Elsevier

NeuroImage

Volume 42, Issue 1, 1 August 2008, Pages 60-69
NeuroImage

High resolution three-dimensional brain atlas using an average magnetic resonance image of 40 adult C57Bl/6J mice

https://doi.org/10.1016/j.neuroimage.2008.03.037Get rights and content

Abstract

Detailed anatomical atlases can provide considerable interpretive power in studies of both human and rodent neuroanatomy. Here we describe a three-dimensional atlas of the mouse brain, manually segmented into 62 structures, based on an average of 32 μm isotropic resolution T2-weighted, within skull images of forty 12 week old C57Bl/6J mice, scanned on a 7 T scanner. Individual scans were normalized, registered, and averaged into one volume. Structures within the cerebrum, cerebellum, and brainstem were painted on each slice of the average MR image while using simultaneous viewing of the coronal, sagittal and horizontal orientations. The final product, which will be freely available to the research community, provides the most detailed MR-based, three-dimensional neuroanatomical atlas of the whole brain yet created. The atlas is furthermore accompanied by ancillary detailed descriptions of boundaries for each structure and provides high quality neuroanatomical details pertinent to MR studies using mouse models in research.

Introduction

High resolution magnetic resonance imaging (MRI) has recently seen increased use in mouse phenotyping studies (Redwine et al., 2003, Bock et al., 2006, Badea et al., 2007b) as reviewed in Nieman et al., 2005, Anderson and Frank, 2007. MRI provides detailed neuroanatomical information covering the whole brain and, in combination with advanced image processing, can rapidly localize regions of the cerebrum differing by genotype. The use of structural segmentations can enhance analyses by (1) aiding in localizing statistical peaks and (2) providing tissue volumes for those structures. Creating a detailed three-dimensional atlas of the brain from MRI is, however, not a trivial task, requiring detailed understanding of the underlying anatomy along with high-resolution MR scans upon which to base the segmentation.

To date, a few structural murine brain atlases have been created with the aid of MRI on neonatal and postnatal mice; each study using differing methodologies and characteristics. Table 1 summarizes the studies conducted previously in creating mouse brain atlases and gives a concise comparison of our present work with those of other investigators. Using this table of previous murine brain MR studies we can conclude the following: (1) higher resolution and signal to noise ratio (SNR), as should be obvious, allow a greater number of anatomical details to be discerned; (2) fixed brain MRI provides greater detail than in-vivo as reviewed in Benveniste and Blackband (2006); (3) removing brains from the skull for ex-vivo scanning distorts the underlying anatomy (Kovacevic et al., 2005, Ma et al., 2005); and (4), using a model independent average of a population of mice rather than one individual allows for the capture of within strain variance and provides greater SNR (Kovacevic et al., 2005). These points are elaborated upon in the Discussion section of this paper.

The present study provides an atlas, based on the commonly used C57Bl/6J mouse, which complements and improves on previous works in this field. The novelty of this work is represented by a manually segmented atlas based on 32 μm3 resolution images created from average MRI of 40 mouse brains: 20 male and 20 female at 12 weeks in age. A total of 62 brain structures which could be clearly visualized at this resolution was manually traced on each slice. These, to the best of our knowledge, are the most number of neuroanatomical structures manually segmented on a mouse average MRI. We show three-dimensional representations of complex brain relationships that can be directly visualized using our atlas. This atlas will be freely available (http://www.mouseimaging.ca/research/mouse_atlas.html) to the research community and the Supplementary data of this paper will provide the users with the detailed descriptions of the boundaries for each structure.

Section snippets

Materials and methods

The mice used in this study were chosen to be 12 week old since this is the most common age group used in behavioural and neurobiological research. The brains were scanned post-mortem within the skull, keeping the brain as close to the normal shape as would be found in-vivo. Post-mortem scanning yielded better contrast and image resolution compared to in-vivo methods which are prone to motion artifacts due to breathing and cardiac movements. It also allows for closer placement of MR coils and

Results

The process described above resulted in a neuroanatomical atlas consisting of 62 structures (Table 2), which will be freely available to the research community. Based on the average mouse atlas presented in this paper, the volume for each individual mouse brain was computed by integrating over the Jacobian determinants for each voxel that belongs to a structure. This process is equivalent to resampling the atlas towards each mouse brain in native MRI space and getting the volume of the

Discussion

The atlas constitutes 62 structures that were clear to view at the resolution of 32 μm, with the smallest structure being the habenular commissure. Some structures which are described by their individual sub-components in histological atlases (Hof et al., 2000, Paxinos and Franklin, 2001, Swanson, 2004, Lein et al., 2007, Dong, 2008), but were indecipherable at this resolution are presented as aggregates, such as the amygdala, the thalamus and the hypothalamus. Similarly, the histological

Conclusion

The atlas created here is intended for general neuroanatomical guidance; in other words to assist in finding structure locations, understanding three-dimensional relationships in mouse neuroanatomy, and for structural segmentation techniques. It could be useful to co-register the atlas to other mouse brains for aid in segmentation when needed; however, some adjustments may need to be made to the final product as the labels may not fit perfectly, especially when registered to individual brains

Acknowledgments

The authors would like to thank the Sunnybrook Health Sciences Support Funds and the Ontario Research and Development Challenge Fund Grant.

References (29)

Cited by (379)

View all citing articles on Scopus
1

Senior authorship on this paper is equally shared by Drs. Kabani and Henkelman.

View full text