Texture-based Mouse Brain Atlas

(MICCAI 2017 paper, PhD thesis defense slides, SfN 2016 poster)

A brain atlas is a quantitative description of the organization of anatomical structures of a brain. Nissl-stained serial sections reveal cellular texture (cytoarchitecture) and are the gold standard for defining structures. The increasing ease of brain image acquisition and the adoption of automated registration algorithms transforms the role of an atlas from a static map to a spatial database where a wide variety of experiment data acquired from different subjects can be integrated in a common anatomical framework. Registration is the central computational issue in this effort.

Traditional registration methods rely on maximizing the similarity between downsampled voxel intensities (> 10 μm), neglecting fine-scale textures that are characteristic of histological data (0.5 μm resolution) and therefore often fail to accurately align structures that are defined by cytoarchitecture rather than graylevel, such as many brainstem nuclei. The lack of a registration tool that strongly utilizes texture and a nucleus-level atlas to assist such registration has stifled comparison of data across experiments.

We demonstrate a data-driven atlas system that automatically aligns brains based on cytoarchitectural landmarks. Our approach combines discriminative texture detectors based on convolutional neural network (CNN) features with a reference atlas that describes structure shapes as probabilistic volumes and their locations as Gaussian distributions. Histological serial sections are reconstructed in 3-D and converted to structure probability maps by classifiers trained to differentiate the texture inside versus outside each structure. Registration is achieved by maximization of the correlation between the probability maps and the reference atlas using a global affine transform followed by deformation interpolated from structure-specific rigid transforms. Initialized from annotation by expert neuroanatomists, the atlas is continuously refined after incorporation of new brains in a semi-supervised fashion.

Based on automated registration of twelve specimens, we developed an atlas for adult mouse brainstem which defines 28 distinct structures in both hemispheres. The system’s utility in advancing brain circuitry study is demonstrated by the precise mapping of neuronal projections in cytoarchitecturally ill-defined regions across brains from different animals. Quantitative results showed that our use of CNN features leads to highly accurate classifications, superior to using LBP, GLCM or mean graylevels. This allowed precise and confident registration and significantly reduced human labor.

Unsupervised Discovery of Robust Landmarks from Brain Section Series


Brightfield and fluorescent imaging of whole brain sections are fundamental tools of research in mouse brain study. As sectioning and imaging becomemore efficient, there is an increasing need to automate the post-processing of sections for alignment and three dimensional visualization. There is a further needto facilitate the development of a digital atlas, i.e. a brain-wide map annotatedwith cell type and tract tracing data, which would allow the automatic registration of images stacks to a common coordinate system. Currently, registration ofslices requires manual identification of landmarks. In this work we describe the first steps in developing a semi-automated system to construct a histology atlas of mouse brainstem that combines atlas-guided annotation, landmark-based registration and atlas generation in an iterative framework. We describe an unsupervised approach for identifying and matching region and boundary landmarks,based on modelling texture. Experiments show that the detected landmarks correspond well with brain structures, and matching is robust under distortion. These results will serve as the basis for registration and atlas building.