster result of preset no cancer picture, extraction obtains the cancer feature cluster of cancer picture, the cancer feature cluster that finally correspondence markings extraction obtains in the artwork of cancer picture.Work efficiency is high by the present invention, and accuracy is higher, saves a large amount of financial resource and material resource, and application cost is low, can be widely applied in the process field of medical image.
Description
Technical field
The present invention relates to medical image process fields, are marked automatically more particularly to a kind of thyroid cancer ultrasound picture
Injecting method and system.
Background technology
With the high speed development and maturation of computer storage capacity and computing capability, the relevant technologies of artificial intelligence obtain
Greatly development, the especially the relevant technologies of computer vision and natural language processing.Meanwhile artificial intelligence also constantly participates in
In different fields, the production efficiency and working efficiency of every field related industry are improved, wherein including just medicine and people
The combination of work intelligence.
Currently, the combination of artificial intelligence and medicine be mainly reflected in machine to doctor diagnosis when auxiliary on.Pass through people
Work intelligence, using computation vision technology and deep learning, machine can assist completing the Diseases diagnosis to medical image, such as first
The judgement of shape gland cancer disease.However, due to the limitation of technology, currently, being completed to thyroid cancer ultrasound figure in training artificial intelligence
When the identification of piece, mainly chooses a large amount of cancer picture and be trained as training set, this mode has the following problems：1)
Training pattern needs a large amount of thyroid cancer ultrasound picture as training sample；2) it is used for the training set of training pattern, is needed
There is doctor's hand labeled to go out the cancerous area on every pictures.This side that a large amount of picture indicia work is completed by doctor
Formula undoubtedly will produce serious time loss, while also can largely overstock the time of doctor, cause the waste of hospital resources.This
Outside, this mode when the accuracy of judgement for continuing to optimize model being needed to spend, needs the number of pictures marked that will continue to rise, this
Mode working efficiency is low, and needs to expend more manpower and materials, and practical application has little significance.
Explanation of nouns
ROI：Full name region of interest, area-of-interest；In machine vision, image procossing, from processed
Image sketches the contours of region to be treated in a manner of box, circle, ellipse, irregular polygon etc.；
K-means++ algorithms：A kind of cluster algorithm, be go to optimize on the basis of K-means algorithms it is initial random
Point into order algorithm.Wherein, K-means algorithms are hard clustering algorithms, are the typical object function clustering methods based on prototype
Representative, it is data point to certain object function of distance as an optimization of prototype, is obtained using the method that function seeks extreme value
The adjustment rule of interative computation.
Selective Search algorithms：Selective search algorithm, a kind of picture search algorithm, for a given figure
Piece, selective search algorithm will find out ROI region on picture.
Invention content
In order to solve the above technical problems, the object of the present invention is to provide a kind of thyroid cancer ultrasound pictures to mark automatically
Injecting method and system.
The technical solution adopted by the present invention to solve the technical problems is：
A kind of thyroid cancer ultrasound picture automatic marking method, includes the following steps：
S1, pending cancer image data collection is pre-processed, extracts the ROI sub-graph datas of every cancer picture
Collection；
S2, feature extraction is carried out to ROI sub-graph data collection using VGG16 deep learnings network model；
S3, the feature obtained to extraction using K-means++ algorithms are clustered；
After S4, the result that will cluster acquisition are compared with the benchmark cluster result of preset no cancer picture, extraction obtains
Obtain the cancer feature cluster of cancer picture；
S5, the cancer feature cluster that correspondence markings extraction obtains in the artwork of cancer picture.
Further, in the step S4, the benchmark cluster result of the preset no cancer picture is obtained by following steps
：
S01, no cancer image data collection is pre-processed, extracts every ROI sub-graph data collection without cancer picture；
S02, feature extraction is carried out to the ROI sub-graph data collection of no cancer picture using VGG16 deep learnings network model,
Obtain reference characteristic；
S03, the reference characteristic obtained to extraction using K-means++ algorithms are clustered.
Further, the step S1 is specially：
For every cancer picture that pending cancer image data is concentrated, contrast, brightness and clear are carried out to it
After degree adjustment, cutting is carried out to the cancer picture after adjustment using selective search algorithm and obtains multiple ROI subgraphs, and is recorded every
A ROI subgraphs after the coordinate of present position, obtain corresponding ROI sub-graph datas collection in the artwork of cancer picture.
Further, the step S2 is specially：
Using ROI sub-graph datas collection as input data, it is input in advance trained VGG16 deep learnings network model
It is calculated, and extracts the feature for the 5th section of convolutional layer that VGG16 deep learnings network model obtains in learning process.
Further, the step S3 is specially：
After the number k of the specified cluster cluster to be obtained, the feature obtained to extraction using K-means++ algorithms is gathered
Class obtains the conjunction of feature gathering.
Further, the step S4 is specially：
Each cluster in the feature gathering conjunction that cluster is obtained, with the benchmark cluster result of preset no cancer picture
After the cluster heart that feature gathering is closed carries out Euclidean distance calculating, the maximum preceding n cluster of chosen distance is special as the cancer of cancer picture
Levy cluster；
Wherein, n is preset constant.
Further, the step S3, specifically includes following steps：
The number k of S31, the specified cluster cluster to be obtained；
S32, the feature obtained for extraction, randomly choose one as after cluster centre, the feature that extraction is obtained into
Row traversal calculates the distance between the cluster centre of residue character and selection, and chosen distance is maximum characterized by newest poly-
Class center；
S33, judge whether the sum of cluster centre reaches k, if so, thening follow the steps S34, otherwise return and continue to execute
Step S32；
S34, all features for non-cluster center, assign it in the cluster of k cluster centre so that Mei Geju
The quadratic sum that the corresponding feature gathering of class is closed is minimum；
S35, after recalculating the cluster centre that each feature gathering is closed, S34 is returned to step, until in all clusters
The heart no longer changes.
Further, the step S5 is specially：
Each feature in the cancer feature cluster obtained for extraction, according to its corresponding coordinate information, in cancer figure
Corresponding position in the artwork of piece is marked this feature using preset tag format.
The present invention solves another technical solution used by its technical problem：
A kind of thyroid cancer ultrasound picture automatic marking system, including：
At least one processor；
At least one processor, for storing at least one program；
When at least one program is executed by least one processor so that at least one processor is realized
A kind of thyroid cancer ultrasound picture automatic marking method.
The beneficial effects of the invention are as follows：After the present invention obtains the ROI sub-graph data collection of every cancer picture by extraction, adopt
Feature extraction is carried out to ROI sub-graph datas collection with VGG16 deep learnings network model, then uses K-means++ algorithms to carrying
After the feature taken is clustered, cluster result is compared with benchmark cluster result, can automatically extract and obtain cancer picture
Cancer feature cluster, the cancer feature cluster that finally automatically correspondence markings extraction obtains in the artwork of cancer picture is entire to mark
Note process is that automation carries out, and does not need manpower intervention, saves a large amount of financial resource and material resource, and application cost is low, and automatic
Change that work efficiency is high, accuracy is higher, the annotation results of acquisition doctor can be facilitated quickly and intuitively to obtain there may be cancerations
Region, have preferable auxiliary reference meaning.
Description of the drawings
Fig. 1 is a kind of flow chart of thyroid cancer ultrasound picture automatic marking method of the present invention；
Fig. 2 is a kind of structure diagram of thyroid cancer ultrasound picture automatic marking system of the present invention.
Specific implementation mode
Embodiment of the method
Referring to Fig.1, a kind of thyroid cancer ultrasound picture automatic marking method, including following step are present embodiments provided
Suddenly：
S1, pending cancer image data collection is pre-processed, extracts the ROI sub-graph datas of every cancer picture
Collection；
S2, feature extraction is carried out to ROI sub-graph data collection using VGG16 deep learnings network model；
S3, the feature obtained to extraction using K-means++ algorithms are clustered；
After S4, the result that will cluster acquisition are compared with the benchmark cluster result of preset no cancer picture, extraction obtains
Obtain the cancer feature cluster of cancer picture；
S5, the cancer feature cluster that correspondence markings extraction obtains in the artwork of cancer picture.
After this programme obtains the ROI sub-graph data collection of every cancer picture by extraction, using VGG16 deep learning networks
Model carries out feature extraction to ROI sub-graph data collection, after then being clustered to the feature of extraction using K-means++ algorithms,
Cluster result is compared with benchmark cluster result, the cancer feature cluster for obtaining cancer picture can be automatically extracted, finally certainly
The dynamic ground cancer feature cluster that correspondence markings extraction obtains in the artwork of cancer picture, entire annotation process be automate into
Row, does not need manpower intervention, saves a large amount of financial resource and material resource, and automatically working is efficient, accuracy is higher, acquisition
Annotation results can facilitate doctor quickly and intuitively to obtain the region there may be canceration.
It is further used as preferred embodiment, in the step S4, the benchmark of the preset no cancer picture clusters
As a result it is obtained by following steps：
S01, no cancer image data collection is pre-processed, extracts every ROI sub-graph data collection without cancer picture；
S02, feature extraction is carried out to the ROI sub-graph data collection of no cancer picture using VGG16 deep learnings network model,
Obtain reference characteristic；
S03, the reference characteristic obtained to extraction using K-means++ algorithms are clustered.
Specifically, the concrete processing procedure of step S01~S03 is identical as step S1~S3, the two is only to deal with objects
Difference that is, in advance after the processing procedure for executing step S01~S03 to no cancer image data collection, establish without cancer picture number
According to collecting corresponding reference characteristic cluster result, when subsequent execution this method, treated using step S1~S3 of same process
The cancer image data collection of processing carries out clustering processing, consequently facilitating carrying out cancer feature extraction, finally realizes the cancer of this programme
Disease feature marks purpose.
This method cuts the ROI subgraphs of cancer picture by using selective search algorithm, relative to blindness cutting subgraph
Method, obtained picture effect is more preferable, picture scale also smaller, to improve efficiency and accuracy rate in subsequent operation.
It is further used as preferred embodiment, the step S1 is specially：
For every cancer picture that pending cancer image data is concentrated, contrast, brightness and clear are carried out to it
After degree adjustment, cutting is carried out to the cancer picture after adjustment using selective search algorithm and obtains multiple ROI subgraphs, and is recorded every
A ROI subgraphs after the coordinate of present position, obtain corresponding ROI sub-graph datas collection in the artwork of cancer picture.Record ROI
When figure coordinate, coordinate origin is set as the artwork upper left corner.
Since ultrasonoscopy itself is there are color dullness, texture is unintelligible, the relatively low situation of pixel value, by this step
After carrying out contrast, brightness and clarity adjustment to cancer picture, it is made more to adapt to selective search algorithm, to pass through choosing
Selecting property searching algorithm can obtain more satisfactory ROI subgraphs with cutting.
It is further used as preferred embodiment, the step S2 is specially：
Using ROI sub-graph datas collection as input data, it is input to through public image data set trained VGG16 in advance
It is calculated in deep learning network model, and extracts the 5th section of volume that VGG16 deep learnings network model in learning process obtains
The feature of lamination.
VGG16 deep learning network models be a kind of convolutional neural networks (Convolutional Neural Network,
CNN), there is level depth, the good feature of effect to have preferable image processing effect.
In this step, during VGG16 deep learning network models are calculated, show as to input data i.e. ROI
Sub-graph data collection carries out convolutional calculation, to extract the feature of the 5th section of convolutional layer as feature extraction result.
Relative to traditional convolutional neural networks, for VGG16 in the lower convolutional layer of level, feature is fairly simple.With
The intensification of level, VGG16 can extract the feature of higher order, these features are more abstract, while also have more semantic special
Sign.Therefore, this step can preferably extract the feature of image using VGG16.This step uses VGG16 deep learning networks
Model chooses the output of the deeper 5th section of convolutional layer of level as feature, relative to traditional spy as feature extractor
Extracting method is levied, more abstract and high-order feature can be extracted, to preferably complete subsequent operation.
In the present invention, the training goal of VGG16 deep learning network models is to carry out feature extraction, and is not limited to refer to
Determine cancer picture, therefore, is trained by public image data set and training goal can be realized.Public image data set is deep
The shared image data collection of learning training is spent, general deep learning training can download the data set and be trained.
It is further used as preferred embodiment, the step S3 is specially：
After the number k of the specified cluster cluster to be obtained, the feature obtained to extraction using K-means++ algorithms is gathered
Class obtains the conjunction of feature gathering.
It is further used as preferred embodiment, the step S4 is specially：
Each cluster in the feature gathering conjunction that cluster is obtained, with the benchmark cluster result of preset no cancer picture
After the cluster heart that feature gathering is closed carries out Euclidean distance calculating, the maximum preceding n cluster of chosen distance is special as the cancer of cancer picture
Levy cluster；
Wherein, n is preset constant.Preferably, in the present embodiment, the value of n is 3.
Euclidean distance is bigger, indicates that the similarity between cluster is smaller, and Euclidean distance is smaller, indicates that the similarity between cluster is got over
Greatly.The calculation formula of Euclidean distance is as follows：
Wherein, the Euclidean distance between two clusters x and y, x are indicatediAnd yiIndicate that the element of cluster x and y, n indicate in cluster respectively
The sum of feature.
It is further used as preferred embodiment, the step S3 specifically includes following steps：
The number k of S31, the specified cluster cluster to be obtained；
S32, the feature obtained for extraction, randomly choose one as after cluster centre, the feature that extraction is obtained into
Row traversal calculates the distance between the cluster centre of residue character and selection, and chosen distance is maximum characterized by newest poly-
Class center；Here, residue character refers to the feature except cluster centre；
S33, judge whether the sum of cluster centre reaches k, if so, thening follow the steps S34, otherwise return and continue to execute
Step S32；
S34, all features for non-cluster center, assign it in the cluster of k cluster centre so that Mei Geju
The quadratic sum that the corresponding feature gathering of class is closed is minimum；
Specific assigning process, using following formula：
Wherein,Indicate ith cluster cluster, k indicates the number of the final obtained cluster of clustering algorithm, j expressions between
1 to the arbitrary positive integer between k, xpIndicate feature,Indicate ith cluster center,Indicate j-th of cluster centre,
Dist indicates that the distance between two features, t indicate algorithm iteration number；
The formula has reacted foundation of the K-means++ algorithms in cluster process, i.e., is allocated to multiple features so that
Quadratic sum in cluster is minimum.
S35, after recalculating the cluster centre that each feature gathering is closed, S34 is returned to step, until in all clusters
The heart no longer changes.
The cluster centre that each feature gathering is closed is recalculated especially by following formula：
Indicate the cluster centre obtained after the feature gathering conjunction at ith cluster center is recalculated,Indicate i-th
A clustering cluster, xjExpression is subordinated toAll features, wherein j indicates that between 1 to the arbitrary positive integer between k, t indicates to calculate
Method iterations.
In the present invention, cluster refers to clustering some grouping obtained in calculating process, and cluster centre refers to that cluster operation obtains
The central point for the grouping arrived, the i.e. central point of cluster.
It is further used as preferred embodiment, the step S5 is specially：
Each feature in the cancer feature cluster obtained for extraction, according to its corresponding coordinate information, in cancer figure
Corresponding position in the artwork of piece is marked this feature using preset tag format.
Specific annotation process is as follows：By the title of feature find corresponding ROI subgraphs, then by being preserved in step S1
Coordinate information obtains its position in artwork in conjunction with the length and width of ROI subgraphs.By program by ROI subgraphs in artwork
Location information is recorded in xml document, to carry out automatic marking according to preset tag format.
System embodiment
With reference to Fig. 2, a kind of thyroid cancer ultrasound picture automatic marking system is present embodiments provided, including：
At least one processor 100；
At least one processor 200, for storing at least one program；
When at least one program is executed by least one processor 100 so that at least one processor
100 realize a kind of thyroid cancer ultrasound picture automatic marking method.
The thyroid cancer ultrasound picture automatic marking system of the present embodiment, executable the method for the present invention embodiment are provided
Thyroid cancer ultrasound picture automatic marking method, the arbitrary combination implementation steps of executing method embodiment have the party
The corresponding function of method and advantageous effect.
It is to be illustrated to the preferable implementation of the present invention, but the invention is not limited to the implementation above
Example, those skilled in the art can also make various equivalent variations or be replaced under the premise of without prejudice to spirit of that invention
It changes, these equivalent modifications or replacement are all contained in the application claim limited range.
Claims (9)
1. a kind of thyroid cancer ultrasound picture automatic marking method, which is characterized in that include the following steps：
S1, pending cancer image data collection is pre-processed, extracts the ROI sub-graph data collection of every cancer picture；
S2, feature extraction is carried out to ROI sub-graph data collection using VGG16 deep learnings network model；
S3, the feature obtained to extraction using K-means++ algorithms are clustered；
S4, it after being compared the result that cluster obtains with the benchmark cluster result of preset no cancer picture, extracts and obtains cancer
The cancer feature cluster of disease picture；
S5, the cancer feature cluster that correspondence markings extraction obtains in the artwork of cancer picture.
2. thyroid cancer ultrasound picture automatic marking method according to claim 1, which is characterized in that the step S4
In, the benchmark cluster result of the preset no cancer picture is obtained by following steps：
S01, no cancer image data collection is pre-processed, extracts every ROI sub-graph data collection without cancer picture；
S02, feature extraction is carried out to the ROI sub-graph data collection of no cancer picture using VGG16 deep learnings network model, obtained
Reference characteristic；
S03, the reference characteristic obtained to extraction using K-means++ algorithms are clustered.
3. thyroid cancer ultrasound picture automatic marking method according to claim 1, which is characterized in that the step
S1 is specially：
For every cancer picture that pending cancer image data is concentrated, contrast, brightness and clarity tune are carried out to it
After whole, cutting is carried out to the cancer picture after adjustment using selective search algorithm and obtains multiple ROI subgraphs, and is recorded each
ROI subgraphs after the coordinate of present position, obtain corresponding ROI sub-graph datas collection in the artwork of cancer picture.
4. thyroid cancer ultrasound picture automatic marking method according to claim 1, which is characterized in that the step
S2 is specially：
Using ROI sub-graph datas collection as input data, it is input in advance trained VGG16 deep learnings network model and carries out
It calculates, and extracts the feature for the 5th section of convolutional layer that VGG16 deep learnings network model obtains in learning process.
5. thyroid cancer ultrasound picture automatic marking method according to claim 1, which is characterized in that the step
S3 is specially：
After the number k of the specified cluster cluster to be obtained, the feature obtained to extraction using K-means++ algorithms is clustered,
Obtain the conjunction of feature gathering.
6. thyroid cancer ultrasound picture automatic marking method according to claim 5, which is characterized in that the step
S4 is specially：
Each cluster in the feature gathering conjunction that cluster is obtained, the feature with the benchmark cluster result of preset no cancer picture
After the cluster heart that gathering is closed carries out Euclidean distance calculating, cancer feature cluster of the maximum preceding n cluster of chosen distance as cancer picture；
Wherein, n is preset constant.
7. thyroid cancer ultrasound picture automatic marking method according to claim 5, which is characterized in that the step
S3 specifically includes following steps：
The number k of S31, the specified cluster cluster to be obtained；
S32, the feature obtained for extraction, randomly choose one as after cluster centre, are carried out time to the feature that extraction obtains
It goes through, calculates the distance between the cluster centre of residue character and selection, and chosen distance is maximum characterized by newest cluster
The heart；
S33, judge whether the sum of cluster centre reaches k, if so, thening follow the steps S34, otherwise return and continue to execute step
S32；
S34, all features for non-cluster center, assign it in the cluster of k cluster centre so that each cluster pair
The quadratic sum that the feature gathering answered is closed is minimum；
S35, after recalculating the cluster centre that each feature gathering is closed, S34 is returned to step, until all cluster centres are equal
No longer change.
8. thyroid cancer ultrasound picture automatic marking method according to claim 1, which is characterized in that the step
S5 is specially：
Each feature in the cancer feature cluster obtained for extraction, according to its corresponding coordinate information, in cancer picture
Corresponding position in artwork is marked this feature using preset tag format.
9. a kind of thyroid cancer ultrasound picture automatic marking system, which is characterized in that including：
At least one processor；
At least one processor, for storing at least one program；
When at least one program is executed by least one processor so that at least one processor is realized as weighed
Profit requires a kind of thyroid cancer ultrasound picture automatic marking method of 1-8 any one of them.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810298494.4A CN108681731A (en) | 2018-04-03 | 2018-04-03 | A kind of thyroid cancer ultrasound picture automatic marking method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810298494.4A CN108681731A (en) | 2018-04-03 | 2018-04-03 | A kind of thyroid cancer ultrasound picture automatic marking method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108681731A true CN108681731A (en) | 2018-10-19 |
Family
ID=63800797
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810298494.4A Pending CN108681731A (en) | 2018-04-03 | 2018-04-03 | A kind of thyroid cancer ultrasound picture automatic marking method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108681731A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126470A (en) * | 2019-12-18 | 2020-05-08 | 创新奇智(青岛)科技有限公司 | Image data iterative clustering analysis method based on depth metric learning |
| CN111897984A (en) * | 2020-05-28 | 2020-11-06 | 广州市玄武无线科技股份有限公司 | Picture labeling method and device, terminal equipment and storage medium |
| WO2021189900A1 (en) * | 2020-10-14 | 2021-09-30 | 平安科技（深圳）有限公司 | Medical image analysis method and apparatus, and electronic device and readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120008839A1 (en) * | 2010-07-07 | 2012-01-12 | Olympus Corporation | Image processing apparatus, method of processing image, and computer-readable recording medium |
| CN103324853A (en) * | 2013-06-25 | 2013-09-25 | 上海交通大学 | Similarity calculation system and method based on medical image features |
| CN105139390A (en) * | 2015-08-14 | 2015-12-09 | 四川大学 | Image processing method for detecting pulmonary tuberculosis focus in chest X-ray DR film |
| CN105427296A (en) * | 2015-11-11 | 2016-03-23 | 北京航空航天大学 | Ultrasonic image low-rank analysis based thyroid lesion image identification method |
| CN105654490A (en) * | 2015-12-31 | 2016-06-08 | 中国科学院深圳先进技术研究院 | Lesion region extraction method and device based on ultrasonic elastic image |
| CN106023239A (en) * | 2016-07-05 | 2016-10-12 | 东北大学 | Breast lump segmentation system and method based on mammary gland subarea density clustering |
-
2018
- 2018-04-03 CN CN201810298494.4A patent/CN108681731A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120008839A1 (en) * | 2010-07-07 | 2012-01-12 | Olympus Corporation | Image processing apparatus, method of processing image, and computer-readable recording medium |
| CN103324853A (en) * | 2013-06-25 | 2013-09-25 | 上海交通大学 | Similarity calculation system and method based on medical image features |
| CN105139390A (en) * | 2015-08-14 | 2015-12-09 | 四川大学 | Image processing method for detecting pulmonary tuberculosis focus in chest X-ray DR film |
| CN105427296A (en) * | 2015-11-11 | 2016-03-23 | 北京航空航天大学 | Ultrasonic image low-rank analysis based thyroid lesion image identification method |
| CN105654490A (en) * | 2015-12-31 | 2016-06-08 | 中国科学院深圳先进技术研究院 | Lesion region extraction method and device based on ultrasonic elastic image |
| CN106023239A (en) * | 2016-07-05 | 2016-10-12 | 东北大学 | Breast lump segmentation system and method based on mammary gland subarea density clustering |
Non-Patent Citations (1)
| Title |
|---|
| DAVID ARTHUR，ET AL.: ""k-means++: the advantages of careful seeding"", 《SODA "07: PROCEEDINGS OF THE EIGHTEENTH ANNUAL ACM-SIAM SYMPOSIUM ON DISCRETE ALGORITHMS》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126470A (en) * | 2019-12-18 | 2020-05-08 | 创新奇智(青岛)科技有限公司 | Image data iterative clustering analysis method based on depth metric learning |
| CN111897984A (en) * | 2020-05-28 | 2020-11-06 | 广州市玄武无线科技股份有限公司 | Picture labeling method and device, terminal equipment and storage medium |
| WO2021189900A1 (en) * | 2020-10-14 | 2021-09-30 | 平安科技（深圳）有限公司 | Medical image analysis method and apparatus, and electronic device and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111476292B (en) | Small sample element learning training method for medical image classification processing artificial intelligence | |
| CN110503654A (en) | A kind of medical image cutting method, system and electronic equipment based on generation confrontation network | |
| Yang et al. | Deep plastic surgery: Robust and controllable image editing with human-drawn sketches | |
| CN104850633B (en) | A kind of three-dimensional model searching system and method based on the segmentation of cartographical sketching component | |
| US10346728B2 (en) | Nodule detection with false positive reduction | |
| Saha et al. | Skeletonization: Theory, methods and applications | |
| WO2022001623A1 (en) | Image processing method and apparatus based on artificial intelligence, and device and storage medium | |
| CN109740652A (en) | A kind of pathological image classification method and computer equipment | |
| Schultz et al. | Open-box spectral clustering: applications to medical image analysis | |
| CN102651128B (en) | Image set partitioning method based on sampling | |
| CN108416776A (en) | Image-recognizing method, pattern recognition device, computer product and readable storage medium storing program for executing | |
| CN110147483A (en) | A kind of title method for reconstructing and device | |
| CN108681731A (en) | A kind of thyroid cancer ultrasound picture automatic marking method and system | |
| CN106874489A (en) | A kind of Lung neoplasm image block search method and device based on convolutional neural networks | |
| Casaca et al. | Laplacian coordinates: Theory and methods for seeded image segmentation | |
| CN110796135A (en) | Target positioning method and device, computer equipment and computer storage medium | |
| Seidl et al. | Automated classification of petroglyphs | |
| Zhao et al. | Fine-grained diabetic wound depth and granulation tissue amount assessment using bilinear convolutional neural network | |
| CN110334566A (en) | Fingerprint extraction method inside and outside a kind of OCT based on three-dimensional full convolutional neural networks | |
| Wang et al. | Segmenting crop disease leaf image by modified fully-convolutional networks | |
| Kang et al. | Image-based modeling of plants and trees | |
| CN114419087A (en) | Focus image generation method and device, electronic equipment and storage medium | |
| CN110363103A (en) | Identifying pest method, apparatus, computer equipment and storage medium | |
| Zhang et al. | Classification of benign and malignant pulmonary nodules based on deep learning | |
| Lengauer et al. | A sketch-aided retrieval approach for incomplete 3D objects |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181019 |