A groupwise mutual information metric for cost efficient selection of a suitable reference in cardiac computational atlas construction

Abstract

Computational atlases based on nonrigid registration have found much use in the medical imaging community. To avoid bias to any single element of the training set, there are two main approaches: using a (random) subject to serve as an initial reference and posteriorly removing bias, and a true groupwise registration with a constraint of zero average transformation for direct computation of the atlas. Major drawbacks are the possible selection of an outlier on one side, and an initialization with an invalid instance on the other. In both cases there is great potential for affecting registration performance, and producing a final average image in which the structure of interest deviates from the central anatomy of the population under study. We propose an inexpensive means of reference selection based on a groupwise correspondence measure, which avoids the selection of an outlier and is independent from the atlas construction approach that follows. Thus, it improves tractability of reference selection and robustness of automated atlas construction. We illustrate the method using a set of 20 cardiac multislice computed tomography volumes.

A High-Resolution Atlas and Statistical Model of the Human Heart From Multislice CT

“…The solid lines (connected to the associated dash-dot lines using dotted lines) illustrate the mean ranks for these subjects. It shows that high and low ranks are relatively stable [29].…”

“…Using pairwise registrations starts out from a single reference image. Typically this one is chosen from the population at hand, either by an expert [27], or by an automated method; this can be based on subspace exploration [28] or on an information-theoretical basis [29].…”

“…1 is therefore an attempt to select the reference already close to the unknown mean. For this we use a heuristic based on a Groupwise Mutual Information (GWMI) score, an extension of Mutual Information [55] that describes the amount of information between a single image and a set of images [29]. We compute the GWMI score over affinely registered volumes.…”

Atlas-Based Quantification of Myocardial Motion Abnormalities: Added-Value for Understanding the Effect of Cardiac Resynchronization Therapy

“…The velocities extracted from each sequence were then spatiotemporally aligned to a common reference anatomy, chosen among the set of healthy volunteers using the group-wise normalized mutual information metric (Hoogendoorn et al 2010). Temporal alignment was based on piecewise linear warping of the timescale based on the timing of physiological events (onset of QRS complex [beginning and end of the cycle], located on the ECG using tools from the EchoPac software [GE Healthcare, Milwaukee, WI, USA], and the aortic valve opening and closure, determined using continuous wave Doppler imaging on the aortic valve).…”

Atlas-Based Quantification of Myocardial Motion Abnormalities: Added-value for the Understanding of CRT Outcome?

“…2.1. The choice of a reference among the set of healthy volunteers was addressed using the group-wise normalized mutual information metric proposed in [16], and criteria based on image quality (LV fully visible along the whole sequence, and low heart rate to achieve a higher temporal resolution of the atlas). Statistics on myocardial velocities were computed locally in time and space, at every anatomical location (x, t) of the reference anatomy.…”