CN109635871A - A kind of capsule endoscope image classification method based on multi-feature fusion - Google Patents
A kind of capsule endoscope image classification method based on multi-feature fusion Download PDFInfo
- Publication number
- CN109635871A CN109635871A CN201811515944.7A CN201811515944A CN109635871A CN 109635871 A CN109635871 A CN 109635871A CN 201811515944 A CN201811515944 A CN 201811515944A CN 109635871 A CN109635871 A CN 109635871A
- Authority
- CN
- China
- Prior art keywords
- image
- capsule endoscope
- feature
- method based
- color
- 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
- 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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06V2201/03—Recognition of patterns in medical or anatomical images
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Endoscopes (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
A kind of capsule endoscope image classification method based on multi-feature fusion classifies to capsule endoscope image using classifier by blending image color characteristic and textural characteristics.Compared with prior art, the textural characteristics that this method uses combine colouring information.The detection of inactive area is the adaptive detection algorithm based on super-pixel, can automatically detect out dark space and clear zone in image, realizes and carry out feature extraction in effective coverage, avoids the interference of invalid information.Fusion Features use distinguishing ability analytical technology, and the new feature of fusion is made to have more discrimination.The present invention can automatically classify capsule endoscope image, so as to shorten the time of review, alleviate the burden of doctor.
Description
Technical field
The invention belongs to technical field of image processing more particularly to a kind of capsule endoscopes based on multi-feature fusion
Image classification method.
Background technique
Conventional hand-held endoscopy can not be related to entire complete alimentary canal, and gastroscope can check upper digestive tract,
And colonoscopy can only check colon and rectum, therefore be that conventional endoscope can not be visited and be arrived there are also partial enteral.Capsule endoscope
Appearance so that the entire alimentary canal of human body can be observed.In general, patient swallows capsule endoscope from oral cavity, pass through stomach
The wriggling of enteron aisle, capsule endoscope stop about 8 hours in vivo, and shoot image by the rate of 2 frame per second.Patient understands satellite and takes
One reception device of band, capsule endoscope shooting image will be transmitted on this receiver, finally all endoscopic images with
The mode of video sequence exports to computer, is analyzed by the software of profession for clinician.Surpass however, capsule endoscope will generate
50000 images are crossed, this is the extremely time-consuming work for taking energy again for doctor's review.Therefore, it develops and is peeped in capsule
Mirror image automated organ categorizing system has important practical value for mitigating the burden of doctor.
Summary of the invention
In order to overcome the shortcomings of that the prior art can not effectively realize capsule endoscope image, the present invention provides a kind of effective reality
The capsule endoscope image classification method based on multi-feature fusion of existing capsule endoscope image, by extracting characteristics of image, instruction
Practice classifier, capsule endoscope image is divided into esophagus, cardia, stomach, several main positions such as pylorus and duodenum, thus
The burden for mitigating doctor's review, improves efficiency.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of capsule endoscope image classification method based on multi-feature fusion, comprising the following steps:
A. it obtains capsule endoscope image set: by extracting capsule endoscope video frame, converting image for video sequence
Collection；
B. image preprocessing: pre-processing capsule endoscope image, generates inactive area mask (mask)；
C. color of image feature extraction: under HSI color space, by the method for color histogram, the face of image is extracted
Color characteristic；
D. image texture characteristic extracts: extracting image texture characteristic under color pattern image；
E. multi-features: the color of image and textural characteristics merge and dimensionality reduction；
F. image classification: fused new feature is trained classifier, then using classification to the figure newly obtained
As classifying, marks and which organoid be belonging respectively to.
Further, in the step b, image preprocessing includes the detection of dark areas and bright area, and constructs one in vain
The mask in region.
Further, in the step c, color characteristic is under HSI color space using form and aspect-saturation histogram
The extraction of color characteristic is carried out with discrete cosine transform (DCT).
Further, the step d, middle textural characteristics are mentioned using the colored constant three value mode (CSILTP) of part of gray scale
Take image texture characteristic.
In the step e, merge simultaneously dimensionality reduction to color and textural characteristics using distinguishing ability analysis (DPA)；
It in the step f, is trained using random forest grader, then utilizes trained classifier by new glue
Capsule endoscopic images are divided into respectively affiliated organ.
In the step f, it is trained using random forest grader, steps are as follows:
1) under the guidance of clinician, capsule endoscope image is classified；
2) endoscope non-image areas is marked with an overall situation mask；
3) inactive area is marked using dark space and clear zone detection algorithm, constructs inactive area mask；
4) color and textural characteristics of all images are extracted at global mask and inactive area mask；
5) Fusion Features are carried out using DPA technology, forms characteristic data set；
6) characteristic data set is randomly divided into 10 parts, takes the training of 9 parts of carry out classifier therein every time, remaining 1 part
For the classification performance of testing classification device；
7) by 10 subseries, results are averaged, compares with the classification results of doctor, completes after reaching requirement
The training of classifier；
In the step 1), according to the guidance of doctor, capsule endoscope image is divided into esophagus, cardia, stomach, pylorus and ten
Two duodenum 12.
Compared with the prior art, the present invention has the following beneficial effects:
1. the present invention utilizes the technology of machine learning, by existing capsule endoscope image zooming-out feature, training one
Then classifier classifies automatically to the image newly obtained, highly shortened the time of doctor's review, mention and alleviate doctor
Burden, improve efficiency.
2. the present invention is using adaptive detection algorithm label dark space and clear zone based on super-pixel, detection accuracy is high, and
Image border is more smooth.
3., according to the experience of clinician, color characteristic is main differentiation Different Organs present invention employs color characteristic
Judgment basis, and form and aspect-saturation histogram (HS) to different color have more identification.
4. the present invention extracts image texture characteristic using CSILTP, CSILTP feature has robustness to illumination variation, and
And to local insensitive for noise.
5. the present invention merges color characteristic using DPA technology, the problem for avoiding single features classifying quality bad,
The distinguishing ability for improving feature simultaneously, reduces characteristic dimension.
Detailed description of the invention
Fig. 1 is the flow chart of heretofore described capsule endoscope image classification.
Fig. 2 is heretofore described adaptive dark space clear zone detection algorithm flow chart.
Fig. 3 is heretofore described CSILTP Texture Segmentation Algorithm flow chart.
Fig. 4 is heretofore described DPA Feature Fusion Algorithm flow chart.
Fig. 5 is application result of the heretofore described adaptive dark space clear zone detection algorithm in practical endoscopic images.Its
Middle figure (a) is an antrum by juxtapyloric endoscopic images, and figure (b) is image (the inactive area use for marking inactive area
Black display).
Fig. 6 is classification results of the heretofore described CSILTP textural characteristics for endoscopic images, be compared traditional
Local binary patterns (LBP), the features such as uniform local binary patterns (ULBP) and homogeneous color local binary patterns (CULBP).
Fig. 7 is performance of the heretofore described DPA Feature Fusion Algorithm for different dimensions, compared traditional principal component point
Analyse the performance of (PCA).
Specific embodiment
The present invention is described further with reference to the accompanying drawing.
Referring to Fig.1~Fig. 7, a kind of capsule endoscope image classification method based on multi-feature fusion, comprising the following steps:
It obtains capsule endoscope image set: obtaining Wireless capsule endoscope video, by extracting endoscopic video frame, will regard
Frequency sequence is converted into image set, under the guidance of clinician, carries out manual sort to endoscopic images.
Image preprocessing: in capsule endoscope image, due to the movement of light source and the tubular structure of organ, endoscopic images
It will appear the high region in relatively darker region and brightness, the feature of these extracted regions can generate unfavorable shadow to classification
It rings, so to be pre-processed to capsule endoscope image；According to the imaging arrangement of original endoscope, to endoscope non-image areas
Domain construction overall situation mask marks inactive area by adaptive dark space clear zone detection algorithm, generates inactive area mask.
Color of image feature extraction: under HSI color space, by the method for color histogram, the color of image is extracted
Feature；Color characteristic is the main analysis foundation of clinician, and different organs has different degrees of discrimination in color.
Image texture characteristic extracts: extracting image texture characteristic under color pattern image；Textural characteristics are usually in ash
Feature extraction is carried out under degree image model, gray level image loses a large amount of color informations of original image, it is therefore desirable in conjunction with endoscope
Color information to extract texture special.
Multi-features: the color of image and textural characteristics merge simultaneously dimensionality reduction；Single features divide image
Class is unsatisfactory, so the scheme combined using multiple features, but the higher dimension of multiple features will cause dimension disaster, generate
Fitting phenomenon, it is therefore desirable to dimension-reduction treatment be carried out to multiple features, using principal component analysis (PCA) technology classification effect and paid no attention to
Think, therefore the present invention carries out fusion dimensionality reduction to multiple features using DPA, improves classification accuracy.
Image classification: fused new feature is trained classifier, adjusts optimal classification by cross validation
Then device parameter classifies to the image newly obtained using trained classifier, marks and which organoid be belonging respectively to.
Utilize above-mentioned capsule endoscope image classification method based on multi-feature fusion, comprising the following steps:
A. it obtains capsule endoscope image set: by extracting capsule endoscope video frame, converting image for video sequence
Collection；
B. image preprocessing: pre-processing capsule endoscope image, generates inactive area mask (mask)；
C. color of image feature extraction: under HSI color space, by the method for color histogram, the face of image is extracted
Color characteristic；
D. image texture characteristic extracts: extracting image texture characteristic under color pattern image；
E. multi-features: the color of image and textural characteristics merge and dimensionality reduction；
F. image classification: fused new feature is trained classifier, then using classification to the figure newly obtained
As classifying, marks and which organoid be belonging respectively to.
The method that the present invention uses machine learning, it is special by the color and texture of extracting capsule endoscope effective image area
Sign, after DPA Fusion Features, endoscopic images are divided into corresponding organ classification by training study random forest grader, from
And mitigate the review burden of doctor, it improves efficiency.
As shown in Fig. 2, adaptive shade and highlight regions detection algorithm are the algorithms based on super-pixel segmentation in step b.
First according to capsule endoscope imaging modality, overall situation mask is constructed.Super-pixel point is carried out to endoscopic images at global mask
It cuts, then converts the image into HSI color space, under luminance picture, calculate the average value of each piece of super-pixel, pass through searching
The smallest super-pixel of brightness is as a reference point in image, calculates dark space threshold value according to the brightness value of reference point, will be less than dark space threshold
All label is the super-pixel of value, finally exports dark space mask.Similarly, under saturation degree image, clear zone threshold value is calculated, it is defeated
Clear zone mask out.Finally merge dark space mask and clear zone mask, exports inactive area mask.
Further, threshold value I in above-mentioned steps bnCalculation method be to be obtained by the mathematical model of a piecewise function:
In formula (1), I is the value of any one reference point, TmaxIt is max-thresholds, that is, is more than after this threshold value, is no longer dark
Area or clear zone.A, b, c are three parameters in formula (1), according to condition:
1) threshold value is no more than Tmax；
2) difference of threshold value and reference point reduces with the increase of reference point；
3) the corresponding reference point of maximum threshold value is the smallest reference point in all samples.
It can be determined by the following by calculating these three parameters:
In formula (2), IsIt is the smallest reference point in all samples.So far, available interior using adaptive detection algorithm
The inactive area mask of sight glass image can reduce the interference of invalid information when extracting feature to avoid calculating.
As shown in figure 3, the extraction of textural characteristics is on the basis of the constant three value mode (SILTP) of part of gray scale in step d
On, by the R to color image, tri- components of G, B calculate 2 norms, obtain norm grayscale image Vnorm, by VnormIt extracts
SILTP feature obtains color texture feature CSILTP.
As shown in figure 4, Fusion Features are by calculating in training sample between identical inter-class variance and inhomogeneity in step e
The ratio of variance calculates the distinguishing ability of every one-dimensional characteristic component, by the selection biggish component of distinguishing ability to test data
Feature selecting is carried out, to play the effect of Fusion Features and dimensionality reduction.
As shown in figure 5, figure (a) is a true capsule endoscope image, location is antrum by juxtapyloricly
Square, there are highlight regions and dark spaces in figure.Figure (b) is after adaptive clear zone dark space detection algorithm of the present invention processing
As a result, the inactive area in figure marks out with black, compare original image, clear zone and dark space are effectively detected out.
As shown in fig. 6, using the classification results of CSILTP textural characteristics of the present invention in accuracy rate, accurate rate and call together
It is all higher than traditional LBP texture and extended version in the rate of returning.
As shown in fig. 7, all having reached relatively steady after the 10th dimensional feature using DPA Feature Fusion Algorithm of the present invention
Fixed classification results, and it is all higher than traditional PCA dimension reduction method in accuracy rate.
Table 1 is application of the heretofore described capsule endoscope image classification algorithms in true picture collection as a result, having recorded
Single category classification performance of each organ.Table 2 is heretofore described capsule endoscope image classification algorithms in true picture collection
Application as a result, have recorded the overall performance of the method, and compared the conventional methods such as HS and HS+LBP.
Table 1
Table 2
As shown in table 1, using endoscopic images classification method of the present invention, esophagus, cardia, stomach, pylorus and 12
Single class classification results of duodenum 12 have all reached 97% or more average accuracy and recall rate.
As shown in table 2, using endoscopic images classification method of the present invention, accuracy rate, accurate rate and recall rate are all
Reach 99% or more, compared to traditional method, performance is improved.
The content that this specification is not described in detail belongs to the prior art well known to professional and technical personnel in the field.
The scheme of the present embodiment is not directly used in human object, and the direct purpose of method is also not the diagnosis of disease,
It is only to realize capsule endoscope image classification based on multi-feature fusion.
Claims (10)
1. a kind of capsule endoscope image classification method based on multi-feature fusion, which is characterized in that the method includes following
Step:
A. it obtains capsule endoscope image set: by extracting capsule endoscope video frame, converting image set for video sequence；
B. image preprocessing: pre-processing capsule endoscope image, generates inactive area mask mask；
C. color of image feature extraction: under HSI color space, by the method for color histogram, the color for extracting image is special
Sign；
D. image texture characteristic extracts: extracting image texture characteristic under color pattern image；
E. multi-features: the color of image and textural characteristics merge and dimensionality reduction；
F. image classification: fused new feature is trained classifier, then using classification to the image newly obtained into
Row classification, marks and which organoid is belonging respectively to.
2. a kind of capsule endoscope image classification method based on multi-feature fusion as described in claim 1, it is characterised in that:
In the step b, image preprocessing includes constructing global mask, marks endoscope non-image areas；Using based on super-pixel
Adaptive dark space clear zone detection method, generates inactive area mask, avoids the feature for extracting irrelevant information.
3. a kind of capsule endoscope image classification method based on multi-feature fusion as claimed in claim 2, it is characterised in that:
The adaptive dark space clear zone detection method based on super-pixel is to obtain global minima reference point by super-pixel segmentation, is utilized
Piecewise function calculates dark space clear zone threshold value, goes out inactive area by Threshold segmentation.
4. a kind of capsule endoscope image classification method based on multi-feature fusion as described in one of claims 1 to 3, special
Sign is: in the step c, color of image feature extraction is that form and aspect and saturation degree two dimension histogram are extracted under HSI color space
Figure is used as color characteristic.
5. a kind of capsule endoscope image classification method based on multi-feature fusion as described in one of claims 1 to 3, special
Sign is: in the step d, image texture characteristic extraction is to be peeped using the constant three value mode CSILTP of part of colored gray scale to interior
Mirror image zooming-out textural characteristics.
6. a kind of capsule endoscope image classification method based on multi-feature fusion as claimed in claim 5, it is characterised in that:
The colour constant three value mode CSILTP of part of gray scale is the R to endoscopic images, and tri- components of G, B calculate 2 norms, then
The constant three value mode of part of gray scale is calculated again obtains CSILTP.
7. a kind of capsule endoscope image classification method based on multi-feature fusion as described in one of claims 1 to 3, special
Sign is: in the step e, multi-features are to be melted using distinguishing ability analysis DPA to multiple images local feature
Conjunction and dimensionality reduction.
8. a kind of capsule endoscope image classification method based on multi-feature fusion as claimed in claim 7, it is characterised in that:
Distinguishing ability analysis DPA is the mirror for calculating the ratio of every one-dimensional characteristic component inter-class variance and variance within clusters as the component
Other ability, according to the distinguishing ability of the every one-dimensional characteristic of training data, the feature for choosing test data carries out Fusion Features and dimensionality reduction.
9. a kind of capsule endoscope image classification method based on multi-feature fusion as described in one of claims 1 to 3, special
Sign is: in the step f, image classification is that capsule endoscope image is divided into corresponding organ using trained classifier
Classification.
10. a kind of capsule endoscope image classification method based on multi-feature fusion as described in claim 1, feature exist
In: in the step f, image classification is carried out using trained classifier, steps are as follows:
1) under the guidance of clinician, capsule endoscope image is classified；
2) endoscope non-image areas is marked with an overall situation mask；
3) inactive area is marked using dark space and clear zone detection algorithm, constructs inactive area mask；
4) color and textural characteristics of all images are extracted at global mask and inactive area mask；
5) Fusion Features are carried out using DPA technology, forms characteristic data set；
6) characteristic data set is randomly divided into 10 parts, takes the training of 9 parts of carry out classifier therein every time, remaining 1 part is used to
The classification performance of testing classification device；
7) by 10 subseries, results are averaged, compares with the classification results of doctor, completes classifier after reaching requirement
Training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811515944.7A CN109635871B (en) | 2018-12-12 | 2018-12-12 | Capsule endoscope image classification method based on multi-feature fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811515944.7A CN109635871B (en) | 2018-12-12 | 2018-12-12 | Capsule endoscope image classification method based on multi-feature fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109635871A true CN109635871A (en) | 2019-04-16 |
| CN109635871B CN109635871B (en) | 2021-06-18 |
Family
ID=66073063
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811515944.7A Active CN109635871B (en) | 2018-12-12 | 2018-12-12 | Capsule endoscope image classification method based on multi-feature fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109635871B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110119710A (en) * | 2019-05-13 | 2019-08-13 | 广州锟元方青医疗科技有限公司 | Cell sorting method, device, computer equipment and storage medium |
| CN110232408A (en) * | 2019-05-30 | 2019-09-13 | 清华-伯克利深圳学院筹备办公室 | A kind of endoscopic image processing method and relevant device |
| CN110367913A (en) * | 2019-07-29 | 2019-10-25 | 杭州电子科技大学 | Wireless capsule endoscope image pylorus and ileocaecal sphineter localization method |
| CN111325709A (en) * | 2019-12-26 | 2020-06-23 | 联博智能科技有限公司 | Wireless capsule endoscope image detection system and detection method |
| CN111767958A (en) * | 2020-07-01 | 2020-10-13 | 武汉楚精灵医疗科技有限公司 | Real-time enteroscopy withdrawal time monitoring method based on random forest algorithm |
| CN112837275A (en) * | 2021-01-14 | 2021-05-25 | 长春大学 | Capsule endoscope image organ classification method, device, equipment and storage medium |
| CN113393449A (en) * | 2021-06-25 | 2021-09-14 | 上海市第一人民医院 | Endoscope video image automatic storage method based on artificial intelligence |
| CN113920042A (en) * | 2021-09-24 | 2022-01-11 | 深圳市资福医疗技术有限公司 | Image processing system and capsule endoscope |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1862106A1 (en) * | 2005-03-22 | 2007-12-05 | Osaka University | Capsule endoscope image display controller |
| CN105469383A (en) * | 2014-12-30 | 2016-04-06 | 北京大学深圳研究生院 | Wireless capsule endoscopy redundant image screening method based on multi-feature fusion |
| CN105512612A (en) * | 2015-11-24 | 2016-04-20 | 中科院合肥技术创新工程院 | SVM-based image classification method for capsule endoscope |
| CN107292347A (en) * | 2017-07-06 | 2017-10-24 | 中冶华天南京电气工程技术有限公司 | A kind of capsule endoscope image-recognizing method |
| CN107730489A (en) * | 2017-10-09 | 2018-02-23 | 杭州电子科技大学 | Wireless capsule endoscope small intestine disease variant computer assisted detection system and detection method |
| CN107767365A (en) * | 2017-09-21 | 2018-03-06 | 华中科技大学鄂州工业技术研究院 | A kind of endoscopic images processing method and system |
| CN108292366A (en) * | 2015-09-10 | 2018-07-17 | 美基蒂克艾尔有限公司 | The system and method that suspect tissue region is detected in endoscopic surgery |
-
2018
- 2018-12-12 CN CN201811515944.7A patent/CN109635871B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1862106A1 (en) * | 2005-03-22 | 2007-12-05 | Osaka University | Capsule endoscope image display controller |
| CN105469383A (en) * | 2014-12-30 | 2016-04-06 | 北京大学深圳研究生院 | Wireless capsule endoscopy redundant image screening method based on multi-feature fusion |
| CN108292366A (en) * | 2015-09-10 | 2018-07-17 | 美基蒂克艾尔有限公司 | The system and method that suspect tissue region is detected in endoscopic surgery |
| CN105512612A (en) * | 2015-11-24 | 2016-04-20 | 中科院合肥技术创新工程院 | SVM-based image classification method for capsule endoscope |
| CN107292347A (en) * | 2017-07-06 | 2017-10-24 | 中冶华天南京电气工程技术有限公司 | A kind of capsule endoscope image-recognizing method |
| CN107767365A (en) * | 2017-09-21 | 2018-03-06 | 华中科技大学鄂州工业技术研究院 | A kind of endoscopic images processing method and system |
| CN107730489A (en) * | 2017-10-09 | 2018-02-23 | 杭州电子科技大学 | Wireless capsule endoscope small intestine disease variant computer assisted detection system and detection method |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110119710A (en) * | 2019-05-13 | 2019-08-13 | 广州锟元方青医疗科技有限公司 | Cell sorting method, device, computer equipment and storage medium |
| CN110232408A (en) * | 2019-05-30 | 2019-09-13 | 清华-伯克利深圳学院筹备办公室 | A kind of endoscopic image processing method and relevant device |
| CN110232408B (en) * | 2019-05-30 | 2021-09-10 | 清华-伯克利深圳学院筹备办公室 | Endoscope image processing method and related equipment |
| CN110367913A (en) * | 2019-07-29 | 2019-10-25 | 杭州电子科技大学 | Wireless capsule endoscope image pylorus and ileocaecal sphineter localization method |
| CN110367913B (en) * | 2019-07-29 | 2021-09-28 | 杭州电子科技大学 | Wireless capsule endoscope image pylorus and ileocecal valve positioning method |
| CN111325709A (en) * | 2019-12-26 | 2020-06-23 | 联博智能科技有限公司 | Wireless capsule endoscope image detection system and detection method |
| CN111767958A (en) * | 2020-07-01 | 2020-10-13 | 武汉楚精灵医疗科技有限公司 | Real-time enteroscopy withdrawal time monitoring method based on random forest algorithm |
| CN112837275A (en) * | 2021-01-14 | 2021-05-25 | 长春大学 | Capsule endoscope image organ classification method, device, equipment and storage medium |
| CN112837275B (en) * | 2021-01-14 | 2023-10-24 | 长春大学 | Capsule endoscope image organ classification method, device, equipment and storage medium |
| CN113393449A (en) * | 2021-06-25 | 2021-09-14 | 上海市第一人民医院 | Endoscope video image automatic storage method based on artificial intelligence |
| CN113920042A (en) * | 2021-09-24 | 2022-01-11 | 深圳市资福医疗技术有限公司 | Image processing system and capsule endoscope |
| CN113920042B (en) * | 2021-09-24 | 2023-04-18 | 深圳市资福医疗技术有限公司 | Image processing system and capsule endoscope |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109635871B (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109635871A (en) | A kind of capsule endoscope image classification method based on multi-feature fusion | |
| JP6657480B2 (en) | Image diagnosis support apparatus, operation method of image diagnosis support apparatus, and image diagnosis support program | |
| CN107920722B (en) | Reconstruction by object detection for images captured from a capsule camera | |
| CN106023151B (en) | Tongue object detection method under a kind of open environment | |
| JP7229996B2 (en) | Speckle contrast analysis using machine learning to visualize flow | |
| CN110189303B (en) | NBI image processing method based on deep learning and image enhancement and application thereof | |
| CN107767365A (en) | A kind of endoscopic images processing method and system | |
| CN105657580B (en) | A kind of capsule endoscope video abstraction generating method | |
| CN110049709A (en) | Image processing apparatus | |
| CN110738655B (en) | Image report generation method, device, terminal and storage medium | |
| CN106535736B (en) | Image processing apparatus, image processing method and image processing program | |
| Cui et al. | Bleeding detection in wireless capsule endoscopy images by support vector classifier | |
| Yuan et al. | Automatic bleeding frame detection in the wireless capsule endoscopy images | |
| CN115049666B (en) | Endoscope virtual biopsy device based on color wavelet covariance depth map model | |
| Mathew et al. | Transform based bleeding detection technique for endoscopic images | |
| KR101063343B1 (en) | Color Information Segmentation Method Using Tongue Image | |
| Mackiewicz et al. | Colour and texture based gastrointestinal tissue discrimination | |
| Ezatian et al. | Image quality enhancement in wireless capsule endoscopy with adaptive fraction gamma transformation and unsharp masking filter | |
| Bogo et al. | Automated detection of new or evolving melanocytic lesions using a 3D body model | |
| CN110969603B (en) | Relative positioning method and device for lesion position and terminal equipment | |
| Ciaccio et al. | Recommendations to quantify villous atrophy in video capsule endoscopy images of celiac disease patients | |
| US20240065540A1 (en) | Apparatus and method for detecting cervical cancer | |
| CN102124471A (en) | Methods for enhancing vascular patterns in cervical imagery | |
| Gadermayr et al. | Getting one step closer to fully automatized celiac disease diagnosis | |
| Li et al. | Computer aided detection of bleeding in capsule endoscopy images |
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 |