CN114119511A - Colonoscope image evaluation method and system based on EfficientNet structure - Google Patents
Colonoscope image evaluation method and system based on EfficientNet structure Download PDFInfo
- Publication number
- CN114119511A CN114119511A CN202111339539.6A CN202111339539A CN114119511A CN 114119511 A CN114119511 A CN 114119511A CN 202111339539 A CN202111339539 A CN 202111339539A CN 114119511 A CN114119511 A CN 114119511A
- Authority
- CN
- China
- Prior art keywords
- image
- colon
- efficientnet
- images
- intestinal tract
- 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
- 238000011156 evaluation Methods 0.000 title claims abstract description 40
- 210000001072 colon Anatomy 0.000 claims abstract description 38
- 210000001035 gastrointestinal tract Anatomy 0.000 claims abstract description 26
- 210000003384 transverse colon Anatomy 0.000 claims abstract description 23
- 238000002360 preparation method Methods 0.000 claims abstract description 19
- 230000003749 cleanliness Effects 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000000034 method Methods 0.000 claims description 18
- 230000004913 activation Effects 0.000 claims description 9
- 230000002440 hepatic effect Effects 0.000 claims description 7
- 239000002131 composite material Substances 0.000 claims description 6
- 230000003993 interaction Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 3
- 230000005284 excitation Effects 0.000 claims description 3
- 238000003384 imaging method Methods 0.000 claims description 3
- 238000002052 colonoscopy Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 210000004185 liver Anatomy 0.000 description 4
- 210000000952 spleen Anatomy 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 210000001815 ascending colon Anatomy 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 210000001731 descending colon Anatomy 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 210000003405 ileum Anatomy 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000001079 digestive effect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000003608 fece Anatomy 0.000 description 1
- 230000037406 food intake Effects 0.000 description 1
- 235000012631 food intake Nutrition 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 235000011389 fruit/vegetable juice Nutrition 0.000 description 1
- 210000003767 ileocecal valve Anatomy 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000000968 intestinal effect Effects 0.000 description 1
- 210000000936 intestine Anatomy 0.000 description 1
- 210000002429 large intestine Anatomy 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000010412 perfusion Effects 0.000 description 1
- 210000000664 rectum Anatomy 0.000 description 1
- 238000013341 scale-up Methods 0.000 description 1
- 210000001599 sigmoid colon Anatomy 0.000 description 1
- 238000001356 surgical procedure Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30028—Colon; Small intestine
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention relates to a colonoscope image evaluation method based on an EfficientNet structure, which comprises the following steps of: step S1: acquiring colorectal sequence images, and preliminarily classifying the colorectal sequence images into three subsequences of images of a left half colon, a transverse colon and a right half colon; step S2, preprocessing the subsequence image; step S3, obtaining the optimal classification based on the scalable EfficientNet network model according to the preprocessed subsequence images; step S4: based on BBPS intestinal tract preparation evaluation criteria, namely, the three subsequences of images of the left half colon, the transverse colon and the right half colon are respectively scored, the score of the image with the worst intestinal tract cleanliness of each segment is accumulated, and whether the colonoscope image is qualified or not is further evaluated. The invention effectively realizes the automatic and accurate evaluation of the intestinal tract preparation and can greatly reduce the burden of doctors.
Description
Technical Field
The invention relates to the field of image mode identification and classification, in particular to a colonoscope image evaluation method and system based on an EfficientNet structure.
Background
Before colonoscopy surgery, the process of cleaning the colon by removing feces and digestive juices and emptying the bowel of foreign matter is known as bowel preparation. Bowel preparation is an important prerequisite for successful colonoscopy. It is common to use a combination of stopping food intake prior to colonoscopy and oral bowel cleansing to cleanse the bowel. However, the domestic related documents report that the intestinal tract preparation inadequacy rate of outpatients is 12% -25%, and the intestinal tract preparation inadequacy rate of the elderly patients is higher and is 17% -32%. Foreign documents report that the incidence of insufficient preparation of intestinal tracts of patients subjected to colonoscopy is 20% -25%. Therefore, it is important to accurately and objectively evaluate the quality of the preparation of the intestinal tract.
Disclosure of Invention
In view of the above, the present invention provides a colonoscope image evaluation method and system based on the EfficientNet structure, which can effectively implement automatic and accurate evaluation of intestinal tract preparation and greatly reduce the burden of doctors.
In order to achieve the purpose, the invention adopts the following technical scheme:
a colonoscope image evaluation method based on an EfficientNet structure comprises the following steps:
step S1: acquiring colorectal sequence images, and preliminarily classifying the colorectal sequence images into three subsequences of images of a left half colon, a transverse colon and a right half colon;
step S2, preprocessing the subsequence image;
step S3, obtaining the optimal classification based on the scalable EfficientNet network model according to the preprocessed subsequence images;
step S4: based on BBPS intestinal tract preparation evaluation criteria, namely, the three subsequences of images of the left half colon, the transverse colon and the right half colon are respectively scored, the score of the image with the worst intestinal tract cleanliness of each segment is accumulated, and whether the colonoscope image is qualified or not is further evaluated.
Further, the preprocessing comprises: reject the failing image and reset the resolution to fit the network.
Further, the EfficientNet network model comprises a plurality of MBConv volume blocks, and the expansion method is used for solving the composite coefficients of three dimensions, and specifically comprises the following steps: setting the coefficients of network depth d, network width w and image resolution r to be respectively subjected to parametersMeasured as formula 1, with the constraint of formula 2
Further, the MBConv volume block is specifically as follows: caching a Swish activation function in a ReLU activation function; a drop _ connect method is adopted to replace the traditional drop method; BN is placed between the activation function and the convolutional layer; a structure similar to residual linking is used, and an SE layer is used in a short connection part; the SE module considers the interdependency between model channels and comprises two parts of Squeeze and Excitation.
A colonoscope image evaluation system based on an EfficientNet structure comprises a manual interaction unit, a data classification unit and an evaluation unit;
the artificial interaction unit is used for separating the colorectal sequence image into three subsequences of images of a left half colon, a transverse colon and a right half colon by artificially identifying a hepatic flexure image and a splenic flexure image, and rejecting an obvious image which is unqualified in imaging;
the data classification unit takes EfficientNet as a basic network, and can manually scale a network model according to computational power configuration to obtain an optimal classification effect;
the evaluation unit is used for respectively scoring three subsequences of images of the left half colon, the transverse colon and the right half colon based on BBPS intestinal tract evaluation standards, and accumulating the image score with the worst cleanliness of each section of intestinal tract, so that whether the intestinal tract preparation is sufficient or not is evaluated.
Compared with the prior art, the invention has the following beneficial effects:
the invention realizes the automatic and accurate evaluation of the intestinal tract preparation and can greatly reduce the burden of doctors.
Drawings
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is a schematic diagram of filename labels of images of liver and spleen koji in an embodiment of the present invention;
FIG. 3 is a schematic view of an interface of an automatic evaluation system for BBPS bowel preparation in accordance with an embodiment of the present invention;
FIG. 4 is a schematic representation of the output of the system to a colonoscopy report in accordance with an embodiment of the present invention;
FIG. 5 is an MBConv module according to an embodiment of the present invention;
FIG. 6 is an SE module in an embodiment of the invention.
Detailed Description
The invention is further explained below with reference to the drawings and the embodiments.
Referring to fig. 1, the present invention provides a colonoscope image evaluation system based on the EfficientNet structure, which comprises
The artificial interaction unit is used for separating the colorectal sequence image into three subsequences of images of a left half colon, a transverse colon and a right half colon by artificially identifying the hepatic flexure image and the splenic flexure image, and rejecting an obvious image which is unqualified in imaging;
the data classification unit is used for configuring a network model which can be manually scaled according to calculation power by taking EfficientNet as a basic network so as to obtain the optimal classification effect;
and the evaluation unit is used for evaluating intestinal tracts based on BBPS (BBPS), namely respectively scoring three subsequences of images of the left half colon, the transverse colon and the right half colon, and accumulating the image score with the worst cleanliness of each intestinal tract so as to evaluate whether the intestinal tract preparation is sufficient or not.
In this embodiment, the colon image acquisition generally adopts a retroscopic photograph acquisition mode, i.e., the image is firstly taken from the hepatic portal to the end of the ileum, and the image is checked and photographed by using a freezing button in the process of going back the transpedicular lens; there are images for each structure and fixed area in the large intestine, from the end of the ileum, ileocecal valve, appendix recess, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending colon, sigmoid colon to rectum there are at least 1-2 high quality images, and because of the long range of ascending colon, transverse colon, descending colon, more images will be acquired, 20-100 images in total.
In this example, referring to fig. 2, the colon is divided into 3 segments of left colon, transverse colon and right colon by hepatic flexure and splenic flexure. For example, the colon sequence of a patient is 49 images, the initial file names of the images are '0. bmp', 1.bmp ', 2. bmp', n.bmp ',. 49. bmp', the doctor can quickly browse and find the images of the liver koji and the spleen koji, the file name of the image of the liver koji is automatically changed into 'G.bmp' by clicking the right button of a mouse, and the letter 'G' is added after the serial number to be used as a mark; similarly, the serial number of the file name of the spleen song image is added with a letter "P" as a mark, as shown in FIG. 2. Thus, the evaluation of the segments can be performed based on the file names.
Referring to fig. 3, the interface of manual interaction includes the following operation steps:
(1) "open file" imports a sample of the subject's sequential colon images into a file list, clicks on the images in the list, and can view the images one by one in an "image viewing area".
(2) Right clicking the image in the file list can mark the 'liver/spleen koji' image; or unqualified images caused by improper photographing can be eliminated.
(3) Clicking an artificial intelligence automatic evaluation button, and obtaining the cleanliness score of each image by a score column, namely, classifying (0, 1, 2, 3) scores according to a deep learning model 4.
(4) The "result evaluation area" gives the lowest results of evaluation for each of the three segments of the left colon, the transverse colon and the right colon, and a BBPS total score is obtained.
(5) If the manual evaluation is found to be wrong, the evaluation result of each segment can be corrected by operating the files in the file list, and correct evaluation is obtained.
The content of the examination report is shown in fig. 4 below.
Preferably, in this embodiment, the colonoscopy generally acquires 20-100 color images in total, and the dynamic range is large, i.e. the sequence of image detection is completed when the examination report of one patient is completed, so that the test speed of the system is high in the process of clinical practice. Therefore, the adopted deep learning network can simultaneously take the precision and the speed into consideration.
To improve the detection accuracy of the image, the method can be improved from three dimensions of network depth, network width and image resolution, networks such as ResNet and VGG improve the detection accuracy by increasing the network depth, networks such as GoogleNet and MobileNet widen the network width by increasing the number of channels to achieve the purpose of improving the accuracy, and although the detection result is optimized, the method brings adverse conditions such as network degradation or abrupt increase of calculation amount.
Preferably, in this embodiment, EfficientNet is selected as the deep learning network of the system artificial intelligence module. The network adopts a composite model expansion method and combines a neural structure search technology, the model is zoomed from three dimensions of network depth, network width and image resolution, an optimal group of composite coefficients can be obtained, namely the zoom ratios for balancing the three dimensions of the network depth, the network width and the image resolution are obtained, the speed is higher than that of other networks, and the precision is higher.
Preferably, in this embodiment, the composite model expansion method of the EfficientNet network structure is used for solving composite coefficients of three dimensions, and is a solution process with constraints. Setting the coefficients of network depth d, network width w and image resolution r to be respectively subjected to parametersBy weight, as in equation 1, with constraintsProvided that the formula 2
WhereinThe squaring is constrained because if the width or resolution is increased by a factor of 2, the amount of computation is increased by a factor of 4, but if the depth is increased by a factor of 2, the amount of computation is increased by a factor of only 2.
When the scaling factor is fixedSetting the calculated amount to be 2 times of the original amount, and obtaining the optimal parameters by the grid search technologyThe resulting basic model was named: EfficientNet-B0.
When fixedParameter values, using differencesBy amplifying the network, a sequence model of EfficientNet-B1, EfficientNet-B7 can be obtained.
In the embodiment, the task of classifying the colonoscope image into 4 is related, and through public data set experiments, the first generation model EfficientNet-B0 can meet the requirements of the accuracy and the speed of the system at the same time on a machine with a GPU video card, and the evaluation of an average examinee can be completed within 1-3 seconds. The model parameter of the model is only about 5M, and the average classification accuracy can reach more than 95%. In practical clinical application, the network can be adjusted manually when the computational power configuration allowsScaling factorTo scale up the network for better classification accuracy. The EfficientNet-B0 structure is shown in Table 2 below.
TABLE 2 EfficientNet-B0 network architecture
It can be seen that the inside of the EfficientNet model is realized by a plurality of MBConv convolution blocks, and the specific structure of each MBConv convolution block is shown in fig. 5 below. Wherein, the Swish activation function is cached in the ReLU activation function; a drop _ connect method is adopted to replace the traditional drop method; BN is placed between the activation function and the convolutional layer; a structure similar to the residual linking is used and the SE layer is used in the short connection part. The SE module considers the interdependency between model channels, and comprises two parts of Squeeze (compression) and Excitation, as shown in FIG. 6.
To verify the validity of the network model, a public data set of colonoscope images, Nerthus, with cleanliness markers collected by norwegian B æ rum hospital was chosen. The data set was prepared by extracting 5525 frames of images from 21 video of subjects, with a resolution of 720 × 576, in jpg format, and by an experienced endoscopist, performing an evaluation of the cleanliness of the intestine according to the Boston scale, and was divided into 4 categories from 0 to 3, with the number distribution of each category shown in table 3.
TABLE 3 Nerthus data set Classification
The experiment is carried out by adopting an EfficientNet-B0 model, the data set (80% of the training set and 20% of the testing set) is tested, the classification accuracy is high, and all indexes are shown in Table 4.
Table 4 results of experiments on Nerthus dataset
Method | Acc | Prec | Rec | Spec | F1 score | Mcc |
Comparison scheme | 0.949 | 0.901 | 0.901 | 0.960 | 0.899 | 0.863 |
The invention | 0.977 | 0.956 | 0.954 | 0.979 | 0.953 | 0.939 |
Acc (Accuracy), accuracy, and sample proportion representing the coincidence of true values and predicted values in all samples; prec (precision), the precision, which indicates the proportion of samples with positive true values in samples with positive predicted values; recall (Recall), the recall rate, which indicates the proportion of samples with positive true values and positive predicted values in samples with positive true values; specificity, which is a sample proportion that the true value and the predicted value are both negative in a sample with a negative true value; f1 score, which is the index of the harmonic mean of the precision rate and the recall rate and the comprehensive performance; mcc (matthews correlation coefficient), a mazis correlation coefficient, represents the classification correlation coefficient between the predicted value and the true value of the sample, and the return value is between-1 and +1, and the closer to +1, the better the classification performance.
In this example, the Boston intestinal Preparation Scale (BBPS) shows that the cleanliness of the intestinal tract can be classified into 4 categories as described in Table 1, and the categories can be respectively rated as 0 to 3.
TABLE 1 Boston intestinal tract preparation evaluation Scale (BBPS)
The BBPS evaluation index divides the colon into 3 segments of left colon, transverse colon and right colon by taking hepatic flexure and splenic flexure as boundaries, and performs sectional evaluation after cleaning treatment such as extraction and perfusion of intestinal tracts. Each colon is evaluated for 0 to 3 points, and the total point is 0 to 9 points; the evaluation of each segment of colon is more than or equal to 2, and the intestinal tract is fully prepared; a total score of < 6 for 3 segments or < 2 for any segment is considered to be inadequate bowel preparation.
The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.
Claims (5)
1. A colonoscope image evaluation method based on an EfficientNet structure is characterized by comprising the following steps:
step S1: acquiring colorectal sequence images, and preliminarily classifying the colorectal sequence images into three subsequences of images of a left half colon, a transverse colon and a right half colon;
step S2, preprocessing the subsequence image;
step S3, obtaining the optimal classification based on the scalable EfficientNet network model according to the preprocessed subsequence images;
step S4: based on BBPS intestinal tract preparation evaluation criteria, namely, the three subsequences of images of the left half colon, the transverse colon and the right half colon are respectively scored, the score of the image with the worst intestinal tract cleanliness of each segment is accumulated, and whether the colonoscope image is qualified or not is further evaluated.
2. The colonoscope image evaluation method based on the EfficientNet structure according to claim 1, wherein said preprocessing comprises: reject the failing image and reset the resolution to fit the network.
3. According to the rightThe colonoscope image evaluation method based on the EfficientNet structure according to claim 1, characterized in that the EfficientNet network model comprises a plurality of MBConv volume blocks, and the expansion method is used for solving the composite coefficient of three dimensions, specifically as follows: setting the coefficients of network depth d, network width w and image resolution r to be respectively subjected to parametersMeasured as formula 1, with the constraint of formula 2
4. A colonoscope image evaluation method based on the EfficientNet structure according to claim 3, wherein the MBConv volume block is as follows: caching a Swish activation function in a ReLU activation function; a drop _ connect method is adopted to replace the traditional drop method; BN is placed between the activation function and the convolutional layer; a structure similar to residual linking is used, and an SE layer is used in a short connection part; the SE module considers the interdependency between model channels and comprises two parts of Squeeze and Excitation.
5. A colonoscope image evaluation system based on an EfficientNet structure is characterized by comprising a manual interaction unit, a data classification unit and an evaluation unit;
the artificial interaction unit is used for separating the colorectal sequence image into three subsequences of images of a left half colon, a transverse colon and a right half colon by artificially identifying a hepatic flexure image and a splenic flexure image, and rejecting an obvious image which is unqualified in imaging;
the data classification unit takes EfficientNet as a basic network, and can manually scale a network model according to computational power configuration to obtain an optimal classification effect;
the evaluation unit is used for respectively scoring three subsequences of images of the left half colon, the transverse colon and the right half colon based on BBPS intestinal tract evaluation standards, and accumulating the image score with the worst cleanliness of each section of intestinal tract, so that whether the intestinal tract preparation is sufficient or not is evaluated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111339539.6A CN114119511A (en) | 2021-11-12 | 2021-11-12 | Colonoscope image evaluation method and system based on EfficientNet structure |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111339539.6A CN114119511A (en) | 2021-11-12 | 2021-11-12 | Colonoscope image evaluation method and system based on EfficientNet structure |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114119511A true CN114119511A (en) | 2022-03-01 |
Family
ID=80379115
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111339539.6A Pending CN114119511A (en) | 2021-11-12 | 2021-11-12 | Colonoscope image evaluation method and system based on EfficientNet structure |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114119511A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114596288A (en) * | 2022-03-11 | 2022-06-07 | 山东大学齐鲁医院 | Ulcerative colitis assessment system under endoscope, medium and electronic equipment |
CN115082448A (en) * | 2022-07-26 | 2022-09-20 | 青岛美迪康数字工程有限公司 | Method and device for scoring cleanliness of intestinal tract and computer equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110916606A (en) * | 2019-11-15 | 2020-03-27 | 武汉楚精灵医疗科技有限公司 | Real-time intestinal cleanliness scoring system and method based on artificial intelligence |
KR102222547B1 (en) * | 2020-07-15 | 2021-03-04 | 주식회사 웨이센 | AI(Artificial Intelligence) -based colonoscopy image analysis method |
CN113516659A (en) * | 2021-09-15 | 2021-10-19 | 浙江大学 | Medical image automatic segmentation method based on deep learning |
-
2021
- 2021-11-12 CN CN202111339539.6A patent/CN114119511A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110916606A (en) * | 2019-11-15 | 2020-03-27 | 武汉楚精灵医疗科技有限公司 | Real-time intestinal cleanliness scoring system and method based on artificial intelligence |
KR102222547B1 (en) * | 2020-07-15 | 2021-03-04 | 주식회사 웨이센 | AI(Artificial Intelligence) -based colonoscopy image analysis method |
CN113516659A (en) * | 2021-09-15 | 2021-10-19 | 浙江大学 | Medical image automatic segmentation method based on deep learning |
Non-Patent Citations (3)
Title |
---|
吉娜・伊力;高峰;: "肠道准备评分量表在结肠镜检查中的应用", 国际消化病杂志, no. 05, 25 October 2015 (2015-10-25) * |
边坤鹏: "基于卷积神经网络的乳腺肿瘤良恶性分类方法研究", 中国优秀硕士学位论文全文数据库 (医药卫生科技辑), 31 July 2021 (2021-07-31), pages 23 - 24 * |
阿依木克地斯・亚力孔;庄惠军;蔡世伦;牛雪静;谭伟敏;颜波;姚礼庆;周平红;钟芸诗;: "基于深度学习人工智能在结肠镜检查中应用研究", 中国实用外科杂志, no. 03, 1 March 2020 (2020-03-01) * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114596288A (en) * | 2022-03-11 | 2022-06-07 | 山东大学齐鲁医院 | Ulcerative colitis assessment system under endoscope, medium and electronic equipment |
CN114596288B (en) * | 2022-03-11 | 2024-05-10 | 山东大学齐鲁医院 | Ulcerative colitis evaluation system under endoscope, medium and electronic equipment |
CN115082448A (en) * | 2022-07-26 | 2022-09-20 | 青岛美迪康数字工程有限公司 | Method and device for scoring cleanliness of intestinal tract and computer equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Mishra et al. | Diabetic retinopathy detection using deep learning | |
CN108898595B (en) | Construction method and application of positioning model of focus region in chest image | |
CN114119511A (en) | Colonoscope image evaluation method and system based on EfficientNet structure | |
CN111160135A (en) | Urine red blood cell lesion identification and statistical method and system based on improved Faster R-cnn | |
CN113781489B (en) | Polyp image semantic segmentation method and device | |
WO2023004546A1 (en) | Traditional chinese medicine constitution recognition method and apparatus, and electronic device, storage medium and program | |
CN108305253A (en) | A kind of pathology full slice diagnostic method based on more multiplying power deep learnings | |
CN112151167A (en) | Intelligent screening method for six-age dental caries of children based on deep learning | |
CN111524124A (en) | Digestive endoscopy image artificial intelligence auxiliary system for inflammatory bowel disease | |
CN113222957A (en) | Multi-class focus high-speed detection method and system based on capsule lens image | |
CN111754485A (en) | Artificial intelligence ultrasonic auxiliary system for liver | |
CN115526834A (en) | Immunofluorescence image detection method and device, equipment and storage medium | |
CN113160151B (en) | Panoramic sheet decayed tooth depth identification method based on deep learning and attention mechanism | |
CN116188946B (en) | Blood glucose concentration prediction system based on neural network | |
Popa et al. | Automatic Diagnosis of High-Resolution Esophageal Manometry using Artificial Intelligence. | |
CN112784924A (en) | Rib fracture CT image classification method based on grouping aggregation deep learning model | |
CN112085742A (en) | NAFLD ultrasonic video diagnosis method based on context attention | |
CN111652840B (en) | Turbid screening and classifying device for X-ray chest X-ray image lung | |
CN113130050B (en) | Medical information display method and display system | |
CN113379724A (en) | Age inference system and method based on curved surface fault slice and deep learning | |
CN113553991A (en) | Tongue picture recognition method based on deep learning | |
Liu et al. | HFANet: hierarchical feature fusion attention network for classification of glomerular immunofluorescence images | |
CN113409319A (en) | Rib fracture detection model training system, method, detection system and detection method | |
CN112807008A (en) | Method and system for identifying actual mycoplasma pneumoniae and streptococcus pneumoniae of children based on imaging omics | |
CN117351317B (en) | Automatic identification method and system for last stool character picture |
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 |