Convolutional neural network weight optimization method for retinal lesion classification

A convolutional neural network and retinopathy technology, applied in the field of medical information intelligent processing, can solve the problem of gradient descent algorithm falling into local optimal solution, etc., to achieve the effect of improving detection accuracy, improving execution efficiency, and reducing initial loss value

Active Publication Date: 2020-03-27
NANTONG UNIVERSITY
View PDF2 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the complexity of fundus images, the use of traditional convolutional neural networks for multi-label classification of fundus images can easily cause the gradient descent algorithm to fall into a local optimal solution.

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
  • Convolutional neural network weight optimization method for retinal lesion classification
  • Convolutional neural network weight optimization method for retinal lesion classification
  • Convolutional neural network weight optimization method for retinal lesion classification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] The present invention will be described in further detail below in conjunction with the accompanying drawings.

[0044] Such as figure 1 Described, a convolutional neural network weight optimization method for classification of retinal lesions, comprising the following steps:

[0045] Step 1, input the fundus image training set and label, the training set is X=(x 1 , x 2 ,...,x n ), n=1, 2, 3..., labeled as B=(b 1 , b 2 ,...,bn ), n=1, 2, 3..., the label b corresponding to the fundus image i Perform one-hot encoding to get the real value y_true i ;

[0046] Step 2. Initialize the weight parameters of the convolutional neural network, use the standard normal distribution to generate m frogs, sort them by fitness value, and find the optimal frog f b and worst frog f w , continuously update the position of the worst frog and reorder until the single-population leapfrog algorithm meets the convergence conditions, and the global optimal frog f is obtained q , put f...

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 field of medical information intelligent processing, in particular to a convolutional neural network weight optimization method for retinal lesion classification. The method comprises the following steps: firstly, acquiring a fundus image training set and a corresponding multi-lesion label; searching an optimal initial weight value through a single-population leapfrogalgorithm, then constructing a convolution layer, a pooling layer and a full connection layer in the convolutional neural network, and taking the optimal initial weight value as a parameter of first forward propagation calculation; performing cross entropy loss calculation and summation on the four predicted values of the four lesions in the retina and the true value to obtain a loss value, judging whether the loss value is abnormal or not, if the loss value is abnormal, generating a frog group around the weight of the previous forward propagation, and searching an optimal frog updating network weight; otherwise, updating the network weight by adopting a gradient descent algorithm; and finally, optimizing the final weight. According to the method, the accuracy of fundus image multi-lesiondetection can be effectively improved, and the method has high application value for retinal diseases and adjuvant therapy.

Description

technical field [0001] The invention relates to the field of intelligent processing of medical information, in particular to a convolutional neural network weight optimization method for retinal lesion classification. Background technique [0002] Color fundus images are the most basic examination method for diagnosing ophthalmic diseases. At the same time, fundus images can enable people to detect various eye diseases as early as possible, such as glaucoma, optic neuritis, macular degeneration, etc., which is convenient for timely and effective treatment. Early diagnosis and timely treatment can effectively reduce the prevalence. However, due to the large population in China and the relatively limited number of ophthalmologists, it takes a lot of time to rely solely on doctors to diagnose eye diseases, so other methods are urgently needed for large-scale screening. Computer-aided diagnosis can not only greatly reduce the workload of doctors, but also has the advantages of...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/00G06T7/00
CPCG06N3/006G06T7/0012G06T2207/10024G06T2207/20081G06T2207/30041G06N3/045G06F18/241
Inventor 丁卫平任龙杰孙颖鞠恒荣丁帅荣曹金鑫张毅冯志豪李铭文万志胡彬赵理莉
Owner NANTONG UNIVERSITY
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