CN114359131A - Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof - Google Patents
Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof Download PDFInfo
- Publication number
- CN114359131A CN114359131A CN202111338011.7A CN202111338011A CN114359131A CN 114359131 A CN114359131 A CN 114359131A CN 202111338011 A CN202111338011 A CN 202111338011A CN 114359131 A CN114359131 A CN 114359131A
- Authority
- CN
- China
- Prior art keywords
- helicobacter pylori
- module
- video
- image
- anatomical
- 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
Abstract
The invention belongs to the field of medical data mining, and particularly relates to a helicobacter pylori stomach video full-automatic intelligent analysis system and a marking method thereof. According to the invention, a doctor does not need to step down a pedal to input a specific picture to identify helicobacter pylori infection in the test process, manual intervention is not needed in the whole process, and full automation and high intelligence can be achieved.
Description
Technical Field
The invention belongs to the field of medical data mining, and particularly relates to a helicobacter pylori stomach video full-automatic intelligent analysis system and a marking method thereof.
Background
Helicobacter pylori or helicobacter pylori, the english name helicobacter pylori. Helicobacter pylori causes progressive damage to the gastric mucosa, which has a pathogenic role in several important diseases, including duodenal ulcers, gastric adenocarcinoma, and gastric mucosa-associated lymphoid tissue (MALT) lymphoma. Almost all gastric cancer patients have a history of helicobacter pylori infection, and timely discovery and eradication of helicobacter pylori infection can effectively reduce the risk of gastric cancer. Furthermore, according to an international risk assessment analysis, it has been shown that even after eradication of H.pylori, the patient is at a higher risk of developing gastric cancer than those who have never developed H.pylori, so that regular endoscopy is recommended after H.pylori eradication surgery. Therefore, it is becoming increasingly important to examine the condition of helicobacter pylori infection endoscopically.
An experienced digestive tract endoscopy physician can judge the helicobacter pylori infection condition based on the endoscope image characteristics. According to Kyoto global helicobacter pylori gastritis consensus guidelines, the signs of helicobacter pylori infection include diffuse redness, mucosal swelling, plica swollen snake shape, white turbid mucus, punctate redness, chicken skin-like gastritis, etc., and signs of helicobacter pylori non-infection are shown by regular arrangement of aggregated venules, and after eradication, multiple map redness, atrophy, etc. are shown. The degree of infection varies, as do the signs under the endoscopic image. If the judgment is carried out only by naked eyes, the requirement on the experience level of a doctor is high, and the doctor in the junior can not identify the doctor correctly.
Patent CN112651375A discloses a helicobacter pylori stomach image recognition classification system based on a deep learning model, which comprises an image acquisition device and a calculation host, wherein the calculation host comprises an image preprocessing module and a deep learning model module, and the image acquisition module, the image preprocessing module and the deep learning model module are connected in sequence; the image acquisition device is used for acquiring image data of a to-be-detected position of the stomach; the image preprocessing module is used for preprocessing an image; the deep learning model module is used for extracting the features of the images and then classifying the images according to the extracted features. The system can realize two functions: 1. features of the endoscope image are extracted through a deep learning model, the category of the image is judged, and the image is divided into three categories of infection, cured after infection and non-infected. 2. The characteristics of the endoscope image are extracted through the deep learning model, the stomach position collected by the image is identified, and the image is classified according to the identified stomach position so as to detect whether the position is missed and avoid the detection error of the endoscope.
However, the recognition classification system is trained and tested based on static pictures, the robustness of video data is poor, the physician is required to select a specific video frame with infection signs in each gastroscopy video during the construction of a training set, the time and the labor are consumed, and the physician is required to input a specific picture into the system for analysis through a foot switch during the test, so the workload of normal endoscopy is increased. Therefore, it is necessary to develop a video-training-based helicobacter pylori gastric video full-automatic intelligent auxiliary analysis system which does not need intervention of a doctor during testing, can quickly judge the helicobacter pylori infection condition in real time, reduces biopsy requirements, and provides better auxiliary diagnosis and reference value for endoscopy of a primary doctor.
Disclosure of Invention
The invention aims to solve the problems of high data annotation cost and low inference automation degree in the existing technology for identifying helicobacter pylori under an endoscope, and provides a full-automatic intelligent analysis system for helicobacter pylori stomach video.
Another objective of the invention is to provide a full-automatic real-time helicobacter pylori marking method.
The technical scheme adopted by the invention is as follows:
a helicobacter pylori stomach video full-automatic intelligent analysis system includes:
an image acquisition module to acquire an input stomach conventional white light endoscopic video stream from an endoscopic image system;
the image preprocessing module is used for preprocessing images of the video stream;
the effective frame module is used for classifying and screening images of the video stream to obtain an effective frame with clear visual field;
an anatomical location locating module to identify a particular location within the stomach at which the endoscope is currently located;
the deep learning module is used for predicting helicobacter pylori infection probability of pictures at different anatomical positions, then generating a position prediction result at each anatomical position through a multi-example voting mechanism, and finally obtaining real-time helicobacter pylori infection probability of the gastroscope through voting again by combining the prediction results of the plurality of anatomical positions; and
a marker display module to display anatomical location information and helicobacter pylori video infection probability.
Further, the preprocessing of the image preprocessing module is to process the image into an input form required by the deep learning module, and the processing includes one or more of image normalization, invalid pixel clipping and image scaling.
Further, the operation of the active frame module includes: the video frames are classified into 4 types of in-vitro frames, narrow-band endoscopy frames, invalid examination frames and valid examination frames by utilizing a deep learning technology, so that interference images are filtered out, and endoscopy images with clear visual fields are reserved.
Further, the anatomical location module uses deep learning techniques to identify a particular location within the stomach where the endoscope is currently located, including one or more of the fundus, body, angle, antrum 4.
Further, the deep learning module is used for helicobacter pylori infection prediction, the training process of the deep learning module is based on a multi-example learning mechanism and video-level labels, the deep learning module comprises a model training unit and a model testing unit, the model training unit is used for building a deep learning model, and the model testing unit is used for predicting the helicobacter pylori infection rate of gastroscope video in real time.
Further, the model training unit proceeds by:
s200, collecting a plurality of gastroscopic videos to correspond to each examination14C, using the urea breath test and pathological tissue section examination results as a video grade helicobacter pylori infection label;
s201, screening a single-frame image with clear view in a video through an effective frame module, and then performing data amplification;
s202, dividing the single-frame image into different anatomical positions through an anatomical position positioning module;
s203, based on a multi-example learning mechanism, a doctor does not need to screen specific infection frame pictures, and only utilizes video-level helicobacter pylori infection labels to respectively train a deep learning model for predicting helicobacter pylori infection probability of a single-frame image at each anatomical position.
Further, the model test unit is performed by:
s300, screening and reserving pictures with clear visual fields by using an effective frame module for the input video stream;
s301, obtaining the anatomical position category of each frame of image by using an anatomical position positioning module;
s302, obtaining the helicobacter pylori infection probability of each frame of picture by using a helicobacter pylori infection prediction model of the corresponding anatomical position, then generating a position prediction result at each anatomical position through a multi-example voting mechanism, and finally voting the infection probabilities of all the anatomical positions again to obtain the helicobacter pylori infection probability of the patient level.
Further, the operation of the indicia display module includes: and refreshing and displaying the infection probability of the current frame, the whole infection probability of the current video, the maximum value of the infection probability under each anatomical position and the corresponding picture according to the output information of the deep learning module.
Compared with the prior art, the full-automatic intelligent stomach video analysis system for helicobacter pylori provided by the invention has the following beneficial effects:
1) the helicobacter pylori infection condition under the gastroscope is judged in real time in a full-automatic manner through a deep learning module;
2) the method comprises the steps that a prediction model is trained on the basis of video level labels, a multi-example learning mechanism is adopted, the training data of a deep learning module is constructed without manually screening and labeling the helicobacter pylori infection labels of single-frame pictures by doctors, and the video level infection labels of all inspection videos are obtained only according to pathology and expiration results;
3) according to the invention, a doctor does not need to step down a pedal to input a specific picture to identify helicobacter pylori infection in the test process, and the whole process does not need manual intervention, so that full automation and high intellectualization can be achieved;
4) the mark display module refreshes and displays the anatomical position information and the helicobacter pylori video infection probability on the display, refreshes and prompts the non-checked anatomical position according to the checked position queue, can assist and relieve the high strength of doctors, can help to read the film for a long time, avoids the error of subjective judgment of the doctors caused by the working strength and the working time, reduces the workload of the doctors, and improves the efficiency of medical diagnosis work.
The full-automatic real-time helicobacter pylori marking method adopts the intelligent analysis system, and comprises the following specific steps:
s100, obtaining a conventional white light endoscope video stream of the stomach of a detected person through an image acquisition module and an endoscope image system device, and refreshing and displaying the video stream on a mark display module;
s101, preprocessing a video stream through an image preprocessing module;
s102, classifying and screening the video stream through an effective frame module to obtain an effective frame image;
s103, displaying the type of the anatomical position of the current frame by the anatomical position positioning module, refreshing and displaying the coverage rate of the current checked position, and prompting the remaining unchecked positions;
s104, predicting the helicobacter pylori infection probability of all effective frame images through a deep learning module;
and S105, refreshing and displaying the infection probability of the current frame, the whole infection probability of the current video, the maximum value of the infection probability under each anatomical position and the corresponding picture according to the prediction result of the deep learning module.
The intelligent analysis system is adopted in the full-automatic real-time helicobacter pylori marking method provided by the invention, and the beneficial effects of the intelligent analysis system and the like are explained in detail in the characters, so that the detailed description is omitted.
Drawings
FIG. 1 is a schematic diagram of a helicobacter pylori stomach video full-automatic intelligent analysis system provided in embodiment 1 of the present invention;
FIG. 2 is a structural diagram of a model training unit of a helicobacter pylori stomach video full-automatic intelligent analysis system according to embodiment 1 of the present invention;
FIG. 3 is a structural diagram of a model test unit of a helicobacter pylori stomach video full-automatic intelligent analysis system provided in embodiment 1 of the present invention;
FIG. 4 is a schematic diagram illustrating a display result of a marker display module in a helicobacter pylori stomach video full-automatic intelligent analysis system according to embodiment 1 of the present invention;
FIG. 5 is a flowchart of a method for automatically marking helicobacter pylori in real time according to embodiment 2 of the present invention.
Detailed Description
In the description of the present invention, it is to be understood that the terms "one end", "the other end", "outside", "upper", "inside", "horizontal", "coaxial", "central", "end", "length", "outer end", and the like, indicate orientations or positional relationships based on those shown in the drawings, and are used only for convenience in describing the present invention and for simplicity in description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed in a particular orientation, and be operated, and thus, should not be construed as limiting the present invention.
The invention will be further explained with reference to the drawings.
Example 1
Fig. 1 shows a schematic diagram of a helicobacter pylori stomach video full-automatic intelligent analysis system of the present embodiment, which runs on a computer system and includes an image acquisition module, an image preprocessing module, an effective frame module, an anatomical position locating module, a deep learning module and a mark display module, which are connected in sequence.
The image acquisition module acquires an input stomach conventional white light endoscope video stream in real time through an endoscope image system.
The image preprocessing module is used for processing images of the video stream into an input form required by the deep learning model, and the image enhancement processing comprises the following steps: image normalization, invalid pixel clipping, and image scaled pixel clipping.
The effective frame module classifies video frames into 4 types of in-vitro frames, narrow-band endoscope frames, invalid examination frames and effective examination frames by utilizing a deep learning technology, so that non-white light images, interference images such as motion blur, instrument operation and blood stain are filtered, and endoscope examination images with clear vision are reserved.
The dissection position locating module is used for obtaining stomach position information of the current frame, and the stomach position information comprises 4 main stomach dissection positions of a stomach fundus, a stomach body, a stomach antrum and a pylorus.
The deep learning module is used for predicting the helicobacter pylori infection condition of the endoscope video in real time and comprises a model training unit and a model testing unit.
Continuing with FIG. 2, an illustration of the model training unit: because helicobacter pylori infection may occur at each anatomical position of the stomach, and the anatomical position structure of the stomach and the helicobacter pylori infection symptoms have characteristics, the anatomical positions are directly placed together to train a deep learning model, and the convergence direction of the network is greatly interfered by the structural information of the anatomical positions, so that the helicobacter pylori infection information cannot be effectively learned, therefore, in the embodiment, in order to decouple the anatomical position information from the helicobacter pylori infection information, respective multi-instance-based learning helicobacter pylori infection prediction models are trained respectively for different anatomical positions, and the specific process is as follows:
s200, collecting a plurality of gastroscopic videos to correspond to each examination14C, using the urea breath test and pathological tissue section examination results as a video grade helicobacter pylori infection label;
s201, screening clear images in each video through the effective frame module of the embodiment, and then performing data amplification through translation transformation, mirror image turning or random cutting to expand a data set;
s202, dividing the image into different anatomical positions including four main anatomical positions of a fundus, a body, an angle and a antrum by using the anatomical position locating module of the embodiment;
s203, based on a weak supervision mechanism of multi-example learning, for the helicobacter pylori infection video, in all frame images included in the video, the image with the front prediction probability should have morphological characteristics related to the helicobacter pylori infection, such as mucus, red swelling and the like, and the image with the rear prediction probability does not necessarily have infection characteristics; on the premise that all frame images contained in the negative video do not have infection characteristics, namely the images with the highest prediction probability sequence do not contain helicobacter pylori infection characteristics, the invention adopts a multi-example learning mechanism, and in each iteration of network training, the invention uses a two-step training mode:
Step1：
for each anatomical position, calculating the infection probability of all video frames of each patient by using a deep convolutional network model, then sequencing the infection probabilities, and selecting a picture of the infection probability top-10 of each anatomical position of each patient;
Step2:
and taking the label corresponding to the video as the pseudo label of the selected top-10 picture, calculating the loss of each picture, and performing back propagation to update the weight parameter of the convolution network model.
The steps are circularly carried out until the network converges.
The convolution network model adopted by the invention takes ResNet18 as a basic network and carries out weight initialization based on the pre-training model of Image-Net.
Generally, a doctor is required to invest a large amount of time and energy to select pictures with helicobacter pylori infection signs in the construction of a training data set of a helicobacter pylori recognition system under an endoscope, and the embodiment is based on a multi-example learning mechanism, only results obtained through expiration and pathological examination reports are used as a label training deep learning model of each video, no marking work is required to be carried out by the doctor, no marking historical data of passing hospitals can be fully utilized, and the method has the advantage of large-scale popularization.
In addition, in order to relieve the interference of the unbalanced variety, in each iteration of the model, the invention respectively reads a plurality of pictures from the positive picture data loader and the negative picture loader to form batch training data with balanced positive and negative proportions, so that the model is well balanced in the aspects of prediction specificity and sensitivity and has better robustness.
Continuing with FIG. 3, an illustration of the model prediction unit: is responsible for real-time and full-automatic prediction of helicobacter pylori infection in gastroscopy. Generally, a helicobacter pylori recognition system under an endoscope usually needs a doctor to step on a specific frame through operation of a foot switch and the like and input the frame into an analysis system for helicobacter pylori infection recognition, while the embodiment fully adopts a multi-example learning mechanism based on weak supervision, firstly voting a video frame result of a prediction probability top-10 under each anatomical part to obtain an anatomical position level prediction probability, then voting the infection probability of each anatomical position again to obtain the video level prediction probability, and does not need the doctor to input a specific frame, so that the helicobacter pylori recognition system has the characteristic of full automation, and the main flow is as follows:
s300, screening and reserving pictures with clear visual fields by using an effective frame module for the input video stream;
s301, determining the anatomical position category of each frame of image through an anatomical position positioning module;
s302, obtaining the helicobacter pylori infection probability of each frame of picture by using a helicobacter pylori infection prediction model corresponding to the anatomical position, and respectively carrying out accumulated voting on the pictures of the infection probability top-10 at each anatomical position to obtain the accumulated average infection probability at the current moment of each anatomical position;
s303, voting the infection probabilities of all anatomical positions to obtain the cumulative average infection probability of the helicobacter pylori at the current moment of the gastroscopy video.
This example has the following advantages compared with a similar endoscopic helicobacter pylori infection analysis system
1. The helicobacter pylori infection prediction model in the embodiment is based on a deep learning model of a multi-example learning mechanism, a training set of the model does not need a doctor to manually select and label a single-frame static picture for construction, only helicobacter pylori infection results of a patient need to be obtained from pathology and exhalation reports as corresponding video-level labels, and labeling workload of the doctor is greatly reduced, so that the model is different from any previous helicobacter pylori infection prediction model based on an endoscope, belongs to weak supervised learning, and has good adaptability and popularization for a large batch of video data sets;
2. the multi-stage voting mechanism is divided into three levels for predicting the helicobacter pylori infection, namely, a picture level, an anatomical position level and a patient level, has very good stability and robustness, and can effectively avoid the interference of outliers and extreme values;
3. the whole analysis and prediction process is fully automatic, a doctor does not need to select a specific picture (similar systems often need to step on a foot switch by the doctor to input a designated picture for model analysis) as input, and the helicobacter pylori infection prediction can be carried out under the condition of not increasing the examination workload of an endoscope doctor aiming at the full video stream analysis.
Example 2
Referring to fig. 5, a method for fully automatically marking helicobacter pylori in real time, which employs the intelligent analysis system, includes the following steps:
s100, obtaining a conventional white light endoscope video stream of the stomach of a detected person through an image acquisition module and an endoscope image system device, and refreshing and displaying the video stream on a mark display module;
s101, preprocessing a video stream through an image preprocessing module;
s102, classifying and screening the video stream through an effective frame module to obtain an effective frame image;
s103, displaying the type of the anatomical position of the current frame by the anatomical position positioning module, refreshing and displaying the coverage rate of the current checked position, and prompting the remaining unchecked positions;
s104, predicting the helicobacter pylori infection probability of all effective frame images through a deep learning module;
and S105, refreshing and displaying the infection probability of the current frame, the whole infection probability of the current video, the maximum value of the infection probability under each anatomical position and the corresponding picture according to the prediction result of the deep learning module.
Further, in S100, the obtained conventional white light endoscope video stream of the stomach of the subject is refreshed on the left half part of the screen for the observation of the doctor;
further, in S105, the helicobacter pylori infection probability of the current image is obtained and added to the probability value sequence of the corresponding anatomical position, top-10 helicobacter pylori infection probability of the current anatomical position is calculated and updated, the infection probability of the anatomical position is obtained by voting, then the helicobacter pylori infection probability of the checked position is voted again to obtain the cumulative infection average probability of the helicobacter pylori of the current video, and simultaneously the frame image with the maximum infection probability at each anatomical position is updated, and finally, the information and the image are updated and displayed at the lower part of the right side of the screen to remind a doctor of the currently checked helicobacter pylori infection condition.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (9)
1. The utility model provides a helicobacter pylori stomach video full-automatic intelligent analysis system which characterized in that includes:
an image acquisition module to acquire an input stomach conventional white light endoscopic video stream from an endoscopic image system;
the image preprocessing module is used for preprocessing images of the video stream;
the effective frame module is used for classifying and screening video stream images to acquire effective frames with clear visual fields;
an anatomical location locating module to identify a particular location within the stomach at which the endoscope is currently located;
the deep learning module is used for predicting helicobacter pylori infection probability of pictures at different anatomical positions, then generating a position prediction result at each anatomical position through a multi-example voting mechanism, and finally obtaining real-time helicobacter pylori infection probability of the gastroscope through voting again by combining the prediction results of the plurality of anatomical positions; and
a marker display module to display anatomical location information and helicobacter pylori video infection probability.
2. The system of claim 1, wherein the pre-processing of the image pre-processing module is processing the image into an input form required by the deep learning module, and the processing includes one or more of image normalization, invalid pixel clipping, and image scaling.
3. The system of claim 1, wherein the operations of the valid frame module comprise: the video frames are classified into 4 types of in-vitro frames, narrow-band endoscopy frames, invalid examination frames and valid examination frames by utilizing a deep learning technology, so that interference images are filtered out, and endoscopy images with clear visual fields are reserved.
4. The system of claim 1, wherein the anatomical location module uses a deep learning technique to identify a specific location in the stomach where the endoscope is currently located.
5. The helicobacter pylori stomach video full-automatic intelligent analysis system according to any one of claims 1 to 4, wherein the deep learning module is used for helicobacter pylori infection prediction, the training process is based on a multi-example learning mechanism and a video level label, and the deep learning module comprises a model training unit and a model testing unit, the model training unit is used for establishing a deep learning model, and the model testing unit is used for predicting the helicobacter pylori infection rate.
6. The helicobacter pylori stomach video full-automatic intelligent analysis system according to claim 5, wherein the model training unit is used for performing the following steps:
s200, collecting a plurality of gastroscopic videos to correspond to each examination14C, using the urea breath test and pathological tissue section examination results as a video grade helicobacter pylori infection label;
s201, screening a single-frame image with clear view in a video through an effective frame module, and then performing data amplification;
s202, dividing the single-frame image into different anatomical positions through an anatomical position positioning module;
s203, respectively training the deep learning model at each anatomical position based on a multi-example learning mechanism.
7. The helicobacter pylori stomach video full-automatic intelligent analysis system according to claim 6, wherein the model test unit is used for performing the following steps:
s300, filtering an image with unclear visual field by using an effective frame module for an input video stream, and screening and retaining a picture with clear visual field;
s301, obtaining the anatomical position category of each frame of image by using an anatomical position positioning module;
s302, obtaining the helicobacter pylori infection probability of each frame of picture by using a helicobacter pylori infection prediction model of the corresponding anatomical position, then generating a position prediction result at each anatomical position through a multi-example voting mechanism, and finally voting again on the infection probabilities of all the anatomical positions to obtain the helicobacter pylori infection probability of the patient level.
8. The system for fully-automatic intelligent video analysis of helicobacter pylori stomach according to any one of claims 1 to 4, wherein the operation of the marker display module comprises: and refreshing and displaying the infection probability of the current frame, the whole infection probability of the current video, the maximum value of the infection probability under each anatomical position and the corresponding picture according to the output information of the deep learning module.
9. A full-automatic real-time helicobacter pylori marking method, which is characterized in that the intelligent analysis system as claimed in any one of claims 1 to 8 is adopted, and the method comprises the following specific steps:
s100, obtaining a conventional white light endoscope video stream of the stomach of a detected person through an image acquisition module and an endoscope image system device, and refreshing and displaying the video stream on a mark display module;
s101, preprocessing a video stream through an image preprocessing module;
s102, classifying and screening the video stream through an effective frame module to obtain an effective frame image;
s103, displaying the type of the anatomical position of the current frame by the anatomical position positioning module, refreshing and displaying the coverage rate of the current checked position, and prompting the remaining unchecked positions;
s104, predicting the helicobacter pylori infection probability of all effective frame images through a deep learning module;
and S105, refreshing and displaying the infection probability of the current frame, the whole infection probability of the current video, the maximum value of the infection probability under each anatomical position and the corresponding picture according to the prediction result of the deep learning module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111338011.7A CN114359131A (en) | 2021-11-12 | 2021-11-12 | Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111338011.7A CN114359131A (en) | 2021-11-12 | 2021-11-12 | Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114359131A true CN114359131A (en) | 2022-04-15 |
Family
ID=81096046
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111338011.7A Pending CN114359131A (en) | 2021-11-12 | 2021-11-12 | Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114359131A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114678121A (en) * | 2022-05-30 | 2022-06-28 | 上海芯超生物科技有限公司 | HP spherical deformation diagnosis model and construction method thereof |
| CN117671573A (en) * | 2024-02-01 | 2024-03-08 | 苏州凌影云诺医疗科技有限公司 | Helicobacter pylori infection state identification method and device based on gastroscope image |
-
2021
- 2021-11-12 CN CN202111338011.7A patent/CN114359131A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114678121A (en) * | 2022-05-30 | 2022-06-28 | 上海芯超生物科技有限公司 | HP spherical deformation diagnosis model and construction method thereof |
| CN114678121B (en) * | 2022-05-30 | 2022-09-09 | 上海芯超生物科技有限公司 | Method and system for constructing HP spherical deformation diagnosis model |
| CN117671573A (en) * | 2024-02-01 | 2024-03-08 | 苏州凌影云诺医疗科技有限公司 | Helicobacter pylori infection state identification method and device based on gastroscope image |
| CN117671573B (en) * | 2024-02-01 | 2024-04-12 | 苏州凌影云诺医疗科技有限公司 | Helicobacter pylori infection state identification method and device based on gastroscope image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7404509B2 (en) | Gastrointestinal early cancer diagnosis support system and testing device based on deep learning | |
| CN109146884B (en) | Endoscopic examination monitoring method and device | |
| US7319781B2 (en) | Method and system for multiple passes diagnostic alignment for in vivo images | |
| WO2020215593A1 (en) | Artificial intelligence-based automatic evaluation method and system on quality check of gastrointestinal endoscopy | |
| US20220172828A1 (en) | Endoscopic image display method, apparatus, computer device, and storage medium | |
| CN114359131A (en) | Helicobacter pylori stomach video full-automatic intelligent analysis system and marking method thereof | |
| US20050075537A1 (en) | Method and system for real-time automatic abnormality detection for in vivo images | |
| US20050196023A1 (en) | Method for real-time remote diagnosis of in vivo images | |
| CN113610847B (en) | Method and system for evaluating stomach markers in white light mode | |
| CN109447985A (en) | Colonoscopic images analysis method, device and readable storage medium storing program for executing | |
| CN111091559A (en) | Depth learning-based auxiliary diagnosis system for small intestine sub-scope lymphoma | |
| CN110867233B (en) | System and method for generating electronic laryngoscope medical test reports | |
| CN112801958A (en) | Ultrasonic endoscope, artificial intelligence auxiliary identification method, system, terminal and medium | |
| CN109859827A (en) | Gastrointestinal Endoscopes operation horizontal points-scoring system and method in real time | |
| WO2019098415A1 (en) | Method for determining whether subject has developed cervical cancer, and device using same | |
| CN109446627A (en) | Endoscopic images recognition methods and device | |
| CN111144271A (en) | Method and system for automatically identifying biopsy parts and biopsy quantity under endoscope | |
| CN111839428A (en) | Method for improving detection rate of colonoscope adenomatous polyps based on deep learning | |
| CN114842000A (en) | Endoscope image quality evaluation method and system | |
| Wang et al. | Localizing and identifying intestinal metaplasia based on deep learning in oesophagoscope | |
| CN113139937B (en) | Digestive tract endoscope video image identification method based on deep learning | |
| CN110772210B (en) | Diagnosis interaction system and method | |
| CN111047582A (en) | Crohn's disease auxiliary diagnosis system under enteroscope based on degree of depth learning | |
| CN114332858A (en) | Focus detection method and device and focus detection model acquisition method | |
| CN114581408A (en) | Gastroscope polyp detection method based on YOLOV5 |
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 |