, Daniel Valentino
, Guogang Hu, Jenaro Felix, Michael Mega
, Allan MacKenzie-Graham, Seth Riffins¶, David Rex, Arthur TogaLONI, Department of Neurology, UCLA School of Medicine, Los Angeles, CA 90095, USA.
Department of Radiology, UCLA School of Medicine, Los Angeles, CA 90095, USA.
Alzheimers Disease Research Center, UCLA School of Medicine, Los Angeles, CA 90095, USA.
¶Department of Biology, California Institute of Technology, Pasadena, CA 91125, USA.
Subject: Modeling & Analysis
Abstract
Many state-of-the-art neuroanatomical labeling schemes differ significantly in their hierarchical nomenclature organization. There are developmental cephalic organizations where the brain is separated, or tessellated, into anatomically disjoint regions based on the cellular lineage. There also exist neuro-labeling approaches that systematically organize the hierarchy of structures based on cytoarchitectonic, functional or chemo-architechtonic connectivity. Moreover, there are variations within each of these schemes, variations in naming between different investigators and in different studies. For example, an investigator devising a study that uses the Paxinos labeling scheme [Paxinos and Watson, 1998] may desire cross-validation with studies utilizing Swanson hierarchical nomenclature [Swanson, 1998].
Previously we had developed a tree-based hierarchical data structure, the LONI BrainTree, to address the need for linking anatomical, functional and contextual neuroscientific information. The BrainTree approach was successfully used in conducting both volumetric studies [Crabtree, et al., 2000] and functional activation studies [Dinov, et al., 2001] in Alzheimer’s disease neuroimaging data.
The BrainTree data model introduced: Graphical and interactive organization of brain anatomy; A common coordinate-to-label reference frame; Linking neuroimaging, neurogenetic, neuropsychiatric and contextual data; Hierarchical representation of neuroanatomical names in an accessible and user friendly manner; Region anatomical location (containment by, and of, other neighboring regions; An axtremely conceptually and computationally attractive data representation.
In the original BrainTree, we used a tree-based, relational data model because storing, retrieving and manipulating Tree structures is well understood [Tinhofer, 1990] and computationally tractable. We found that the tree structure was limited because in many situations different paths exist between two remote structures that are not necessary descendants of each other.
To address these limitations we extended the BrainTree as a general, flexible, graph-based data model, The BrainGraph, that integrates, organizes and provides direct access to external structural, functional, histological, genetic and contextual brain information. Since pure hierarchical organizations based on anatomical containment are insufficient to represent complex circular, dynamic and study-specific interrelations between different regions in the brain,
the BrainGraph is required to provide this functionality. It allows simultaneous storage of multiple ROI labeling schemes; Provides study-specific graph traversal schemes; Each node (ROI) and each edge (connection link) has a number of predefined (or user specifiable) description categories (e.g., functional connectivity, anatomical relations to its neighbors, developmental information, genetic information, literature references and other external contextual information).
References:
[Paxinos and Watson, 1998] Paxinos, G, Watson, C. The Rat Brain in Stereotaxic Coordinates. Academic Press. 1998.
[Swanson, 1998] Swanson, LW. Brain Maps: Structure of the Rat Brain 2nd Edition. Amsterdam, Elsvier Science Publishers BV. 1998.
[Crabtree, et. al., 2000] Crabtree EC, Mega MS, Linshield C, Dinov ID, Thompson PM, Felix J, Cummings JL, Toga AW. Alzheimer grey matter loss across time: unbiased assessment using a probabilistic Alzheimer brain atlas. Soc. for Neurosci. Abs. 2000, 26:294.
[Dinov, et al., 2001] Dinov ID, Mega MS, Manese, M, Felix, J, Tran, N, Lindshield, C, Cummings, J, & Toga AW. "Construction of the First Rest-State Functional Sub-Volume Probabilistic Atlas of Normal Variability in the Elderly and Demented Brain", Neurology; Abs., Philadelphia, PA, May 06-11, 2001.
