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

Retinopathy detection system based on lesion attention pyramid convolutional neural network

A convolutional neural network and retinopathy technology, applied in the field of medical image processing, can solve the problems of low detection accuracy of retinopathy, weak interpretability, inability to use multi-resolution image information, etc., and achieve good retinopathy diagnosis performance, The effect of improving accuracy

Active Publication Date: 2022-04-29
HARBIN INST OF TECH
View PDF6 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problem of weak interpretability in the detection of retinal lesions in the existing fundus images based on convolutional neural networks, and the inability to use multi-resolution image information, resulting in low detection accuracy of retinopathy, and proposes a method based on lesion attention Retinopathy detection system with pyramidal convolutional neural network

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
  • Retinopathy detection system based on lesion attention pyramid convolutional neural network
  • Retinopathy detection system based on lesion attention pyramid convolutional neural network
  • Retinopathy detection system based on lesion attention pyramid convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0027] Specific Embodiment 1: In this embodiment, the retinal lesion detection system based on the lesion-focused pyramid convolutional neural network includes:

[0028] Image processing main module, neural network main module, training main module, detection main module;

[0029] The image processing main module is used to collect original retinopathy images, preprocess the collected original retinopathy images, and obtain preprocessed original retinopathy images;

[0030] The main module of the neural network is used to build a lesion-focused pyramid convolutional neural network model;

[0031] The training main module uses the preprocessed original retinal lesion image to train the built lesion-focused pyramid convolutional neural network model, and obtains the trained lesion-focused pyramidal convolutional neural network model;

[0032] The detection main module is used to load a trained lesion-focused pyramid convolutional neural network model to classify images of retin...

specific Embodiment approach 2

[0033] Specific embodiment 2: The difference between this embodiment and specific embodiment 1 is that the image processing main module is used to collect the original retinopathy image, preprocess the collected original retinopathy image, and obtain the preprocessed original retinopathy image; the specific process is:

[0034] Step A1. Preprocessing the original retinopathy image to construct a multi-resolution image; the specific process is:

[0035] Obtain the original retinopathy image set, resize the resolution of each image into a set of multi-resolution images of 1024×1024, 512×512 and 256×256;

[0036] There are some background areas around the retinal images that do not affect the classification results, so each group of multi-resolution images of 1024×1024, 512×512 and 256×256 is cropped to 896×896, 448×448 and 224×224 at the center of the image Multi-resolution images (1024×1024 cropped to 896×896, 512×512 cropped to 448×448, 256×256 cropped to 224×224), each group...

specific Embodiment approach 3

[0043] Specific embodiment three: the difference between this embodiment and specific embodiment one or two is that the main module of the neural network is used to build a lesion-focused pyramid convolutional neural network model; the specific process is:

[0044] The lesion-focused pyramid convolutional neural network model includes a low-resolution network, a medium-resolution network, a high-resolution network, and a weakly supervised localization module;

[0045] B1. The low-resolution network includes a basic module 1, a basic module 3, a basic module 4, a basic module 5, a basic module 6 and a basic module 7 in sequence;

[0046] The medium-resolution network includes a basic module 1, a basic module 3 and a basic module 4 in sequence;

[0047] The high-resolution network sequentially includes a basic module 1, a basic module 2, a basic module 3, a basic module 4, a basic module 5, a basic module 6 and a basic module 7;

[0048] The basic module 1 sequentially includes...

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 retinopathy detection system based on a lesion attention pyramid convolutional neural network, and relates to a retinopathy detection system. The objective of the invention is to solve the problem of low retinopathy detection accuracy in the prior art. The system comprises an image processing main module used for collecting an original retinopathy image and preprocessing the collected original retinopathy image to obtain a preprocessed original retinopathy image; the neural network main module is used for building a lesion attention pyramid convolutional neural network model; the training main module uses the preprocessed original retinopathy image to train a built lesion attention pyramid convolutional neural network model to obtain a trained lesion attention pyramid convolutional neural network model; and the detection main module is used for loading the trained lesion attention pyramid convolutional neural network model and classifying retinopathy images to be tested. The method is applied to the field of medical image processing.

Description

technical field [0001] The invention belongs to the field of medical image processing, and in particular relates to a retinal lesion detection system. Background technique [0002] Diabetic retinopathy is an eye disease caused by complications of diabetes. It can cause retinal damage, vision loss and blindness and is currently the leading cause of visual impairment. Diabetic retinopathy is one of the most challenging diseases in the field of public health, and its diagnosis and classification are of great significance to the clinical treatment process. Early accurate classification can lead to more effective and targeted treatment to avoid visual impairment and further consequences. In the clinical process, doctors mainly classify through fundus images, which contain rich pathological information and pathological markers. Pathological biomarkers used for classification mainly include microaneurysm, hard exudate, soft exudate, and hemorrhage. Ophthalmologists classify dis...

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): G06T7/00G06V10/764G06V10/40G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06T7/0012G06N3/08G06T2207/20016G06T2207/30041G06N3/045G06F18/2415
Inventor 李翔罗浩王豪张九思乔新宇蒋宇辰尹珅
Owner HARBIN INST OF TECH
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