CN112022065A - Method and system for quickly positioning time point of capsule entering duodenum - Google Patents
Method and system for quickly positioning time point of capsule entering duodenum Download PDFInfo
- Publication number
- CN112022065A CN112022065A CN202011016314.2A CN202011016314A CN112022065A CN 112022065 A CN112022065 A CN 112022065A CN 202011016314 A CN202011016314 A CN 202011016314A CN 112022065 A CN112022065 A CN 112022065A
- Authority
- CN
- China
- Prior art keywords
- duodenum
- capsule
- images
- image
- stomach
- 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
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
- A61B1/04—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor combined with photographic or television appliances
- A61B1/041—Capsule endoscopes for imaging
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
- A61B1/00002—Operational features of endoscopes
- A61B1/00004—Operational features of endoscopes characterised by electronic signal processing
- A61B1/00009—Operational features of endoscopes characterised by electronic signal processing of image signals during a use of endoscope
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
- A61B1/273—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor for the upper alimentary canal, e.g. oesophagoscopes, gastroscopes
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
- A61B1/273—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor for the upper alimentary canal, e.g. oesophagoscopes, gastroscopes
- A61B1/2736—Gastroscopes
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Theoretical Computer Science (AREA)
- Pathology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Optics & Photonics (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Veterinary Medicine (AREA)
- Radiology & Medical Imaging (AREA)
- Heart & Thoracic Surgery (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Gastroenterology & Hepatology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Signal Processing (AREA)
- Endoscopes (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a method and a system for quickly positioning a time point when a capsule enters duodenum, wherein the method comprises the following steps: s1, acquiring an initial image set; s2, removing repeated images to obtain a preprocessed image set; s3, acquiring a training set; s4, training a LeNet model; s5, classifying the images shot by the target capsule; s6, starting from the first image classified as being positioned in the duodenum, judging whether the subsequent N images are classified as being positioned in the stomach, if so, entering the step S7, and if not, entering the step S8; s7, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to the step S6; and S8, taking the shooting time of the first image classified as being positioned in the duodenum as the time point when the target capsule enters the duodenum, and finishing positioning. The invention solves the problem that manual inquiry is time-consuming and labor-consuming.
Description
Technical Field
The invention relates to the field of computer vision, in particular to a method and a system for quickly positioning a time point when a capsule enters duodenum.
Background
Because of its painlessness and non-invasiveness, capsule endoscopy is one of the best diagnostic tools for examining small intestine diseases in clinic at present, but one of the disadvantages is that after the capsule is swallowed, its movement in the digestive tract completely depends on the peristaltic push speed of the stomach and intestine, when entering the stomach, if the peristaltic evacuation speed of the stomach is slow, the capsule can be retained in the stomach for a long time, the battery energy is too much consumed, when entering the small intestine, the examination of the whole small intestine can not be completed because of the insufficient battery energy. However, there is no good solution to determine the time point when the capsule enters the duodenum.
During the capsule endoscopy, the patient can take tens of thousands of pictures at the speed of 2 pictures/second (OMOM capsule endoscopy, Jinshan company, China), and generally hundreds of pictures can be obtained before entering the duodenum descending segment. However, the existing method for determining that the capsule firstly reaches the duodenum is that a doctor views the image shot by the capsule for manual resolution, and the problem of time and labor waste exists because the number of the images is large.
Disclosure of Invention
Aiming at the defects in the prior art, the method and the system for quickly positioning the time point of the capsule entering the duodenum provided by the invention solve the problem that the time point of the capsule entering the duodenum is time-consuming and labor-consuming through manual inquiry.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
a method for rapidly locating a time point of entry of a capsule into duodenum is provided, comprising the steps of:
s1, labeling the image shot when the capsule is positioned in the stomach and the image shot when the capsule is positioned in the duodenum to be used as an initial image set;
s2, screening the images in the initial image set by adopting a residual method, and removing repeated images to obtain a preprocessed image set;
s3, increasing the images shot by the capsule in the duodenum in the preprocessed image set by adopting an image augmentation method to obtain a training set;
s4, carrying out classification training on the LeNet model by adopting a training set, so that the trained LeNet model has the function of classifying images in the stomach and in the duodenum;
s5, classifying the images shot by the target capsule by adopting the trained LeNet model, and acquiring each image shot by the target capsule as being positioned in the stomach or in the duodenum;
s6, starting from the first image classified as being positioned in the duodenum, judging whether the subsequent N images are classified as being positioned in the stomach, if so, entering the step S7, and if not, entering the step S8;
s7, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to the step S6;
and S8, taking the shooting time of the first image classified as being positioned in the duodenum as the time point when the target capsule enters the duodenum, and finishing positioning.
Further, the specific method of step S1 is:
the image taken with the capsule in the stomach and the image taken with the capsule in the duodenum were labeled with the photographing positions and the labeled images were normalized in size to 3 × 240 × 256, resulting in an initial image set.
Further, the image augmentation method in step S3 includes a RandomFlip method and a RandomCrop method.
Further, the LeNet model in step S4 includes 4 sets of convolution layers and pooling layers alternately connected, and three full-connection layers sequentially connected after the last pooling layer; wherein the first set of convolutional and pooling layers has a size of 240 × 256 × 3, the second set of convolutional and pooling layers has a size of 118 × 126 × 32, the third set of convolutional and pooling layers has a size of 28 × 30 × 64, and the fourth set of convolutional and pooling layers has a size of 6 × 6 × 128; the node of the first fully-connected layer is 512, the node of the second fully-connected layer is 256, and the node of the third fully-connected layer is 2.
Further, the value N is 3 in step S6.
The system for rapidly positioning the time point when the capsule enters the duodenum comprises an initial image set acquisition module, a preprocessing module, a data equalization module, a training module and an identification module;
the initial image set acquisition module is used for labeling an image shot by the capsule in the stomach and an image shot by the capsule in the duodenum to be used as an initial image set;
the preprocessing module is used for screening the images in the initial image set by adopting a residual method, removing repeated images and obtaining a preprocessed image set;
the data equalization module is used for increasing the images shot by the preprocessed image concentration capsules in duodenum by adopting an image augmentation method to obtain a training set;
the training module is used for carrying out classification training on the LeNet model by adopting a training set, so that the trained LeNet model has the function of classifying images in the stomach and in the duodenum;
an identification module to:
p1, classifying the images shot by the target capsule by adopting the trained LeNet model, and acquiring each image shot by the target capsule as being positioned in the stomach or in the duodenum;
p2, starting from the first image classified as being located in the duodenum, determining whether there are N consecutive images classified as being located in the stomach, if so, proceeding to operation P3, otherwise, proceeding to operation P4;
p3, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to step P2;
p4, taking the time of the first image classified as being located in the duodenum as the time point when the target capsule enters the duodenum, and completing the localization.
Further, the LeNet model comprises 4 groups of convolution layers and pooling layers which are alternately connected, and three full-connection layers which are sequentially connected behind the last pooling layer; wherein the first set of convolutional and pooling layers has a size of 240 × 256 × 3, the second set of convolutional and pooling layers has a size of 118 × 126 × 32, the third set of convolutional and pooling layers has a size of 28 × 30 × 64, and the fourth set of convolutional and pooling layers has a size of 6 × 6 × 128; the node of the first fully-connected layer is 512, the node of the second fully-connected layer is 256, and the node of the third fully-connected layer is 2.
Further, in operation P1, the value N is 3.
The invention has the beneficial effects that:
1. according to the invention, the LeNet model is trained after the images shot by the capsule in the stomach and the images shot by the capsule in the duodenum are processed, and the time for the capsule to enter the duodenum is automatically acquired through the trained LeNet model, so that the problems of time and labor waste caused by manual inquiry are solved.
2. The method solves the problem that a large number of repeated images are obtained due to the fact that the capsule moves slowly in a human body through a residual method, and solves the problem that images in the duodenum are few through an image augmentation method, so that training data of the model are more balanced, the model obtained through training is better in recognition effect, and the method is more beneficial to quickly positioning the time point when the capsule enters the duodenum.
3. According to the method, the threshold value is used for classifying the N continuous images to be positioned in the stomach, so that the result of possible recognition errors is corrected, and the recognition accuracy can be improved.
Drawings
FIG. 1 is a schematic flow diagram of the process;
fig. 2 is a schematic structural diagram of the LeNet model.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in FIG. 1, the method for rapidly locating the time point when the capsule enters the duodenum comprises the following steps:
s1, labeling the image shot when the capsule is positioned in the stomach and the image shot when the capsule is positioned in the duodenum to be used as an initial image set;
s2, screening the images in the initial image set by adopting a residual method, and removing repeated images to obtain a preprocessed image set;
s3, increasing the images shot by the capsule in the duodenum in the preprocessed image set by adopting an image augmentation method to obtain a training set;
s4, carrying out classification training on the LeNet model by adopting a training set, so that the trained LeNet model has the function of classifying images in the stomach and in the duodenum;
s5, classifying the images shot by the target capsule by adopting the trained LeNet model, and acquiring each image shot by the target capsule as being positioned in the stomach or in the duodenum;
s6, starting from the first image classified as being positioned in the duodenum, judging whether the subsequent N images are classified as being positioned in the stomach, if so, entering the step S7, and if not, entering the step S8;
s7, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to the step S6;
and S8, taking the shooting time of the first image classified as being positioned in the duodenum as the time point when the target capsule enters the duodenum, and finishing positioning.
The specific method of step S1 is: the image taken with the capsule in the stomach and the image taken with the capsule in the duodenum were labeled with the photographing positions and the labeled images were normalized in size to 3 × 240 × 256, resulting in an initial image set. The image augmentation method in step S3 includes a RandomFlip method and a RandomCrop method.
As shown in fig. 2, the LeNet model in step S4 includes 4 sets of convolutional layers and pooling layers alternately connected, and three fully-connected layers sequentially connected after the last pooling layer; wherein the first set of convolutional and pooling layers has a size of 240 × 256 × 3, the second set of convolutional and pooling layers has a size of 118 × 126 × 32, the third set of convolutional and pooling layers has a size of 28 × 30 × 64, and the fourth set of convolutional and pooling layers has a size of 6 × 6 × 128; the node of the first fully-connected layer is 512, the node of the second fully-connected layer is 256, and the node of the third fully-connected layer is 2.
The system for rapidly positioning the time point when the capsule enters the duodenum comprises an initial image set acquisition module, a preprocessing module, a data balancing module, a training module and an identification module;
the initial image set acquisition module is used for labeling an image shot by the capsule in the stomach and an image shot by the capsule in the duodenum to be used as an initial image set;
the preprocessing module is used for screening the images in the initial image set by adopting a residual method, removing repeated images and obtaining a preprocessed image set;
the data equalization module is used for increasing the images shot by the preprocessed image concentration capsules in duodenum by adopting an image augmentation method to obtain a training set;
the training module is used for carrying out classification training on the LeNet model by adopting a training set, so that the trained LeNet model has the function of classifying images in the stomach and in the duodenum;
an identification module to:
p1, classifying the images shot by the target capsule by adopting the trained LeNet model, and acquiring each image shot by the target capsule as being positioned in the stomach or in the duodenum;
p2, starting from the first image classified as being located in the duodenum, determining whether there are N consecutive images classified as being located in the stomach, if so, proceeding to operation P3, otherwise, proceeding to operation P4;
p3, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to step P2;
p4, taking the time of the first image classified as being located in the duodenum as the time point when the target capsule enters the duodenum, and completing the localization.
The LeNet model comprises 4 groups of convolution layers and pooling layers which are alternately connected, and three full-connection layers which are sequentially connected behind the last pooling layer; wherein the first set of convolutional and pooling layers has a size of 240 × 256 × 3, the second set of convolutional and pooling layers has a size of 118 × 126 × 32, the third set of convolutional and pooling layers has a size of 28 × 30 × 64, and the fourth set of convolutional and pooling layers has a size of 6 × 6 × 128; the node of the first fully-connected layer is 512, the node of the second fully-connected layer is 256, and the node of the third fully-connected layer is 2.
In one embodiment of the present invention, the value N is 3 and the convolution layer kernel is 5 x 5. The training set comprises images of mucosa of stomach, duodenal bulb, duodenal descending segment, and various interference images encountered in the capsule endoscopy process, such as saliva, gastric juice, intestinal fluid, blood fluid, gastric chyme, bile, dark-field images, blurred images and the like.
According to the method, the basis that the capsule does not return to the stomach and the duodenal bulb when entering the duodenal descending segment is taken as the basis, if the trained LeNet model confirms that the capsule endoscope enters the duodenal descending segment, images behind the confirmation point time are images of the duodenal descending segment, if images behind the confirmation point time are judged to be images of mucosa of the stomach and the duodenal bulb, particularly when more than or equal to 3 images appear, the confirmation point can be judged to be wrong, the confirmation point is corrected again, the identification accuracy of the LeNet model can be improved, and the accuracy of the positioning of the capsule entering the duodenal descending segment is improved.
In conclusion, the LeNet model is trained after the images shot by the capsule in the stomach and the images shot by the capsule in the duodenum are processed, and the time for the capsule to enter the duodenum is automatically acquired through the trained LeNet model, so that the problems of time and labor waste caused by manual inquiry are solved.
Claims (8)
1. A method for rapidly locating the point in time at which a capsule enters the duodenum, comprising the steps of:
s1, labeling the image shot when the capsule is positioned in the stomach and the image shot when the capsule is positioned in the duodenum to be used as an initial image set;
s2, screening the images in the initial image set by adopting a residual method, and removing repeated images to obtain a preprocessed image set;
s3, increasing the images shot by the capsule in the duodenum in the preprocessed image set by adopting an image augmentation method to obtain a training set;
s4, carrying out classification training on the LeNet model by adopting a training set, so that the trained LeNet model has the function of classifying images in the stomach and in the duodenum;
s5, classifying the images shot by the target capsule by adopting the trained LeNet model, and acquiring each image shot by the target capsule as being positioned in the stomach or in the duodenum;
s6, starting from the first image classified as being positioned in the duodenum, judging whether the subsequent N images are classified as being positioned in the stomach, if so, entering the step S7, and if not, entering the step S8;
s7, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to the step S6;
and S8, taking the shooting time of the first image classified as being positioned in the duodenum as the time point when the target capsule enters the duodenum, and finishing positioning.
2. The method for rapidly locating the entering time point of the capsule into the duodenum according to claim 1, wherein the specific method of step S1 is:
the image taken with the capsule in the stomach and the image taken with the capsule in the duodenum were labeled with the photographing positions and the labeled images were normalized in size to 3 × 240 × 256, resulting in an initial image set.
3. The method for rapidly locating the entry time point of a capsule into the duodenum according to claim 1, wherein the image augmentation method in step S3 comprises a RandomFlip method and a RandomCrop method.
4. The method for rapidly locating a duodenal time point of entry of a capsule as recited in claim 1, wherein the LeNet model in step S4 comprises 4 sets of alternately connected convolutional and pooling layers, and three fully-connected layers connected sequentially after the last pooling layer; wherein the first set of convolutional and pooling layers has a size of 240 × 256 × 3, the second set of convolutional and pooling layers has a size of 118 × 126 × 32, the third set of convolutional and pooling layers has a size of 28 × 30 × 64, and the fourth set of convolutional and pooling layers has a size of 6 × 6 × 128; the node of the first fully-connected layer is 512, the node of the second fully-connected layer is 256, and the node of the third fully-connected layer is 2.
5. The method for rapidly localizing the entry point in time of a capsule into the duodenum according to claim 1, wherein the value N in step S6 is 3.
6. A system for rapidly positioning a time point when a capsule enters duodenum is characterized by comprising an initial image set acquisition module, a preprocessing module, a data balancing module, a training module and an identification module;
the initial image set acquisition module is used for labeling an image shot by the capsule in the stomach and an image shot by the capsule in the duodenum to be used as an initial image set;
the preprocessing module is used for screening the images in the initial image set by adopting a residual method, removing repeated images and obtaining a preprocessed image set;
the data equalization module is used for increasing the images shot by the preprocessed image concentration capsules in duodenum by adopting an image augmentation method to obtain a training set;
the training module is used for carrying out classification training on the LeNet model by adopting a training set, so that the trained LeNet model has the function of classifying images in the stomach and in the duodenum;
the identification module is used for performing the following operations:
p1, classifying the images shot by the target capsule by adopting the trained LeNet model, and acquiring each image shot by the target capsule as being positioned in the stomach or in the duodenum;
p2, starting from the first image classified as being located in the duodenum, determining whether there are N consecutive images classified as being located in the stomach, if so, proceeding to operation P3, otherwise, proceeding to operation P4;
p3, removing the images classified as being in the duodenum before the consecutive N or more images classified as being in the stomach, and returning to step P2;
p4, taking the time of the first image classified as being located in the duodenum as the time point when the target capsule enters the duodenum, and completing the localization.
7. The system for rapidly locating a point in time at which a capsule enters the duodenum according to claim 6, wherein the LeNet model comprises 4 sets of alternately connected convolutional and pooling layers, and three fully connected layers connected sequentially after the last pooling layer; wherein the first set of convolutional and pooling layers has a size of 240 × 256 × 3, the second set of convolutional and pooling layers has a size of 118 × 126 × 32, the third set of convolutional and pooling layers has a size of 28 × 30 × 64, and the fourth set of convolutional and pooling layers has a size of 6 × 6 × 128; the node of the first fully-connected layer is 512, the node of the second fully-connected layer is 256, and the node of the third fully-connected layer is 2.
8. The system for rapidly positioning a capsule entry time point into the duodenum according to claim 6, wherein the value N in operation P1 is 3.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011016314.2A CN112022065A (en) | 2020-09-24 | 2020-09-24 | Method and system for quickly positioning time point of capsule entering duodenum |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011016314.2A CN112022065A (en) | 2020-09-24 | 2020-09-24 | Method and system for quickly positioning time point of capsule entering duodenum |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112022065A true CN112022065A (en) | 2020-12-04 |
Family
ID=73573832
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011016314.2A Pending CN112022065A (en) | 2020-09-24 | 2020-09-24 | Method and system for quickly positioning time point of capsule entering duodenum |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112022065A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101365136A (en) * | 2008-09-09 | 2009-02-11 | 深圳市同洲电子股份有限公司 | Method and apparatus for intra-frame prediction |
CN102999912A (en) * | 2012-11-27 | 2013-03-27 | 宁波大学 | Three-dimensional image quality objective evaluation method based on distorted images |
WO2018112255A1 (en) * | 2016-12-14 | 2018-06-21 | Progenity Inc. | Treatment of a disease of the gastrointestinal tract with an immunosuppressant |
CN108960198A (en) * | 2018-07-28 | 2018-12-07 | 天津大学 | A kind of road traffic sign detection and recognition methods based on residual error SSD model |
CN109583325A (en) * | 2018-11-12 | 2019-04-05 | 平安科技(深圳)有限公司 | Face samples pictures mask method, device, computer equipment and storage medium |
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 |
-
2020
- 2020-09-24 CN CN202011016314.2A patent/CN112022065A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101365136A (en) * | 2008-09-09 | 2009-02-11 | 深圳市同洲电子股份有限公司 | Method and apparatus for intra-frame prediction |
CN102999912A (en) * | 2012-11-27 | 2013-03-27 | 宁波大学 | Three-dimensional image quality objective evaluation method based on distorted images |
WO2018112255A1 (en) * | 2016-12-14 | 2018-06-21 | Progenity Inc. | Treatment of a disease of the gastrointestinal tract with an immunosuppressant |
CN108960198A (en) * | 2018-07-28 | 2018-12-07 | 天津大学 | A kind of road traffic sign detection and recognition methods based on residual error SSD model |
CN109583325A (en) * | 2018-11-12 | 2019-04-05 | 平安科技(深圳)有限公司 | Face samples pictures mask method, device, computer equipment and storage medium |
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 |
Non-Patent Citations (1)
Title |
---|
张向荣,冯婕,刘芳,焦李成: "《人工智能前沿技术丛书 模式识别》", 30 September 2019 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021036863A1 (en) | Deep learning-based diagnosis assistance system for early digestive tract cancer and examination apparatus | |
CN109146884B (en) | Endoscopic examination monitoring method and device | |
CN106934799B (en) | Capsule endoscope visual aids diagosis system and method | |
WO2020071677A1 (en) | Method and apparatus for diagnosing gastric lesions by using deep learning on gastroscopy images | |
CN110367913B (en) | Wireless capsule endoscope image pylorus and ileocecal valve positioning method | |
CN107767365A (en) | A kind of endoscopic images processing method and system | |
CN111986196B (en) | Automatic monitoring method and system for retention of gastrointestinal capsule endoscope | |
CN109584229A (en) | A kind of real-time assistant diagnosis system of Endoscopic retrograde cholangio-pancreatiography art and method | |
CN111340094A (en) | Capsule endoscope image auxiliary classification system and classification method based on deep learning | |
CN115564712B (en) | Capsule endoscope video image redundant frame removing method based on twin network | |
CN114399465B (en) | Benign and malignant ulcer identification method and system | |
CN114259197B (en) | Capsule endoscope quality control method and system | |
CN111839428A (en) | Method for improving detection rate of colonoscope adenomatous polyps based on deep learning | |
JP2006223376A (en) | Medical image processing apparatus | |
AU2020101450A4 (en) | Retinal vascular disease detection from retinal fundus images using machine learning | |
CN114782760A (en) | Stomach disease picture classification system based on multitask learning | |
Xu et al. | Upper gastrointestinal anatomy detection with multi‐task convolutional neural networks | |
CN111493805A (en) | State detection device, method, system and readable storage medium | |
CN111768389A (en) | Automatic timing method for digestive tract operation based on convolutional neural network and random forest | |
CN108937871A (en) | A kind of alimentary canal micro-optics coherence tomography image analysis system and method | |
CN114359131A (en) | Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof | |
CN112022065A (en) | Method and system for quickly positioning time point of capsule entering duodenum | |
CN111784669A (en) | Capsule endoscopy image multi-focus detection method | |
Patel et al. | Deep learning in gastrointestinal endoscopy | |
CN111126474B (en) | Confocal laser micro-endoscope digestive tract image identification method and system |
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 |
Application publication date: 20201204 |
|
RJ01 | Rejection of invention patent application after publication |