CN113990456A - Deep learning-based graphical analysis and screening method and system for early cancers of digestive tract - Google Patents
Deep learning-based graphical analysis and screening method and system for early cancers of digestive tract Download PDFInfo
- Publication number
- CN113990456A CN113990456A CN202111317482.XA CN202111317482A CN113990456A CN 113990456 A CN113990456 A CN 113990456A CN 202111317482 A CN202111317482 A CN 202111317482A CN 113990456 A CN113990456 A CN 113990456A
- Authority
- CN
- China
- Prior art keywords
- lesion
- early warning
- digestive tract
- image
- early
- 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
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- 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
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Image Analysis (AREA)
- Endoscopes (AREA)
Abstract
The invention provides a deep learning-based graphical analysis and screening system for early cancers of the digestive tract, which comprises an intercepted image input module, a data acquisition module and a data analysis module, wherein the intercepted image input module is used for intercepting images from a real-time acquisition video of an endoscope as input data of an image recognition module; the lesion area detection module is used for inputting the real-time video screenshot of the endoscope into the trained deep convolutional neural network model for detection; and the prediction output module is used for outputting the acquired lesion type and the percentage of the suspicious lesion matching degree and displaying the percentages on a system interface. Whether a suspected canceration area exists in a gastroscopy image can be identified through the system, the position, the size and the type of a focus are accurately identified and displayed, and the suspected focus is scored, so that a doctor is assisted in screening early cancers of the upper digestive tract, the diagnosis efficiency and the diagnosis accuracy are improved, the screening of the early cancers of the upper digestive tract is facilitated to be popularized to primary medical institutions, and the problem of unbalanced distribution of medical resources in China is balanced.
Description
Technical Field
The invention relates to the field of digestive tract lesion detection, in particular to a digestive tract early cancer graphical analysis screening method and a digestive tract early cancer graphical analysis screening system based on deep learning.
Background
In the early-stage symptom discovery mode of the upper digestive tract, an endoscope is adopted to observe a video acquired in real time by naked eyes, and the primary screening of the upper digestive tract lesion is combined with symptoms such as humps, depressions, redness, ulcers, white fur (leukoplakia), erosion, pachylosis and the like summarized by Paris classification. And then various types of typing after screening means such as dyeing, NBI, amplification and the like are adopted for observation, so that the discovery and judgment of early lesions of diseases such as Batttert esophagus, cardiac carcinoma, cardiac early cancer, reflux esophagitis, chronic non-atrophic gastritis, chronic atrophic gastritis, duodenal ulcer, esophageal early cancer, gastric ulcer and gastric ulcer are achieved.
With the accelerated pace of life, the pressure of life is increased, improper diet for a long time and irregular diet are commonly caused, the incidence rate of chronic diseases of the digestive system is gradually increased, and the canceration rate of the chronic diseases is increased. The early detection of the disease condition is helpful for the treatment of the patient and is more helpful for the rehabilitation of the patient.
The current screening means aiming at the early cancer of the upper digestive tract mainly comprises the technologies of naked eyes (white light), staining, NBI, amplification and the like. If abnormality is found during visual white light examination, further judgment is carried out by adopting means such as dyeing or NBI (negative fluorescence ionization), amplification and the like, or tissue biopsy is taken, but the pathological change stage is judged by visual (white light) completely depending on the experience of a doctor, so the experience of the doctor determines the probability of finding the early cancer pathological change, and for most local hospitals, the problems of low early cancer finding rate and the like caused by insufficient experience and quantity of the doctor are also faced.
Disclosure of Invention
The invention aims to provide a deep learning-based graphical structured analysis screening method and a deep learning-based graphical structured analysis screening system for early cancers of the digestive tract, which can identify whether a suspected lesion area exists in a selected area of an endoscopic image, accurately identify and display the position and type of a lesion, and score the suspected lesion, so that a doctor is assisted in screening the early cancers of the upper digestive tract, the diagnosis efficiency and accuracy are improved, the probability of discovering the precancerous lesions of the doctor is increased, and the problem of unbalanced distribution of medical resources in China is balanced.
In order to achieve the purpose, the invention provides the following technical scheme:
the method for graphically analyzing and screening the early cancer of the digestive tract based on deep learning comprises the following steps:
s1, inputting an account password to log in the system, watching a real-time video and recording a video;
s2, performing local image capture on the current image, marking a suspected lesion area, and storing the image after confirmation and storage in a local disk;
s3, alarming according to the type of the suspected lesion area;
and S4, double-clicking the alarm event, popping up the corresponding folder position, and displaying the related image information.
Preferably, the specific step of step S2 includes:
s21, capturing a suspected lesion area by the AI service of image recognition in the video playing process;
s22, detecting the region through a deep convolutional neural network model trained through target detection based on deep learning to obtain the percentage of the matching degree of the suspicious lesion;
and S23, marking the suspicious lesion area.
Preferably, the specific contents of step S3 include:
setting a suspicious lesion grade early warning threshold value and a lesion type, wherein the high-similarity lesion early warning threshold value is 90%, the important observation early warning threshold value is 80%, the biopsy taking early warning threshold value is 70% and the attention observation early warning threshold value is 50%; the early warning color of high-similarity lesion is red, the early warning color of key observation is yellow, the early warning color of biopsy taking is green, and the early warning color of attention observation is blue; the pathological change types comprise esophagus white light and stomach white light, and the specific setting items comprise: redness, bumps, depressions, white moss, ulcers, erosion, mucosal roughness, suspicious lesions, and significant lesions; and clicking the lesion type module, acquiring and displaying the percentage of the matching degree of the suspicious lesions identified by the image, and flashing corresponding early warning colors and displaying early warning levels when the percentage of the matching degree of the suspicious lesions exceeds an early warning threshold value.
It is another object of the present invention to provide a deep learning based graphical analysis and screening system for early cancer of digestive tract, comprising
The intercepted image input module is used for intercepting an image from a real-time acquisition video of the endoscope as input data of the image identification module;
the lesion area detection module is used for inputting the real-time video screenshot of the endoscope into a trained deep convolutional neural network model for detection, acquiring the detected suspected lesion type and the matching percentage of the suspicious lesion, adding a marking frame to mark the area if the suspected lesion area is detected, and outputting the detected lesion type and the matching percentage of the suspicious lesion;
and the prediction output module is used for outputting the acquired lesion type and the percentage of the suspicious lesion matching degree and displaying the percentages on a system interface.
Preferably, the deep convolutional neural network model consists of three neural networks 76 × 255, 38 × 255, and 19 × 255.
Compared with the prior art, the invention has the beneficial effects that:
by adopting the analysis and screening method and the system, whether a suspected lesion area exists in the selected area of the endoscope image can be identified, the position and the type of the lesion can be accurately identified and displayed, and the suspected lesion can be scored, so that a doctor is assisted in screening early cancers of the upper digestive tract, the diagnosis efficiency and the diagnosis accuracy are improved, the probability of finding the precancerous lesion by the doctor is increased, and the problem of unbalanced distribution of medical resources in China is balanced.
Drawings
FIG. 1 is a schematic block diagram of a deep learning based graphical structured analysis screening system for early cancer of digestive tract according to an embodiment of the present invention;
FIG. 2 is an interface for operating a deep learning based graphical structured analysis and screening system for early cancer of digestive tract according to an embodiment of the present invention;
FIG. 3 is a suspected lesion area type setting interface of a deep learning based graphical structured analysis and screening system for early cancers of digestive tract according to an embodiment of the present invention;
FIG. 4 is a suspected lesion area early warning threshold setting interface of a deep learning based graphical structured analysis and screening system for early cancers of digestive tract according to an embodiment of the present invention;
FIG. 5 is an alarm event list interface of a deep learning based graphical structured analysis screening system for early cancers of the digestive tract in accordance with an embodiment of the present invention;
FIG. 6 is a type alarm interface for suspected diseased regions of a deep learning based graphical structured analysis screening system for early cancers of the digestive tract according to an embodiment of the present invention;
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1-6, the present invention provides a deep learning based graphical structured analysis screening method for early cancer of digestive tract, which comprises the following steps:
s1, inputting an account password to log in the system, watching a real-time video and recording a video;
s2, performing local image capture on the current image, marking a suspected lesion area, and storing the image after confirmation and storage in a local disk;
the specific steps of step S2 include:
s21, capturing a suspected lesion area by the AI service of image recognition in the video playing process;
s22, detecting the region through a deep convolutional neural network model trained through target detection based on deep learning to obtain the percentage of the matching degree of the suspicious lesion;
s23, labeling a suspicious lesion area;
s3, alarming according to the type of the suspected lesion area;
the specific content of step S3 includes: setting suspicious lesion level early warning threshold (for example, high-similarity lesion early warning threshold is 90%, important observation early warning threshold is 80%, biopsy taking early warning threshold is 70% and attention observation early warning threshold is 50%) and lesion type. The early warning color of the high-similarity lesion is red, the early warning color of the important observation is yellow, the early warning color of the biopsy is green, and the early warning color of the attention observation is blue. The pathological change types comprise esophagus white light and stomach white light, and the specific setting items comprise: redness, bumps, depressions, white moss, ulcers, erosion, mucosal roughness, suspicious lesions, and significant lesions; and clicking the lesion type module, acquiring and displaying the percentage of the matching degree of the suspicious lesions identified by the image, and flashing corresponding early warning colors and displaying early warning levels when the percentage of the matching degree of the suspicious lesions exceeds an early warning threshold value.
And S4, double-clicking the alarm event, popping up the corresponding folder position, and displaying the related image information.
The deep learning-based graphical structured analysis and screening system for early cancers of the digestive tract comprises:
the intercepted image input module is used for intercepting an image from a real-time acquisition video of the endoscope as input data of the image identification module;
a lesion area detection module, configured to input the endoscope real-time video screenshot into a trained deep convolutional neural network model (the model is composed of three neural networks, namely 76 × 255, 38 × 255 and 19 × 255), to perform detection, and obtain a detected suspected lesion type and a suspicious lesion matching percentage, that is, the deep convolutional neural network model completes detection through a GPU acceleration model or a CPU compatible model, and if a suspected lesion area is detected, a labeling frame is added to label the area, and the detected lesion type and the suspicious lesion matching percentage are output;
and the prediction output module is used for outputting the acquired lesion type and the percentage of the suspicious lesion matching degree and displaying the percentages on a system interface.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (5)
1. The method for graphically analyzing and screening early cancers of the digestive tract based on deep learning is characterized by comprising the following steps of:
s1, inputting an account password to log in the system, watching a real-time video and recording a video;
s2, performing local image capture on the current image, marking a suspected lesion area, and storing the image after confirmation and storage in a local disk;
s3, alarming according to the type of the suspected lesion area;
and S4, double-clicking the alarm event, popping up the corresponding folder position, and displaying the related image information.
2. The deep learning based graphical analysis screening method for early cancers of digestive tract as claimed in claim 1, wherein the step S2 comprises the following steps:
s21, capturing a suspected lesion area by the AI service of image recognition in the video playing process;
s22, detecting the region through a deep convolutional neural network model trained through target detection based on deep learning to obtain the percentage of the matching degree of the suspicious lesion;
and S23, marking the suspicious lesion area.
3. The deep learning based graphical analysis and screening method for early cancers of digestive tract according to claim 1, wherein the step S3 comprises:
setting a suspicious lesion grade early warning threshold value and a lesion type, wherein the high-similarity lesion early warning threshold value is 90%, the important observation early warning threshold value is 80%, the biopsy taking early warning threshold value is 70% and the attention observation early warning threshold value is 50%; the early warning color of high-similarity lesion is red, the early warning color of key observation is yellow, the early warning color of biopsy taking is green, and the early warning color of attention observation is blue; the pathological change types comprise esophagus white light and stomach white light, and the specific setting items comprise: redness, bumps, depressions, white moss, ulcers, erosion, mucosal roughness, suspicious lesions, and significant lesions; and clicking the lesion type module, acquiring and displaying the percentage of the matching degree of the suspicious lesions identified by the image, and flashing corresponding early warning colors and displaying early warning levels when the percentage of the matching degree of the suspicious lesions exceeds an early warning threshold value.
4. A deep learning based graphical analysis and screening system for early cancers of digestive tract as claimed in any one of claims 1 to 3 comprising
The intercepted image input module is used for intercepting an image from a real-time acquisition video of the endoscope as input data of the image identification module;
the lesion area detection module is used for inputting the real-time video screenshot of the endoscope into a trained deep convolutional neural network model for detection, acquiring the detected suspected lesion type and the matching percentage of the suspicious lesion, adding a marking frame to mark the area if the suspected lesion area is detected, and outputting the detected lesion type and the matching percentage of the suspicious lesion;
and the prediction output module is used for outputting the acquired lesion type and the percentage of the suspicious lesion matching degree and displaying the percentages on a system interface.
5. The deep learning-based graphical analysis and screening system for early cancers of the digestive tract according to claim 4, wherein the deep convolutional neural network model is composed of three neural networks of 76 x 255, 38 x 255 and 19 x 255.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111317482.XA CN113990456A (en) | 2021-11-09 | 2021-11-09 | Deep learning-based graphical analysis and screening method and system for early cancers of digestive tract |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111317482.XA CN113990456A (en) | 2021-11-09 | 2021-11-09 | Deep learning-based graphical analysis and screening method and system for early cancers of digestive tract |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113990456A true CN113990456A (en) | 2022-01-28 |
Family
ID=79747297
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111317482.XA Pending CN113990456A (en) | 2021-11-09 | 2021-11-09 | Deep learning-based graphical analysis and screening method and system for early cancers of digestive tract |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113990456A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049666A (en) * | 2022-08-16 | 2022-09-13 | 浙江卡易智慧医疗科技有限公司 | Endoscope virtual biopsy device based on color wavelet covariance depth map model |
| CN115311268A (en) * | 2022-10-10 | 2022-11-08 | 武汉楚精灵医疗科技有限公司 | Esophagus endoscope image identification method and device |
-
2021
- 2021-11-09 CN CN202111317482.XA patent/CN113990456A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049666A (en) * | 2022-08-16 | 2022-09-13 | 浙江卡易智慧医疗科技有限公司 | Endoscope virtual biopsy device based on color wavelet covariance depth map model |
| CN115049666B (en) * | 2022-08-16 | 2022-11-08 | 浙江卡易智慧医疗科技有限公司 | Endoscope virtual biopsy device based on color wavelet covariance depth map model |
| CN115311268A (en) * | 2022-10-10 | 2022-11-08 | 武汉楚精灵医疗科技有限公司 | Esophagus endoscope image identification method and device |
| CN115311268B (en) * | 2022-10-10 | 2022-12-27 | 武汉楚精灵医疗科技有限公司 | Esophagus endoscope image identification method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7335552B2 (en) | Diagnostic imaging support device, learned model, operating method of diagnostic imaging support device, and diagnostic imaging support program | |
| WO2020071677A1 (en) | Method and apparatus for diagnosing gastric lesions by using deep learning on gastroscopy images | |
| US9230163B2 (en) | Cascade analysis for intestinal contraction detection | |
| CN113990456A (en) | Deep learning-based graphical analysis and screening method and system for early cancers of digestive tract | |
| WO2020224153A1 (en) | Nbi image processing method based on deep learning and image enhancement, and application thereof | |
| WO2020215810A1 (en) | Image recognition-based narrowband image detection method for colonoscopy procedure | |
| EP1875855B1 (en) | Image processing apparatus, image processing method, and image processing program | |
| CN107767365A (en) | A kind of endoscopic images processing method and system | |
| WO2007119297A1 (en) | Image processing device for medical use and image processing method for medical use | |
| JP6883662B2 (en) | Endoscope processor, information processing device, endoscope system, program and information processing method | |
| WO2020071678A2 (en) | Endoscopic apparatus and method for diagnosing gastric lesion on basis of gastroscopy image obtained in real time | |
| Cui et al. | Bleeding detection in wireless capsule endoscopy images by support vector classifier | |
| Ghosh et al. | Automatic bleeding detection in wireless capsule endoscopy based on RGB pixel intensity ratio | |
| KR102095730B1 (en) | Method for detecting lesion of large intestine disease based on deep learning | |
| Ghosh et al. | Block based histogram feature extraction method for bleeding detection in wireless capsule endoscopy | |
| KR102505791B1 (en) | Control method, apparatus, and program of lesion determination system acquired through real-time image | |
| Nakajo et al. | Anatomical classification of pharyngeal and laryngeal endoscopic images using artificial intelligence | |
| US20090080768A1 (en) | Recognition method for images by probing alimentary canals | |
| Ghosh et al. | An automatic bleeding detection scheme in wireless capsule endoscopy based on statistical features in hue space | |
| Liu et al. | Automatic detection of early gastrointestinal cancer lesions based on optimal feature extraction from gastroscopic images | |
| Suman et al. | Detection and classification of bleeding using statistical color features for wireless capsule endoscopy images | |
| CN114332056A (en) | Early gastric cancer endoscope real-time auxiliary detection system based on target detection algorithm | |
| Hwang et al. | Stool detection in colonoscopy videos | |
| JP2021037356A (en) | Processor for endoscope, information processing device, endoscope system, program and information processing method | |
| JP7116849B2 (en) | Endoscope processor, endoscope system, information processing device, program and information processing method |
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 |