Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Enteroscope withdrawal overspeed proportion monitoring method based on two-channel convolutional neural network

A convolutional neural network and colonoscopy technology, applied in the field of medical assistance, can solve the problems of multiple image information, low accuracy, loss, etc., and achieve the effects of strong anti-interference ability, good quality monitoring, and high accuracy

Pending Publication Date: 2020-10-09
WUHAN ENDOANGEL MEDICAL TECH CO LTD
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This method adopts the traditional hash algorithm, which needs to reduce the image to 9*8 size, which will lose more image information, the accuracy is slightly lower, and the anti-interference ability is poor

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Enteroscope withdrawal overspeed proportion monitoring method based on two-channel convolutional neural network
  • Enteroscope withdrawal overspeed proportion monitoring method based on two-channel convolutional neural network
  • Enteroscope withdrawal overspeed proportion monitoring method based on two-channel convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0033] S1. Collect 300 consecutive pictures of different colonoscopic examinations, including 150 cases of two different types of endoscopes. The pictures of each case need to include the rectum, sigmoid colon, descending colon, transverse colon, ascending colon, etc. Pictures of the parts.

[0034] S2. Clean up the collected picture set, and remove unqualified pictures such as fuzzy, incorrect parts, in vitro, and incorrect resolution in the picture set. The cleaned data set is marked by a professional doctor, and every two pictures form a pair. The label of the two pictures with a high degree of similarity is set to 1, and the label of the two pictures is completely different is set to 0. After manual labeling, the training set samples are composed.

[0035] Construct a two-channel convolutional neural network and train it with the training sample set to obtain the model 1 for calculating image similarity. The open source TensorFlow deep learning framework is used to const...

Embodiment 2

[0049] Compared with Embodiment 1, step S2 is optimized and described in detail, and the specific distinguishing techniques are as follows:

[0050] S in step S2 i It is the value of whether the two images are similar after being calculated by the trained model 1, and its range is 0-1, that is, S i =ConvNet(I i , I i-1 ), where I i and I i-1 are two adjacent images.

[0051] Obtain the real-time video of the colonoscopy examination through the endoscopy equipment, parse the video stream into pictures, and capture the pictures at a rate of 10 frames per second, and cache the current frame I 0 Colonoscopy picture I with the previous 9 frames i (where i=1...9).

[0052] the current frame I 0 Colonoscopy picture I with the previous 9 frames i (where i=1...9), form 9 pairs of images, and send them into the two-channel convolutional neural network, and each group of pictures is obtained after the model 1 trained in S3 to obtain the similarity coefficient S of the group of p...

Embodiment 3

[0055] Compared with Embodiment 2, step S3 is optimized and specifically described, and the specific differentiating techniques are as follows:

[0056] In step S3, by analyzing 50 sections of standard colonoscopy videos with withdrawal time>6min, 50 sections with withdrawal time of 4-6min standard colonoscopy videos, and 50 poor-quality colonoscopy videos with withdrawal timebase The value is 40. The current t time point colonoscopy examination speed V t Speed ​​greater than the standard speed V base Recorded as overspeed frame, count up to the current time, the ratio of overspeed frame number to the whole frame number, which is the current overspeed ratio, p is the proportion of overspeeding, PV is the number of overspeeding frames, and PA is the whole number of frames.

[0057] After the colonoscopy examination is completed, the speeding ratio p of the colonoscopy examination is calculated, and converted into the evaluation score of the colonoscopy examination score = 10...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to the technical field of medical assistance, in particular to an enteroscope withdrawal overspeed proportion monitoring method based on a two-channel convolutional neural network. The method comprises the following steps: S1, constructing the two-channel convolutional neural network, and training the two-channel convolutional neural network by using a training sample set toobtain a model 1 for calculating image similarity; S2, decoding a real-time video of enteroscopy into an image, calculating the similarity between the current frame of image and the previous i framesof images by using the model 1 to obtain i similarity coefficients Si, calculating a weighted similarity coefficient of the enteroscopy at the current time point, and converting the weighted similarity coefficient into an enteroscopy withdrawal speed; and S3, recording the image of which the enteroscopy speed Vi is greater than the standard speed at the current time point as an overspeed frame, and counting the overspeed proportion. According to the invention, a two-channel convolutional neural network is adopted to calculate the similarity coefficient of enteroscope images, calculate the enteroscope speed, and count the current enteroscope overspeed percentage. All characteristics of the image can be utilized, image information is not lost, the accuracy is high, and the anti-interferencecapability is high.

Description

technical field [0001] The invention relates to the field of medical assistance technology, in particular to a method for monitoring the overspeed ratio of colonoscope retraction based on a two-channel convolutional neural network. Background technique [0002] Colonoscopy is the most common examination method for detecting colorectal polyps, tumors and other lower gastrointestinal lesions. Colonoscopy withdrawal time refers to the actual time from when the colonoscopy reaches the cecum to when the colonoscopy is withdrawn to the anal canal, minus the time for additional operations such as dyeing examination or biopsy. Studies have shown that with the prolongation of withdrawal time, the polyp detection rate, adenoma detection rate and the average number of polyps found per subject increased significantly in colonoscopy patients. Therefore, the colonoscopy operation guidelines of various countries regard the withdrawal time as an important quality control index. American g...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T7/00G06N3/04G06N3/08
CPCG06T7/0012G06N3/08G06T2207/10024G06T2207/10068G06T2207/20081G06T2207/20084G06T2207/30028G06N3/048G06N3/045
Inventor 李超张阔刘奇为胡珊
Owner WUHAN ENDOANGEL MEDICAL TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com