Elastic Registration of Brain Maps to a 3D Digital Mouse Atlas

Elastic Warping. For many years we have been developing software for elastic image registration. This software allows researchers to transform digital brain maps into the coordinate system of a brain atlas. Warping is applied to the maps to match anatomical boundaries in the brain maps with those in the atlas. These maps can then be compared and averaged across expression, to identify systematic patterns of molecular content and gene expression. This page explains the alignment procedure we developed to align genetic and histologic maps of the mouse into a standardized atlas coordinate system. [Cryosection and Nissl Images by Allan Mackenzie-Graham].

Continuum-Mechanical Matching of Nissl Stained Sections. Raw cryosection images lack the anatomical detail necessary to define individual nuclei and subtle cellular boundaries in the brain. Nissl stained tissue sections (left) have excellent anatomical detail but are distorted relative to the cryosection blockface (right). Placement of anchor curves on corresponding anatomical boundaries in each image allows us to compute a mapping that reconfigures the Nissl image into its original blockface configuration. The algorithm uses a continuum mechanical formulation to find the mapping with minimal elastic energy that guides the distorted section back into its blockface configuration. [Cryosection and Nissl Images by Allan Mackenzie-Graham; Java Graphical User Interface by Lynn Thompson].

Elastic Reconfiguration of Nissl Image. Deforming the section according to the principles of continuum mechanics requires the solution of 1,000,000 simultaneous elliptic partial differential equations. Note the complex pattern of warping applied at a very local level, ensuring the exact matching of anatomical boundaries, across the section, with their counterparts in the cryosection blockface. [Nissl Images by Allan Mackenzie-Graham].

3D Elastic Warping to 3D Mouse MRI. To reconstruct the anatomical and histologic maps into their in vivo configuration, an additional 3D surface based warp is required. This takes the 3D cryosection image, and any maps co-registered with it, into the anatomical configuration observed in vivo using 3D micro-MRI. Details of the elastic matching procedure can be found here. [Mouse micro-MRI Images by Russ Jacobs and his colleagues at CalTech].

Related Publications

Review of Elastic Warping Algorithms and their Applications

Mathematical Principles of Our Algorithm

Introduction to Image Warping

Tutorial on Brain Image Analysis

Article about Brain Atlases

(back to main list)

Contact Information