Joint Longitudinal Modeling of Brain Appearance in Multimodal MRI for the Characterization of Early Brain Developmental Processes - Longitudinal Registration and Shape Modeling - page 49

differences between contrast change trajectories of multiple modalities and have
the potential to deepen our understanding of neurodevelopmental processes.

2 Methodology

The proposed framework for joint multimodal modeling of appearance change in
MRI is outlined in Fig. 2 . The framework consists of four major components : (i)
Spatio-temporal image processing pipeline consisting of co-registration of entire
image data to a common coordinate space, segmentation of images into tissue
classes (gray, white matter, etc.) and parcellation of the brain into major cortical
regions; (ii) generation of white matter and gray matter intensity distributions
and quantification of regional contrast in terms of the Hellinger Distance (HD)
between these distributions; (iii) characterizing contrast direction using SIR and
adding a directional attribute to contrast using CONTDIR; and (iv) multivariate
NLME for joint modeling of multimodal appearance change.

Fig. 2. Complete framework for image-processing (left) and statistical analysis of con-
trast (right)

2.1 Spatio-Temporal Image Processing Pipeline

A joint registration-segmentation pipeline which removes variability due to fac-
tors such as shape, structure, and volume is implemented such that appearance
changes alone remain to be studied. This pipeline consists of intra subject regis-
tration using non-rigid free form deformations [ 11 ] followed by inter subject regis-
tration using LDDMM (Large Deformation Diffeophomorphic Metric Mapping)