Locally Normalized Gradient Fields for Multi-Modal Groupwise Image Registration
The Normalized Gradient Fields (NGF) similarity measure has shown its dominance for registering pairs of images from numerous modalities. Recently, gradient fields-based distance measures for groupwise image registration have been proposed, but these works have shown its usefulness only for mono-modal applications or dynamic imaging from the same modality. In this paper, we propose the groupwise Normalized Gradient Fields (gNGF) similarity measure for groupwise multi-modal image registration. The normalization of gradient fields is one of the key components of the NGF measure where the normalization may be computed locally or globally. In this work, we motivate the local normalization for multi-modal registration where the correlations between gradient fields are maximized at every point in the spatial domain for registering images. We demonstrate the effectiveness and applicability to real-world data of our proposal through experiments on images from two datasets. The results show that gNGF may be a long missing extension of NGF for multi-modal, multi-image applications.
Code (0)
등록된 구현이 없습니다.
Tasks
Image RegistrationSimilar Papers 제목 키워드 기반
Beyond Convexity: Stochastic Quasi-Convex Optimization
Stochastic convex optimization is a basic and well studied primitive in machine learning. It is well known that convex and Lipschitz functions can be minimized efficiently using Stochastic Gradient Descent (SGD). The Nor…
Cross-Sim-NGF: FFT-Based Global Rigid Multimodal Alignment of Image Volumes using Normalized Gradient Fields
Multimodal image alignment involves finding spatial correspondences between volumes varying in appearance and structure. Automated alignment methods are often based on local optimization that can be highly sensitive to t…
global-optimizationCross-Sim-NGF: FFT-Based Global Rigid Multimodal Alignment of Image Volumes using Normalized Gradient Fields
Multimodal image alignment involves finding spatial correspondences between volumes varying in appearance and structure. Automated alignment methods are often based on local optimization that can be highly sensitive to i…
Relative Entropy Gradient Sampler for Unnormalized Distributions
We propose a relative entropy gradient sampler (REGS) for sampling from unnormalized distributions. REGS is a particle method that seeks a sequence of simple nonlinear transforms iteratively pushing the initial samples f…
Regularized directional representations for medical image registration
In image registration, many efforts have been devoted to the development of alternatives to the popular normalized mutual information criterion. Concurrently to these efforts, an increasing number of works have demonstra…
DiversityImage RegistrationMedical Image Registration