Average Brain Models: A Convergence Study

Abstract

We present a completely automatic method to build stable average anatomical models of the human brain using a set of magnetic resonance (MR) images. The models computed present two important characteristics: an average intensity and an average shape, both in a single image. We provide results showing convergence toward the centroid of the image set used for the computation of the model. In particular, the RMS distances between the model and the MR images contained in the set stabilize in a range of 2.88 to 3.36 mm from a range of 4.62 to 5.51 mm initially after only one iteration. As for the influence of the reference image chosen for the model construction, this is minimal with differences of about 1.0 mm, from approximately 3.5 mm initially. These results ensure the usefulness of our approach.

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

“…This rst step is performed using the method developed by Guimond et al [19]. It has the advantage of being faster and simpler than the one proposed by Lorenzen et al [21] and is su cient for our images.…”

“…As the built atlas is unbiased, any image of the database can be selected as R 0 . More details can be found in [19].…”

“…They iterate on two steps: the registration of the images on the reference and the application of the inverse mean transformation to the mean image. Guimond et al [19] have shown that this approach, extended by [20,21] to transformations containing large deformations, is not dependent on the choice of the reference image. Recently, based on this principle, a method has been investigated to generate directly a mean symmetric atlas from a database of images [22].…”

Spatial normalization of diffusion tensor MRI using multiple channels

“…Spatial normalization involves the registration of images and the generation of a stereotaxic atlas that represents the statistical distribution of the group at each voxel (Friston et al, 1995;Mazziotta et al, 1995;Thompson and Toga, 1997;Grenander and Miller, 1998;Guimond et al, 2000). The registration of diffusion tensor images has generally been performed in a similar way to the registration of T1-weighted or SPGR MR images.…”

Detection of diffuse abnormal perfusion in SPECT using a normal brain atlas

“…However, its construction neither uses PCA (Houston et al 1998) nor presegmented anatomical regions (Kovalev et al, 1999;Pagani et al, 2003) but a methodology we have previously developed for MRI anatomical atlas construction (Guimond et al, 1999). This automatic procedure builds a stable average anatomical model of the human brain containing two important features: an average intensity (with the normal variations for each voxel) and an average shape (with the normal shape deformation as an x, y, and z covariance matrix for each voxel).…”