CN110599456A - Method for extracting specific region of medical image - Google Patents
Method for extracting specific region of medical image Download PDFInfo
- Publication number
- CN110599456A CN110599456A CN201910746225.4A CN201910746225A CN110599456A CN 110599456 A CN110599456 A CN 110599456A CN 201910746225 A CN201910746225 A CN 201910746225A CN 110599456 A CN110599456 A CN 110599456A
- Authority
- CN
- China
- Prior art keywords
- candidate
- target
- rectangles
- medical image
- function
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- 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
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention relates to the technical field of medical image processing, in particular to a method for extracting a specific region of a medical image, which comprises the following steps of firstly, acquiring a position coordinate set and a confidence coefficient of a target; secondly, inputting the position coordinate set of the target into a first function to generate a candidate rectangle; inputting the candidate rectangles into an overlapping rate function and calculating to obtain the overlapping rate of the candidate rectangles; thirdly, when the overlapping rate of the candidate rectangles is judged to reach the preset threshold value, the candidate rectangles generate a candidate set according to the second function; accumulating the confidence degrees of the candidate sets, sequencing and updating the candidate sets; and finally, reselecting the candidate rectangle according to the position coordinate set of the target to generate a specific area set. Has the advantages that: according to the technical scheme, a region overlapping calculation method can be adopted, extraction of interest regions of large-scale images is achieved, blanks of related fields are made up, and interference of massive visual fields on effective regions is reduced; unsupervised specific area extraction is realized, training is not needed, and the influence of the training data quality on the algorithm quality is reduced.
Description
Technical Field
The invention relates to the technical field of medical image processing, in particular to a method for extracting a specific region of a medical image.
Background
With the development of computer and electronic technology, the image processing technology can carry out quantitative analysis on the form, and the medical image processing is to apply the digital image processing technology to the medical field and use the image processing technology to carry out the analysis and processing of the image; image processing is mainly to detect and measure objects of interest in an image, and obtain their objective information.
Since the last 90 s, with the development of image analysis technology and the slow penetration of image processing technology into the medical field, a new interdisciplinary formation, i.e., medical image processing, has been developed. The image processing technology is combined with a medical analysis method, the morphological quantification is developed from the morphological quantification aspect, and the method is a brand new analysis method, and can enable the measurement to be more accurate, more efficient and faster. The intersection and fusion of computer technology and medicine has increasingly greater effect on the research of medical research and clinical practice, and makes the diagnosis of medicine more direct and clearer.
However, in the current stage of target detection technology, only a method for extracting interest areas from small images is used, and large images are more and more widely applied with the increase of storage media, and the requirement for randomly selecting interest areas through a simple sliding window cannot be met.
Disclosure of Invention
Aiming at the defects and problems in the prior art, the invention discloses a method for extracting a specific area of a medical image, which is suitable for a large-scale medical image processing process, and comprises the following specific technical scheme:
the method for extracting the specific area comprises the following steps:
step S1, acquiring a position coordinate set and a confidence of the target;
step S2, inputting the position coordinate set of the target into a first function to generate a candidate rectangle;
step S3, inputting the candidate rectangle into an overlapping rate function and calculating to obtain the overlapping rate of the candidate rectangle;
step S4, when the overlapping rate of the candidate rectangles reaches a preset threshold, the candidate rectangles generate a candidate set according to a second function;
step S5, accumulating the confidence degrees of the candidate set, sorting and updating the candidate set;
step S6, reselecting the candidate rectangle according to the position coordinate set of the target to generate a specific area set.
Preferably, the step S1 includes:
step S11, obtaining the position coordinates of each target and the confidence corresponding to the target according to a target detection algorithm;
step S12, a set is formed by the position coordinates of each obtained target and the confidence corresponding to the target to obtain the position coordinate set and the confidence of the target.
Preferably, each of the acquired targets corresponds to four circumscribed rectangles.
Preferably, the first function is
Where R is used to represent the set of candidate rectangles, b1、b2、b3、b4A coordinate set, x, for representing four candidate circumscribed rectangles correspondingly generated by each target1,y1Respectively for representing the abscissa and ordinate, x, of a first reference point of said target2,y2And w and h are respectively used for representing the length and the width of the circumscribed rectangle.
Preferably, the overlap ratio function isWherein the IOU is used to represent the candidateThe overlap ratio of the rectangles; a is used for representing the intersection area of any two candidate rectangles; s is used for representing the area sum of any two candidate rectangles.
Preferably, the preset threshold is 0.5.
Preferably, the second function is [ (R, O) R ∈ R, O ∈ O, IOU ≧ 0.5 ];
wherein R is used to represent the set of candidate rectangles; o is used to represent the set of location coordinates of the target; r is used to represent any element in the set R of candidate rectangles; the O is used to represent any element in the set of location coordinates O of the target.
Has the advantages that: according to the technical scheme, the method for calculating the overlapping rate of the specific area is used, the extraction of the interest area of the large-scale image is realized, the blank of the related field is made up, and the interference of a large number of visual fields on the effective area is reduced; unsupervised specific area extraction is realized, training is not needed, and the influence of the training data quality on the algorithm quality is reduced.
Drawings
FIG. 1 is a flow chart illustrating the steps of a method for extracting a specific region of a medical image according to the present invention;
fig. 2 is a flowchart illustrating the specific steps of step S1 of the method for extracting specific regions of a medical image according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The invention is further described with reference to the following drawings and specific examples, which are not intended to be limiting.
The invention discloses a method for extracting a specific region of a medical image, which comprises the following specific technical scheme:
the method for extracting the specific area comprises the following steps:
step S1, acquiring a position coordinate set and a confidence of the target;
step S2, inputting the position coordinate set of the target into a first function to generate a candidate rectangle;
step S3, inputting the candidate rectangle into an overlapping rate function and calculating to obtain the overlapping rate of the candidate rectangle;
step S4, when the overlapping rate of the candidate rectangles reaches a preset threshold, the candidate rectangles generate a candidate set according to a second function;
step S5, accumulating the confidence degrees of the candidate sets, sequencing and updating the candidate sets;
in step S6, the candidate rectangle is reselected according to the position coordinate set of the target to generate a specific area set.
Specifically, as shown in fig. 1, in the case that the medical image is negative, a corresponding specific region of interest is generated by the following steps;
further, firstly, acquiring a position coordinate set and a confidence coefficient of a target; secondly, inputting the position coordinate set of the target into a first function to generate a candidate rectangle; inputting the candidate rectangle into an overlapping rate function and calculating to obtain the overlapping rate of the candidate rectangle; thirdly, when the overlapping rate of the candidate rectangles is judged to reach a preset threshold value, the candidate rectangles generate a candidate set according to a second function; accumulating the confidence degrees of the candidate sets, sequencing and updating the candidate sets; and finally, reselecting the candidate rectangle according to the position coordinate set of the target to generate a specific area set.
In a preferred embodiment, step S1 includes:
step S11, obtaining the position coordinate of each target and the corresponding confidence of the target according to a target detection algorithm;
step S12, a set is formed by the obtained position coordinates of each acquisition target and the confidence level corresponding to the target to acquire a position coordinate set and a confidence level of the target.
Specifically, as shown in fig. 2, the specific step of acquiring the position coordinate set and the confidence of the target includes: firstly, obtaining the position coordinate of each target and the corresponding confidence of the target according to a target detection algorithm; secondly, the obtained position coordinates of each acquired target and the confidence corresponding to the target are combined into a set to acquire a position coordinate set and a confidence of the target.
In a preferred embodiment, there are four circumscribed rectangles for each acquisition target.
Specifically, each target element O in the target position coordinate set O generates four corresponding circumscribed rectangles.
In a preferred embodiment, the first function is
Where R is used to represent a set of candidate rectangles, b1、b2、b3、b4A coordinate set, x, for representing four candidate circumscribed rectangles generated corresponding to each target1,y1Respectively for representing the abscissa and ordinate, x, of a first reference point of the object2,y2And w, h are respectively used to represent the length and width of the circumscribed rectangle.
Specifically, the position coordinate set of the target is input into a first function to generate a candidate rectangle, and the following steps are performed.
In a preferred embodiment, the overlap ratio function isThe IOU is used for representing the overlapping rate of the candidate rectangles; a is used for representing the intersection area of any two candidate rectangles; s is used to represent the sum of the areas of any two candidate rectangles.
Specifically, in an evaluation system for target detection, a function is called an IOU (Intersection over Union), which is simply an overlap ratio of a predetermined rectangle and a candidate rectangle, and is used for measuring object detection in any size and shape, that is, for measuring a correlation between reality and prediction, and the higher the correlation, the higher the value.
In a preferred embodiment, the predetermined threshold is 0.5.
In a preferred embodiment, the second function is [ (R, O) | R ∈ R, O ∈ O, IOU ≧ 0.5 ];
wherein R is used to represent a set of candidate rectangles; o is used to represent the set of position coordinates of the target; r is used to represent any element in the set R of candidate rectangles; o is used to represent any element in the set of location coordinates O of the object.
Specifically, firstly, when the overlapping rate of the candidate rectangles is judged to reach a preset threshold IOU (input/output) which is not less than 0.5, the candidate rectangles generate a candidate set according to a second function; secondly, accumulating the confidence degrees of the candidate sets, sequencing and updating the candidate sets; and finally, reselecting the candidate rectangle according to the position coordinate set of the target to generate a specific area set.
Further, sorting is carried out according to the numerical value of the confidence coefficient, n elements with the highest numerical value of the confidence coefficient are selected, and the candidate rectangle r repeatedly comprising the target O in the confidence coefficient sorting is deleted according to the uniqueness of any element O in the position coordinate set O of the target.
Has the advantages that: according to the technical scheme, a region overlapping calculation method can be adopted, extraction of interest regions of large-scale images is achieved, blanks of related fields are made up, and interference of massive visual fields on effective regions is reduced; unsupervised specific area extraction is realized, training is not needed, and the influence of the training data quality on the algorithm quality is reduced.
While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.
Claims (7)
1. A method for extracting a specific region of a medical image, which is suitable for a large-scale medical image processing process, is characterized in that the method for extracting the specific region comprises the following steps:
step S1, acquiring a position coordinate set and a confidence of the target;
step S2, inputting the position coordinate set of the target into a first function to generate a candidate rectangle;
step S3, inputting the candidate rectangle into an overlapping rate function and calculating to obtain the overlapping rate of the candidate rectangle;
step S4, when the overlapping rate of the candidate rectangles reaches a preset threshold, the candidate rectangles generate a candidate set according to a second function;
step S5, accumulating the confidence degrees of the candidate set, sorting and updating the candidate set;
step S6, reselecting the candidate rectangle according to the position coordinate set of the target to generate a specific area set.
2. The method for extracting specific region of medical image according to claim 1, wherein said step S1 includes:
step S11, obtaining the position coordinates of each target and the confidence corresponding to the target according to a target detection algorithm;
step S12, a set is formed by the position coordinates of each obtained target and the confidence corresponding to the target to obtain the position coordinate set and the confidence of the target.
3. The method for extracting specific region of medical image according to claim 2, wherein there are four circumscribed rectangles corresponding to each of the acquired targets.
4. The method for extracting specific region of medical image according to claim 3, wherein said first function is
Where R is used to represent the set of candidate rectangles, b1、b2、b3、b4A coordinate set, x, for representing four candidate circumscribed rectangles correspondingly generated by each target1,y1Respectively for representing the abscissa and ordinate, x, of a first reference point of said target2,y2And w and h are respectively used for representing the length and the width of the circumscribed rectangle.
5. The method for extracting specific region of medical image according to claim 1, wherein the overlap ratio function is
Wherein the IOU is used for representing the overlapping rate of the candidate rectangles; a is used for representing the intersection area of any two candidate rectangles; s is used for representing the area sum of any two candidate rectangles.
6. The method for extracting specific region of medical image according to claim 1, wherein the preset threshold is 0.5.
7. The method for extracting the specific region of the medical image according to claim 1, wherein the second function is [ (R, O) | R ∈ R, O ∈ O, IOU ≧ 0.5 ];
wherein R is used to represent the set of candidate rectangles; o is used to represent the set of location coordinates of the target; r is used to represent any element in the set R of candidate rectangles; the O is used to represent any element in the set of location coordinates O of the target.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910746225.4A CN110599456B (en) | 2019-08-13 | 2019-08-13 | Method for extracting specific region of medical image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910746225.4A CN110599456B (en) | 2019-08-13 | 2019-08-13 | Method for extracting specific region of medical image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110599456A true CN110599456A (en) | 2019-12-20 |
| CN110599456B CN110599456B (en) | 2023-05-30 |
Family
ID=68854095
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910746225.4A Active CN110599456B (en) | 2019-08-13 | 2019-08-13 | Method for extracting specific region of medical image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110599456B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112613568A (en) * | 2020-12-29 | 2021-04-06 | 清华大学 | Target identification method and device based on visible light and infrared multispectral image sequence |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170124415A1 (en) * | 2015-11-04 | 2017-05-04 | Nec Laboratories America, Inc. | Subcategory-aware convolutional neural networks for object detection |
| CN108062574A (en) * | 2017-12-31 | 2018-05-22 | 厦门大学 | A kind of Weakly supervised object detection method based on particular category space constraint |
| CN108121986A (en) * | 2017-12-29 | 2018-06-05 | 深圳云天励飞技术有限公司 | Object detection method and device, computer installation and computer readable storage medium |
| CN108256404A (en) * | 2016-12-29 | 2018-07-06 | 北京旷视科技有限公司 | Pedestrian detection method and device |
| CN109344768A (en) * | 2018-09-29 | 2019-02-15 | 南京理工大学 | Pointer breaker recognition methods based on crusing robot |
| CN109697719A (en) * | 2019-03-05 | 2019-04-30 | 百度在线网络技术（北京）有限公司 | A kind of image quality measure method, apparatus and computer readable storage medium |
| CN109977945A (en) * | 2019-02-26 | 2019-07-05 | 博众精工科技股份有限公司 | Localization method and system based on deep learning |
-
2019
- 2019-08-13 CN CN201910746225.4A patent/CN110599456B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170124415A1 (en) * | 2015-11-04 | 2017-05-04 | Nec Laboratories America, Inc. | Subcategory-aware convolutional neural networks for object detection |
| CN108256404A (en) * | 2016-12-29 | 2018-07-06 | 北京旷视科技有限公司 | Pedestrian detection method and device |
| CN108121986A (en) * | 2017-12-29 | 2018-06-05 | 深圳云天励飞技术有限公司 | Object detection method and device, computer installation and computer readable storage medium |
| CN108062574A (en) * | 2017-12-31 | 2018-05-22 | 厦门大学 | A kind of Weakly supervised object detection method based on particular category space constraint |
| CN109344768A (en) * | 2018-09-29 | 2019-02-15 | 南京理工大学 | Pointer breaker recognition methods based on crusing robot |
| CN109977945A (en) * | 2019-02-26 | 2019-07-05 | 博众精工科技股份有限公司 | Localization method and system based on deep learning |
| CN109697719A (en) * | 2019-03-05 | 2019-04-30 | 百度在线网络技术（北京）有限公司 | A kind of image quality measure method, apparatus and computer readable storage medium |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112613568A (en) * | 2020-12-29 | 2021-04-06 | 清华大学 | Target identification method and device based on visible light and infrared multispectral image sequence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110599456B (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108898047B (en) | Pedestrian detection method and system based on blocking and shielding perception | |
| Hoang et al. | Metaheuristic optimized edge detection for recognition of concrete wall cracks: a comparative study on the performances of roberts, prewitt, canny, and sobel algorithms | |
| CN110766095B (en) | Defect detection method based on image gray level features | |
| CN110599537A (en) | Mask R-CNN-based unmanned aerial vehicle image building area calculation method and system | |
| US8019164B2 (en) | Apparatus, method and program product for matching with a template | |
| CN109934224B (en) | Small target detection method based on Markov random field and visual contrast mechanism | |
| CN108986152B (en) | Foreign matter detection method and device based on difference image | |
| CN110033040B (en) | Flame identification method, system, medium and equipment | |
| CN110930427B (en) | Image segmentation method, device and storage medium based on semantic contour information | |
| US20220114724A1 (en) | Image processing model generation method, image processing method and device, and electronic device | |
| JP2014228357A (en) | Crack detecting method | |
| CN107301408A (en) | Human body mask extracting method and device | |
| CN115908988B (en) | Defect detection model generation method, device, equipment and storage medium | |
| CN115375917B (en) | Target edge feature extraction method, device, terminal and storage medium | |
| CN112488999A (en) | Method, system, storage medium and terminal for detecting small target in image | |
| CN112634368A (en) | Method and device for generating space and OR graph model of scene target and electronic equipment | |
| CN111539456B (en) | Target identification method and device | |
| CN111415364A (en) | Method, system and storage medium for converting image segmentation samples in computer vision | |
| CN115861715A (en) | Knowledge representation enhancement-based image target relation recognition algorithm | |
| CN116109812A (en) | Target detection method based on non-maximum suppression threshold optimization | |
| CN108764343B (en) | Method for positioning tracking target frame in tracking algorithm | |
| CN108876776B (en) | Classification model generation method, fundus image classification method and device | |
| CN114219936A (en) | Object detection method, electronic device, storage medium, and computer program product | |
| CN110599456A (en) | Method for extracting specific region of medical image | |
| CN115631370A (en) | Identification method and device of MRI (magnetic resonance imaging) sequence category based on convolutional neural network |
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 |