Vessel‐based registration of fundus and optical coherence tomography projection images of retina using a quadratic registration model

Abstract

The new techniques of three-dimensional (3D)-optical coherence tomography (OCT) imaging is very useful for detecting retinal pathologic changes in various diseases and determining retinal thickness 'abnormalities'. Fundus colour images have been used for several years for detecting retinal abnormalities too. If the two image modalities were combined, the resulted image would be more informative. The first step to combine these two modalities is to register colour fundus images with an en face representation of OCT. In this study, curvelet transform is used to extract vessels for both modalities. Then the extracted vessels from two modalities are registered together in two stages. At first, images are registered using scaling and translation transformations. Then a quadratic transformation model is assumed between two pairs of images; because retina is imaged as a second-order surface. Twenty-two eyes (17 macular and 5 prepapillary), from random patients, were imaged in this study with Topcon 3D OCT1000 instrument. A new registration error is defined which averages the distance between all the corresponding points in two sets of vessels. Results show that registration error after stage one is 6.01 ± 1.82 pixels and after stage two is 1.02 ± 0.02 pixels.

Adjusted Fireworks Algorithm Applied to Retinal Image Registration

“…A method for automatic image registration through histogrambased image segmentation was proposed in [10], while geometrical features were used in [8]. In [9] curvelet transform was used to extract vessels and the extracted vessels from two modalities were registered together in two stages. In [7] a method based on the block matching algorithms was proposed.…”

Multimodal registration of SD-OCT volumes and fundus photographs using histograms of oriented gradients

“…Some of the previous works focused on stitching (mosaicing) images of the same modality with the aim of obtaining a broader field of view [12,17,20]. In addition, there are works that attempted to register multimodal retinal images including fluorescence angiogram and red-free fundus pairs [14,15,21], SLO and color fundus photographs [22,23], SD-OCT and color fundus photographs [7,22,[24][25][26]. The focus of the current work is on multimodal registration of fundus and SD-OCT modalities.…”

“…Hence, in order to benefit from the most dominant structural information that both modalities share (i.e. retinal blood vessels), the current color fundus and OCT registration methods [7,[24][25][26] include a vessel segmentation step as part of their algorithms. The retinal vasculatures are the best candidates for identifying the corresponding points (e.g.…”

Registration of SD-OCT en-face images with color fundus photographs based on local patch matching

“…However, to our best knowledge, there are not a lot of researches [4][5][6][7] on registration of SVPIs with…”

“…Li et al [4] recently presented an algorithm to register SVPIs with CFPs; this method utilized the Euclidean distance and normal direction difference to overcome the above problem. However, the algorithms [6,7] used the global similarity to register the SVPIs and CFPs. The registration accuracy is relatively low because these methods do not consider the influence of distortion.…”