Intrasubject multimodal groupwise registration with the conditional template entropy

Abstract

Highlights• A novel similarity metric for multimodal groupwise registration is proposed.• The proposed method showed equivalent or improved registration accuracy.• Pairwise mutual information is outperformed in terms of transitivity. AbstractImage registration is an important task in medical image analysis. Whereas most methods are designed for the registration of two images (pairwise registration), there is an increasing interest in simultaneously aligning more than two images using groupwise registration. Multimodal registration in a groupwise setting remains difficult, due to the lack of generally applicable similarity metrics. In this work, a novel similarity metric for such groupwise registration problems is proposed. The metric calculates the sum of the conditional entropy between each image in the group and a representative template image constructed iteratively using principal component analysis. The proposed metric is validated in extensive experiments on synthetic and intrasubject clinical image data. These experiments showed equivalent or improved registration accuracy compared to other state-of-the-art (dis)similarity metrics and improved transformation consistency compared to pairwise mutual information.

Groupwise Non-Rigid Registration with Deep Learning: An Affordable Solution Applied to 2D Cardiac Cine MRI Reconstruction

“…This decision is made to avoid the bias that would stem from choosing beforehand one of the images in the sequence as the template [ 25 ]. For SSD, the optimum template is known to be the average of the registered images [ 1 ]. Hence, the SSD is defined as where is the set of transformations, N is the number of frames in the sequence, the n -th frame and is the transformation that maps each material point x in the template onto its corresponding position in the n -th frame.…”

“…As previously stated, the Template update block carries out two different operations, depending on the iteration number. For iterations this module performs a pointwise average of the registered images due to its optimality for SSD [ 1 ]. However, any other template construction policy could be accomplished for other types of similarity measures [ 8 ].…”

“…Image registration is the process of aligning two or more images for subsequent joint processing; it is especially widespread in the medical imaging field either for processing images acquired in different time instants—or with different imaging devices—for the same patient or to carry out comparative analyses with multiple patients [ 1 ]. Registration is a three-tier process, in which (a) a deformation model, (b) a measure of similarity between the entities to be aligned and (c) an optimization procedure, must be defined.…”

The tempest in a cubic millimeter: Image-based refinements necessitate the reconstruction of 3D microvasculature from a large series of damaged alternately-stained histological sections

“…Such co-registrations have clinical impact, e. g., [8,60,90]. A separate topic is a groupwise image registration, typically performed on MRI data, [11,14,15,44,45,86,111].…”

Deep Group-Wise Registration for Multi-Spectral Images From Fundus Images

“…Two recent works [15], [16] present effective strategies to address the problem of group-wise registration of multimodal images by a hierarchical intensity-space subdivision scheme, and the approach of conditional template entropy, respectively. A significant drawback is that the performance of these strategies relies on the intensity information from a collection of images with manually identified landmarks.…”