A method and system for eye disease image recognition based on improved d-s evidence

Abstract

The invention discloses an eye disease image recognition method and system based on improved D-S evidence, collecting fundus images; performing data enhancement on the fundus images to obtain enhanced images; and classifying the enhanced images through the first transfer network and the second transfer network respectively Obtain the first classification result and the second classification result; use the fusion model to fuse the first classification result and the second classification result of the enhanced image to obtain the final classification result; unify the size of the fundus image, appropriately reduce the resolution, and remove redundancy , useless areas, reduce the amount of calculations, and reduce the calculation time; by changing the order of RGB channels to enhance offline data, reduce the risk of over-fitting deep neural networks; improve the D-S evidence theory to eliminate four common paradoxes Theory, and use the improved D-S evidence theory to fuse the two deep neural network models, so that the fused network reduces the inherent deviation between the models, and improves the robustness and accuracy.

Description

technical field [0001] The disclosure belongs to the field of deep learning technology and artificial intelligence technology, and specifically relates to an eye disease image recognition method and system based on improved D-S evidence. Background technique [0002] With the rapid development of medical imaging technology and computer vision, fundus screening methods have high accuracy and precision, but actually rely on expensive equipment and complicated operations. In addition, eye diseases are always silent and irreversible , only early detection and early treatment can save vision from the invisible infestation of vision thieves, so cheap and fast early fundus screening has become a research hotspot in academia and industry. [0003] Deep learning has become a research hotspot in medical aided diagnosis. It has powerful automatic feature extraction, feature selection, feature expression and complex model construction capabilities. More importantly, deep learning can ex...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to propose a method for eye disease image recognition based on improved D-S evidence, to solve one or more technical problems in the prior art, at least provide a beneficial option or create conditions

Images