Fundus image microhemangioma detection device and method thereof and storage medium

Abstract

The invention discloses a fundus image microhemangioma detection device and a method and a storage medium, and relates to the field of medical image processing and machine vision application. The method comprises the following steps: firstly, a to-be-detected image is extracted from an input color retina image, and then a micro-hemangioma candidate region template and positive and negative sample pictures are obtained through operations such as small target removal and geodesic expansion; energy features are extracted for the micro-hemangioma candidate area, features are manually designed for the micro-hemangioma form, and finally, traditional features and the manually designed features are cascaded; and the features and the labels are sent to a trained classifier for classification, so that the position of the micro-hemangioma is detected from the sugar net image. According to the method, the tiny target microhemangioma can be detected through the diabetes eye fundus image, the accuracy is high, and an eye fundus doctor can be assisted to observe the existence of the microhemangioma more conveniently.

Description

technical field [0001] The invention relates to a device, method and storage medium for detecting microvascular tumors in fundus images. Background technique [0002] Diabetic patients will develop retinal lesions in the late stage. Microaneurysm (MA) in fundus images is the initial symptom of diabetic retinopathy (sugar net). . Manual detection of microvascular tumors on fundus images mainly relies on ophthalmologists to directly observe retinal images. However, due to the complex structure of the retina, the small area of ​​microvascular tumors and low local contrast, it is time-consuming, laborious, and remote to observe microvascular tumors with human eyes. There is a lack of experienced ophthalmologists in the region, and the automatic detection of retinal microvascular tumors through computer vision technology will help relieve the pressure on ophthalmologists, and at the same time contribute to the sinking of medical resources, which has far-reaching medical signific...

AI Technical Summary

Problems solved by technology

These methods have not analyzed the characteristics of MA itself, so there are still characteristics of low accuracy and weak robustness.

Therefore, the current microvascular tumor detection method based on computer image processing technology still has problems such as low robustness, low detection accuracy, and difficulty in integrated use.

Images