Retinal effusion segmentation method and system based on semi-supervised deep learning network

A deep learning network and deep learning technology, applied in the field of retinal fluid segmentation based on a semi-supervised deep learning network, can solve problems such as limited segmentation ability, low accuracy, and high cost, so as to enhance generalization ability and improve learning ability , the effect of improving accuracy

Pending Publication Date: 2022-03-11
UNIV OF SHANGHAI FOR SCI & TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these deep learning algorithms require a large amount of labeled data for training, which is expensive
Moreover, most of these methods are limited to a single segmentation of IRF

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

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0042] Example one

[0043] Such as figure 1 As shown, a retinal fluid segmentation method based on semi-supervision deep learning network is provided in this embodiment;

[0044] S1. Labeling the retina OCT image, obtains the label data and the non-marking data, based on the label data and the non-marking data to build a retinal OCT database, based on the retinal OCT database division training data set and test data set;

[0045] Among them, the obtained annotation data and the wound data include:

[0046] The retinal OCT picture is labeled in the pixel grade, wherein the retinal OCT picture of the pixel grade annotation is constructed as labeling data, and the retinal OCT picture of the pixel grade label is constructed to be unmarked data;

[0047] The label data in the retina Oct database is divided into training data sets and test data sets.

[0048] In this embodiment, the 2814 sheet OCT image is acquired from 141 patients with different medical institutions. Two ophthalmologists

Example Embodiment

[0098] Example 2

[0099] Such as Figure 7 As shown in the present embodiment, a retinal fluid division system based on a semi-supervision depth learning network includes: acquisition module, first build module, a second build module, and training module;

[0100] Specific, acquisition module, first build module, the second build module, and the training module are sequentially connected, the acquisition module is used to mark the retina OCT picture, obtain the label data and the non-marking data, based on the label data and the non-marking data to construct a retinal OCT The database is based on the retinal OCT database division training data set and test data set; where the acquisition module also includes a label module; the label module is used to target the retinal OCT picture, wherein the retinal OCT image of the pixel grade label is constructed as a label Data, the retinal OCT picture of the pixel grade label is constructed as a non-label data; the first build module is used t

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 discloses a retina effusion segmentation method and system based on a semi-supervised deep learning network, and the method comprises the steps: carrying out the labeling of a retina OCT image, obtaining labeled data and unlabeled data, and dividing a training data set and a test data set based on the labeled data; establishing a retina effusion segmentation deep learning model; establishing a semi-supervised segmentation framework based on a random selection propagation strategy, obtaining retina OCT picture data with pseudo labels, and expanding a training data set; and training the retinal effusion segmentation deep learning model to obtain a retinal effusion segmentation neural network model. According to the method, a semi-supervised learning framework based on a random selection propagation strategy is adopted to relieve the shortage of high-quality pixel-level annotation OCT image data in an actual clinical environment, the learning ability of the Ref-Net is improved, the accuracy of segmentation of the retina effusion in the OCT image is remarkably improved, the number of images needing to be annotated for network training is reduced, and the network training efficiency is improved. And the generalization ability of the network is enhanced.

Description

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

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
Owner UNIV OF SHANGHAI FOR SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap