Segmentation of the hippocampus from MR images is of particular importance in the study of epilepsy [1] or neuro-degenerative diseases such as dementia of the Alzheimer type (DAT) [2], where hippocampal atrophy has been positively correlated with various stages of the aforementioned pathologies. Volumetric information can be extracted once the structure has been segmented, a tedious task if done manually or with methods requiring manual intervention. Fast segmentation techniques also become critical in the case of prospective or retrospective studies involving large ensembles of data volumes.
Segmentation methods of brain MR images typically employ manual and/or automatic knowledge-based models specific to the structure of interest (SOI). The technique presented here overcomes some of the limitations of current methods. Specifically, it (1) requires no manual intervention, (2) is fast, (3) is fully 3D, and (4) can be applied to any SOI yet remains constrained by some form of structure information. The novelty of this work resides in its a priori Principal Components Analysis (PCA) of non-linear registration data of a volume of interest (VOI), represented by dense 3D deformation fields from ANIMAL [3]. The results are used in an Appearance Model, inspired by Cootes [4], able to segment any SOIs contained within the VOI, in the atlas-independent framework described by Collins [3]. The research presented here focuses on non-supervised segmentation of the hippocampus in a population of normal subjects. Training set data came from the International Consortium for Brain Mapping (ICBM) database, and the resulting model was used to segment the left hippocampus of control subjects, extracted from the same database. These data were also part of a study on aging at the Montreal Neurological Institute, in which a neuro-anatomical expert manually segmented the left and right hippocampi [5].
This paper presents the theoretical basis for segmentation and initial results using this method. Validation against manual segmentation and comparison with ANIMAL are also presented. Its applicability towards shape deformation analysis of neuro-degenerative diseases is discussed.