Target recognition system and method for fundus colorful imaging

Abstract

The invention discloses a target recognition system for fundus colorful imaging. The target recognition system comprises a cloud server and a mobile client PowerRealInfo, the fundus colorful image imaging device is used for collecting images, and the made fundus colorful image serves as a model image to be input. The cloud server comprises a model pre-training module and a model prediction module; in the server, a model pre-training module is connected with a background control end through a network, a model prediction module is also connected with the background control end through the network, and connection between the model pre-training module and the model prediction module is controlled through a background; the mobile client comprises a result display processing module; the mobile client is connected with the model prediction module in the cloud server through a network; and the model prediction module is connected with the result display module. The method is used for realizing high-performance eye fundus image target recognition and helping medical personnel to quickly check the states of various focus points, and the accuracy is high.

Description

technical field [0001] The present invention relates to a fundus image lesion recognition system, and more particularly to a target recognition system for colorful fundus imaging. In addition, the present invention also relates to a method for realizing a target recognition system of colorful fundus imaging. Background technique [0002] The present invention is mainly based on the background of two aspects. One is that diabetic retinopathy (diabetic retinopathy, DR) is the fastest growing cause of blindness in the world at present, and it is one of the complications of diabetic patients. The second is the development of deep learning in the field of image processing, especially in medical image annotation systems, where deep learning is the most widely used and the technology development is the most mature. [0003] On the one hand, diabetic retinopathy is asymptomatic until irreversible visual damage occurs. At present, the main means of preventing blindness is regular i...

AI Technical Summary

Problems solved by technology

[0004] There are some AI systems based on fundus photography in the existing technology. However, due to the imperfect standards of medical AI, there is no experience to refer to, and no actual products have appeared. In this case, how to achieve the goal of high-performance fundus images Identifying and helping medical staff to quickly view the status of various lesions is the main difficulty currently facing

[0005] Compared with the similar patent of patent 2017106632189 "Lesion Recognition System Based on Fundus Image", this patent uses its own result judgment function among many deep learning models to select the most reasonable model through various comparisons, and It will automatically adjust the adaptive parameters and models according to the actual situation, and there is no need to manually change and adjust the model parameters according to the actual situation; compared with the patent 201711439401, the patent 201711439401 uses a fixed judgment device, which connects each module through hardware (or It installs the entire system on one device), although this method can achieve fast transmission speed and stable transmission, but due to the design of too many hardware connections, it has a high dependence on system stability, and the maintenance cost is relatively high. The way to deploy to the server cloud is higher and more cumbersome

And because all systems are carried on hardware devices, the mobility and scalability are relatively restricted.

Images