Microaneurysm detection method in fundus images based on phase consistency

Abstract

The invention relates to a method for detecting eye fundus image MA based on phase equalization. The method comprises the steps that firstly, a green channel of a colorful eye fundus image is extracted; then, feature points are extracted based on a phase equalization model, a candidate with the MA is acquired by combining a threshold segmentation and area filtering method; finally, blood vessels and other unrelated information in the image are removed through the method of constructing a gray level profile map, and therefore the true MA is screened out. The advantages of the phase equalization model are fully used, insensitivity on the brightness and the contrast ratio of the image is achieved, enhancement processing does not need to be carried out on an original image, and robustness is good; the blood vessels and other unrelated information are removed through the method of combining the gray level profile map, classification and identification are not involved, the complexity of an algorithm is greatly reduced, operating time is saved, and the MA on the colorful eye fundus image can be efficiently and automatically detected.

Description

technical field [0001] The invention relates to a method for detecting microaneurysms in fundus images based on phase consistency. The method can accurately detect fundus microaneurysms for fundus images with different image qualities, belongs to the technical field of image processing, and can be applied to medical diagnosis. Background technique [0002] Medical image processing and analysis has always been a focus and hot issue in the field of image processing and analysis. With the powerful means of graphics and image technology, the quality and display methods of medical images have been greatly improved, which has greatly improved the level of diagnosis and treatment. Image processing technology has been introduced into ophthalmology for many years. Through the calculation and analysis of fundus images, it can quantitatively measure important fundus tissues such as optic disc, retinal blood vessels, microaneurysms and macular fovea, and make a clear distinction between...

AI Technical Summary

Problems solved by technology

Although the existing methods realize the automatic detection of MA under certain conditions, there are problems such as high false detection rate, high missed detection rate, and complex calculation.

The fundamental reason is that these methods mainly use gradient information to describe the image. Since the gradient information will change with the change of the contrast and brightness of the image, it is very sensitive to the noise in the image, so these methods have higher requirements on image quality.

Due to factors such as uneven illumination of fundus images, weak contrast, and small size of MA itself, it is difficult to exclude a large number of non-MA points by the processing method based on gradient information, and further classification based on regional feature information often increases the complexity of the system

Therefore, for fundus images with complex backgrounds, it is difficult to achieve ideal detection results based on gradient information processing methods

Images