Hybrid Retinal Image Registration

Chanwimaluang, Thitiporn; Fan, Guoliang; Fransen, Stephen

doi:10.1109/titb.2005.856859

Cited by 146 publications

(90 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Hybrid-based methods combine both feature-and intensity-based methods to overcome the drawbacks of both methods. Methods that combine vascular structures with entropy correlation coefficient [10] and spatial information in regional MI [11] and feature neighborhood MI [12] have been proposed. These methods use covariance matrices to reduce the data complexity instead of high-dimensional histograms, though as the spatial information increases, so commensurately does the corresponding computational cost.…”

Section: Introductionmentioning

confidence: 99%

Hybrid Retinal Image Registration Using Mutual Information and Salient Features

Ju¹,

Loew

Ku³

et al. 2016

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYThis paper presents a method for registering retinal images. Retinal image registration is crucial for the diagnoses and treatments of various eye conditions and diseases such as myopia and diabetic retinopathy. Retinal image registration is challenging because the images have non-uniform contrasts and intensity distributions, as well as having large homogeneous non-vascular regions. This paper provides a new retinal image registration method by effectively combining expectation maximization principal component analysis based mutual information (EMPCA-MI) with salient features. Experimental results show that our method is more efficient and robust than the conventional EMPCA-MI method.

show abstract

Section: Introductionmentioning

confidence: 99%

Hybrid Retinal Image Registration Using Mutual Information and Salient Features

Ju¹,

Loew

Ku³

et al. 2016

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…This type of approaches can usually obtain more accurate registration results, since the advantages of landmark-based and intensity-based registration can be combined. Different hybrid registration models have also been proposed recently [7,17,26,27,44].…”

Section: Introductionmentioning

confidence: 99%

Landmark-Matching Transformation with Large Deformation Via n-dimensional Quasi-conformal Maps

2015

View full text Add to dashboard Cite

We propose a new method to obtain landmark-matching transformations between n-dimensional Euclidean spaces with large deformations. Given a set of feature correspondences, our algorithm searches for an optimal folding-free mapping that satisfies the prescribed landmark constraints. The standard conformality distortion defined for mappings between 2-dimensional spaces is first generalized to the n-dimensional conformality distortion K ( f ) for a mapping f between n-dimensional Euclidean spaces (n ≥ 3). We then propose a variational model involving K ( f ) to tackle the landmark-matching problem in higher dimensional spaces. The generalized conformality term K ( f ) enforces the bijectivity of the optimized mapping and minimizes its local geometric distortions even with large deformations. Another challenge is the high computational cost of the proposed model. To tackle this, we have also proposed a numerical method to solve the optimization problem more efficiently. Alternating direction method with multiplier is applied to split the optimization problem into two subproblems. Preconditioned conjugate gradient method with multi-grid preconditioner is applied to solve one of the sub-problems, while a fixed-point iteration is proposed to solve another subproblem. Experiments have been carried out on both synthetic examples and lung CT images to compute the diffeomorphic landmark-matching transformation with different landmark constraints. Results show the efficacy of our proposed model to obtain a folding-free landmark-matching transformation between n-dimensional spaces with large deformations.

show abstract

“…Most common similarity measures in the context of multimodal RIR are mutual information (MI) [11,12], entropy correlation coefficient (ECC) [13], and phase correlation [14]. However, MI performance degrades when faced with a large amount of changes in the texture of retinal image and changes in scale [15].…”

Section: Introductionmentioning

confidence: 99%

“…MI is also weak in registering image pairs with too small overlaps. Therefore, ECC, a normalized measure of MI, was used on the vascular tree to register small overlapping images [13]. Its dependency on vessels restricts the efficiency of registration techniques for low-quality images.…”

Section: Introductionmentioning

confidence: 99%

An efficient approach for robust multimodal retinal image registration based on UR-SIFT features and PIIFD descriptors

et al. 2013

View full text Add to dashboard Cite

Existing algorithms based on scale invariant feature transform (SIFT) and Harris corners such as edge-driven dual-bootstrap iterative closest point and Harris-partial intensity invariant feature descriptor (PIIFD) respectivley have been shown to be robust in registering multimodal retinal images. However, they fail to register color retinal images with other modalities in the presence of large content or scale changes. Moreover, the approaches need preprocessing operations such as image resizing to do well. This restricts the application of image registration for further analysis such as change detection and image fusion. Motivated by the need for efficient registration of multimodal retinal image pairs, this paper introduces a novel integrated approach which exploits features of uniform robust scale invariant feature transform (UR-SIFT) and PIIFD. The approach is robust against low content contrast of color images and large content, appearance, and scale changes between color and other retinal image modalities like the fluorescein angiography. Due to low efficiency of standard SIFT detector for multimodal images, the UR-SIFT algorithm extracts high stable and distinctive features in the full distribution of location and scale in images. Then, feature points are adequate and repeatable. Moreover, the PIIFD descriptor is symmetric to contrast, which makes it suitable for robust multimodal image registration. After the UR-SIFT feature extraction and the PIIFD descriptor generation in images, an initial cross-matching process is performed and followed by a mismatch elimination algorithm. Our dataset consists of 120 pairs of multimodal retinal images. Experiment results show the outperformance of the UR-SIFT-PIIFD over the Harris-PIIFD and similar algorithms in terms of efficiency and positional accuracy.

show abstract

Hybrid Retinal Image Registration

Cited by 146 publications

References 43 publications

Hybrid Retinal Image Registration Using Mutual Information and Salient Features

Hybrid Retinal Image Registration Using Mutual Information and Salient Features

Landmark-Matching Transformation with Large Deformation Via n-dimensional Quasi-conformal Maps

An efficient approach for robust multimodal retinal image registration based on UR-SIFT features and PIIFD descriptors

Contact Info

Product

Resources

About