CN109948614B - Wrist bone interest area cutting method based on machine learning - Google Patents
Wrist bone interest area cutting method based on machine learning Download PDFInfo
- Publication number
- CN109948614B CN109948614B CN201910231454.2A CN201910231454A CN109948614B CN 109948614 B CN109948614 B CN 109948614B CN 201910231454 A CN201910231454 A CN 201910231454A CN 109948614 B CN109948614 B CN 109948614B
- Authority
- CN
- China
- Prior art keywords
- bone
- wrist
- age
- wrist bone
- machine learning
- 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
Landscapes
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a wrist bone interest area cutting method based on machine learning. According to the method, on the premise that the size of each individual is independent, accurate cutting is performed on each wrist bone picture, and features are extracted. The invention can generate the self-adaptive cutting frame for any height and age, and provides data support for more accurate judgment of the wrist bone grade. The method can be applied to feature region segmentation in the field of image recognition, and greatly improves the recognition rate of a deep learning network.
Description
Technical Field
The invention relates to the technical field of self-adaptive cutting of medical pictures, in particular to a wrist bone interest area cutting method based on machine learning.
Background
Deep learning has a very wide prospect in the fields of image recognition, natural language processing, voice recognition and the like. With the wide application of deep learning technology in medical images, the automatic bone age assessment also becomes a big hotspot.
The maturity set average method based on difference analysis automatic weighting, namely CHN method, is an evaluation method aiming at Chinese human carpal bones, and the method makes corresponding grade evaluation on 14 wrist bones. And learning the data set of 14 wrist bone medical images by combining a deep learning network so as to achieve the purpose of automatically judging the bone age. Because the sizes of 14 bones are unbalanced for wrist bone age images of different heights and different age groups, the larger the age is, the larger the 14 bones are, the more accurate extraction can not be completed through a fixed size area, and the 14 independent characteristic areas separated from the whole wrist bone medical picture need to depend on a large amount of manual cutting images, which wastes time and labor. Therefore, how to improve the feature region image extraction by using the cutting frame size adaptive method is an important issue worthy of continuous and intensive research.
Disclosure of Invention
The invention aims to solve the problem that the sizes of cutting frames required by different heights and ages of wrist bone medical images cannot be uniformly fixed, and provides an automatic cutting method of size self-adaption of a bone age image based on machine learning for segmenting the wrist bone medical images.
The technical scheme adopted by the invention for solving the technical problems is as follows:
A wrist bone interest region cutting method based on machine learning comprises the following steps:
1) selecting a batch of wrist bone samples with different ages and heights according to actual conditions, and calibrating the geometric center point of 14 bones aimed by the CHN bone age assessment method for each sample slice; the 14 bones are respectively: radius, palm 1, palm 3, palm 5, proximal 1, proximal 3, proximal 5, middle 3, middle 5, distal 1, distal 3, distal 5, capitate, hamate;
2) for each sample piece, further marking a rectangular cutting area which has proper length and width and can completely wrap the picture information required by bone age grade identification for each bone based on the geometric central point marked in the previous step, and completing the data set manufacturing required by the machine learning of the next step;
3) taking the age and the height of each wrist bone sample as input, taking the size of the interest area of 14 bones of the wrist bone sample as output, and training a sample set by utilizing a machine learning network to obtain a trained model;
4) when 14 bone interest areas of a new wrist bone piece with the bone age to be evaluated need to be cut, inputting age and height information corresponding to the wrist bone piece into the trained wrist bone cutting network model, and returning the size of the 14 bone interest areas by the model;
5) In the wrist bone piece with the bone age to be evaluated, for each wrist bone, the geometric center point of the wrist bone is manually marked, and then the 14 bones are cut out based on the size of the interest region of the 14 bones output in the step 4).
The invention has the following beneficial effects: the invention utilizes a machine learning method to segment the medical image, and achieves better effect by combining the central point cutting and the size self-adaption of the cutting frame. According to the method, on the premise that the size of each individual is independent, accurate cutting is performed on each wrist bone picture, and features are extracted. The invention can generate the self-adaptive cutting frame for any height and age, and provides data support for more accurate judgment of the wrist bone grade. The method can be applied to feature region segmentation in the field of image recognition, and greatly improves the recognition rate of a deep learning network.
Detailed Description
The technical solution of the present invention is further illustrated by the following examples.
Examples
The embodiment provides a wrist bone interest region cutting method based on machine learning, which comprises the following steps:
1) selecting a batch of wrist bone samples with different ages and heights according to actual conditions, and calibrating the geometric center point of 14 bones aimed at by the CHN bone age assessment method for each sample piece; the 14 bones are respectively: radius, palm 1, palm 3, palm 5, proximal 1, proximal 3, proximal 5, middle 3, middle 5, distal 1, distal 3, distal 5, capitate bone, hamate bone;
2) For each sample piece, further marking a rectangular cutting area which has proper length and width and can completely wrap the picture information required by bone age grade identification for each bone based on the geometric central point marked in the previous step, and completing the data set manufacture required by the machine learning of the next step;
3) taking the age and the height of each wrist bone sample as input, taking the size of the interest area of 14 bones of the wrist bone sample as output, and training a sample set by utilizing a machine learning network to obtain a trained model;
4) when 14 bone interest areas of a new wrist bone piece with the bone age to be evaluated need to be cut, inputting age and height information corresponding to the wrist bone piece into the trained wrist bone cutting network model, and returning the size of the 14 bone interest areas by the model;
5) in the wrist bone piece with the bone age to be evaluated, for each wrist bone, the geometric center point of the wrist bone is manually marked, and then the 14 bones are cut out based on the size of the interest region of the 14 bones output in the step 4).
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (1)
1. A wrist bone interest region cutting method based on machine learning is characterized by comprising the following steps:
1) selecting a batch of wrist bone samples with different ages and heights according to actual conditions, and calibrating the geometric center point of 14 bones aimed at by the CHN bone age assessment method for each sample piece; the 14 bones are respectively: radius, palm 1, palm 3, palm 5, proximal 1, proximal 3, proximal 5, middle 3, middle 5, distal 1, distal 3, distal 5, capitate bone, hamate bone;
2) for each sample piece, further marking a rectangular cutting area which has proper length and width and can completely wrap the picture information required by bone age grade identification for each bone based on the geometric central point marked in the previous step, and completing the data set manufacture required by the machine learning of the next step;
3) taking the age and the height of each wrist bone sample as input, taking the size of the interest area of 14 bones of the wrist bone sample as output, and training a sample set by utilizing a machine learning network to obtain a trained model;
4) when 14 bone interest areas of a new wrist bone piece with the bone age to be evaluated need to be cut, inputting age and height information corresponding to the wrist bone piece into the trained wrist bone cutting network model, and returning the size of the 14 bone interest areas by the model;
5) In the wrist bone fragment with the bone age to be evaluated, for each wrist bone, the geometric center point of the wrist bone is manually marked, and then the 14 bones are cut out based on the sizes of the 14 bone interest areas output in the step 4).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910231454.2A CN109948614B (en) | 2019-03-26 | 2019-03-26 | Wrist bone interest area cutting method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910231454.2A CN109948614B (en) | 2019-03-26 | 2019-03-26 | Wrist bone interest area cutting method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109948614A CN109948614A (en) | 2019-06-28 |
| CN109948614B true CN109948614B (en) | 2022-06-28 |
Family
ID=67011914
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910231454.2A Active CN109948614B (en) | 2019-03-26 | 2019-03-26 | Wrist bone interest area cutting method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109948614B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110782450B (en) * | 2019-10-31 | 2020-09-29 | 北京推想科技有限公司 | Hand carpal development grade determining method and related equipment |
| CN111027571B (en) * | 2019-11-29 | 2022-03-01 | 浙江工业大学 | Wrist reference bone characteristic region self-adaptive extraction method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN107767376A (en) * | 2017-11-02 | 2018-03-06 | 西安邮电大学 | X-ray film stone age Forecasting Methodology and system based on deep learning |
| CN107997778A (en) * | 2016-10-31 | 2018-05-08 | 西门子保健有限责任公司 | The bone based on deep learning removes in computed tomography angiography art |
| CN108836338A (en) * | 2018-04-04 | 2018-11-20 | 浙江康体汇科技有限公司 | A kind of calculating of online stone age and prediction of height method based on web database |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017022908A1 (en) * | 2015-08-04 | 2017-02-09 | 재단법인 아산사회복지재단 | Method and program for bone age calculation using deep neural networks |
-
2019
- 2019-03-26 CN CN201910231454.2A patent/CN109948614B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107997778A (en) * | 2016-10-31 | 2018-05-08 | 西门子保健有限责任公司 | The bone based on deep learning removes in computed tomography angiography art |
| 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 |
| CN107767376A (en) * | 2017-11-02 | 2018-03-06 | 西安邮电大学 | X-ray film stone age Forecasting Methodology and system based on deep learning |
| CN108836338A (en) * | 2018-04-04 | 2018-11-20 | 浙江康体汇科技有限公司 | A kind of calculating of online stone age and prediction of height method based on web database |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109948614A (en) | 2019-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111860533B (en) | Image recognition method and device, storage medium and electronic device | |
| CN109948614B (en) | Wrist bone interest area cutting method based on machine learning | |
| CN109740019A (en) | A kind of method, apparatus to label to short-sighted frequency and electronic equipment | |
| CN103914689A (en) | Picture cropping method and device based on face recognition | |
| CN108846828A (en) | A kind of pathological image target-region locating method and system based on deep learning | |
| CN111400536B (en) | Low-cost tomato leaf disease identification method based on lightweight deep neural network | |
| CN113139977B (en) | Mouth cavity curve image wisdom tooth segmentation method based on YOLO and U-Net | |
| CN109522775A (en) | Face character detection method, device and electronic equipment | |
| CN110751606A (en) | Foam image processing method and system based on neural network algorithm | |
| CN104318565B (en) | Interactive method for retinal vessel segmentation based on bidirectional region growing of constant-gradient distance | |
| CN107679469A (en) | A kind of non-maxima suppression method based on deep learning | |
| CN115984698A (en) | Litchi fruit growing period identification method based on improved YOLOv5 | |
| CN115661187A (en) | Image enhancement method for Chinese medicinal preparation analysis | |
| CN113521683A (en) | Intelligent physical ability comprehensive training control system | |
| CN110189327A (en) | Eye ground blood vessel segmentation method based on structuring random forest encoder | |
| CN117636314A (en) | Seedling missing identification method, device, equipment and medium | |
| CN111899214A (en) | Pathological section scanning analysis device and pathological section scanning method | |
| CN111738151B (en) | Grape fruit stem accurate identification method based on deep learning post-optimization | |
| CN114937172A (en) | Grape disease identification method and system based on improved YOLOV3 | |
| CN102129575A (en) | Pornographic image analysis system based on color space skin color model | |
| CN114022717A (en) | Blade disease identification method and system based on image identification | |
| CN109241885A (en) | A kind of method and device of special area in intelligent recognition video | |
| CN110580702B (en) | Method for abdominal aortic aneurysm boundary segmentation | |
| CN104636741B (en) | formula identification method and device | |
| CN113689438A (en) | Grape fruit grain segmentation method aiming at fruit grain edge |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |