Multi-subject Registration for Unbiased Statistical Atlas Construction

Abstract

Abstract. This paper introduces a new similarity measure designed to bring a population of segmented subjects into alignment in a common coordinate system. Our metric aligns each subject with a hidden probabilistic model of the common spatial distribution of anatomical tissues, estimated using STAPLE. Our approach does not require the selection of a subject of the population as a "target subject", nor the identification of "stable" landmarks across subjects. Rather, the approach determines automatically from the data what the most consistent alignment of the joint data is, subject to the particular transformation family used to align the subjects. The computational cost of joint simultaneous registration of the population of subjects is small due to the use of an efficient gradient estimate used to solve the optimization transform aligning each subject. The efficacy of the approach in constructing an unbiased statistical atlas was demonstrated by carrying out joint alignment of 20 segmentations of MRI of healthy preterm infants, using an affine transformation model and a FEM volumetric tetrahedral mesh transformation model.

Atlas-based delineation of lymph node levels in head and neck computed tomography images

“…In this context, we have chosen not to use the method proposed in [25] because it relies on the registration of manual segmentations without taking into account the CT images. However, in our context, the structures of interest do not cover all of the CT images; hence, the computed transformations outside of these regions will be indeterminate, leading to errors when building the mean CT image.…”

“…Finally, de Craene et al [25] proposed recently a method to build the mean image and compute the mean segmentations together. This is achieved using an EM algorithm which alternates two steps.…”

Efficient Selection of the Most Similar Image in a Database for Critical Structures Segmentation

“…The contours and their variability could then be introduced in the building process to better constrain the atlas creation. Previous work in this direction include the work of [8], where the binary structures were included in the atlas formation process.…”

“…Lorenzen et al [7] improved this framework to cope with large deformations. De Craene et al [8] proposed a coupled estimation of the average segmentations and the average image. Finally, Grabner et al [9] presented an interesting method to compute directly an average symmetric image.…”

“…As in that paper, we did not use the method of [8] because of the variability of the manual segmentations in the dataset. The construction is composed of three main steps.…”