Multi-modal retina image fusion method and system based on image registration

Abstract

The invention provides a multi-modal retina image fusion method and system based on image registration. The method comprises the steps of obtaining a retina image pair; preprocessing, obtaining retina edge image pairs, and extracting a feature point set from each retina edge image pair; combining a plurality of feature descriptors to construct a multi-feature difference descriptor, and guiding feature extraction of the feature point set; updating spatial transformation of the source feature point set according to the corresponding relation matrix; performing image registration, and obtaining an image after spatial transformation in combination with the target image; and fusing the image after spatial transformation and a preset reference image to obtain a fused image. According to the method, the defects of insufficient image registration precision and inaccurate image space transformation in a multi-modal retina image fusion process are overcome, and high-precision and high-stability registration performance can be realized. On the basis of high-precision image registration, effective multi-modal eye fundus image fusion is carried out, so that an ophthalmologist can conveniently observe the change of a lesion area.

Description

technical field [0001] The present invention relates to the field of image processing, in particular, to a multimodal retinal image fusion method and system based on image registration. Background technique [0002] Fundus retinal images are an important basis for the diagnosis of various retinal diseases including glaucoma and age-related macular degeneration. In addition, the fundus, as the only direct-viewing vascular window in the human body, can reflect the hemodynamic changes of other organs in the whole body to a certain extent. Multimodal retinal images usually contain important retinal local structure and temporal information. If the retinal images of different modalities of the same patient can be fused, it can provide doctors with a variety of complementary information on the lesion, thereby providing a more comprehensive and clear basis for doctors to make decisions about diagnosis and treatment. [0003] Due to the complex structure of the vascular network in ...

AI Technical Summary

Problems solved by technology

First, due to the specificity of retinal images (repetitive structure, multimodality), inaccurate feature extraction and description can generate a large number of false feature matches

In addition, in the process of image space transformation, the above research does not use constraints or only uses a single constraint method, which will lead to inaccurate image space transformation

These issues will affect the robustness and accuracy of multimodal retinal image fusion based on image registration

[0006] In summary, the existing retinal image registration methods generally suffer from inaccurate feature extraction and description, which lead to a large number of erroneous feature matches, and the constraint method leads to inaccurate transformation of image controls, which in turn affects the robustness and accuracy of multimodal retinal image fusion. precision

Images