The main goal of this project is to characterize the development of the murine brain, with emphasis on white matter anatomy, using magnetic resonance micro-imaging in conjunction with mathematical methodologies for quantitative image analysis. The traditionally used histological methods for examination of murine brain sections are limited by tissue distortion or loss, by difficulties in constructing a spatially consistent volumetric image from sections, by extensive effort in preparation, and by lack of capability for in vivo examination of the mouse brain. Magnetic resonance imaging (MRI) is emerging as a technology with strengths complementary to histology, with respect to these limitations. In this project, we will develop methods for imaging and analysis of the murine brain, and we will use them to generate normative data for brain development of the C57BL/6J mouse strain. Our emphasis will be on using diffusion tensor imaging (DTI) to characterize the white matter architecture. Building upon current work by several groups in the Human Brain Project, we propose to develop mathematical methodologies for computational anatomy, which complement traditional analysis methods in mainly two ways. First, they can identify very subtle and localized shape characteristics, without the need to know the location of an affected brain region a priori. Second, they are highly automated and quantitative, thus enabling the examination of a large number of animals with minimal effort, using statistical image analysis techniques. Our image analysis methodology will involve shape analysis methods for the reconstruction and spatial normalization of murine brain structures, and it will utilize the well-established framework of stereotaxic space analysis. After mass-preserving spatial normalization of MRI images to a stereotaxic space of the respective developmental stage, the normal anatomic variation of grey and white matter structures will be measured at a number of different developmental stages. This nonnative data will allow subsequent studies aiming to identify regions of abnormal development in neurogenetic mice, by finding regions that fall outside this normal range. Accurate spatial normalization will also enable us to develop methods for reconstruction of major fiber pathways, by using statistical averaging in a stereotaxic space, thereby drastically increasing SNR of the DTI images, which typically are noisy. We will test this methodology on a pilot study of the Emx-1 knockout mouse, a well-characterized strain with abnormal cortical lamination and defasciculated white matter fiber tracts, including the corpus callosum, and we will validate our MR-based measurements using histological sections.
