Joint Atlas-Mapping of Multiple Histological Series combined with Multimodal MRI of Whole Marmoset Brains
Year: 2018
Type: preprint
Abstract: Development of a mesoscale neural circuitry map of the common marmoset is an essential task due to the ideal characteristics of the marmoset as a model organism for neuroscience research. To facilitate this development there is a need for new computational tools to cross-register multi-modal data sets containing MRI volumes as well as multiple histological series, and to register the combined data set to a common reference atlas. We present a fully automatic pipeline for same-subject-MRI guided reconstruction of image volumes from a series of histological sections of different modalities, followed by diffeomorphic mapping to a reference atlas. We show registration results for Nissl, myelin, CTB, and fluorescent tracer images using a same-subject ex-vivo MRI as our reference and show that our method achieves accurate registration and eliminates artifactual warping that may be result from the absence of a reference MRI data set. Examination of the determinant of the local metric tensor of the diffeomorphic mapping between each subject's ex-vivo MRI and resultant Nissl reconstruction allows an unprecedented local quantification of geometrical distortions resulting from the histological processing, showing a slight shrinkage, a median linear scale change of ~-1% in going from the ex-vivo MRI to the tape-transfer generated histological image data.
Source: arXiv (Cornell University)
Authors: Brian C. Lee, Meng Kuan Lin, Yan Fu, Junichi Hata, Michael I. Miller, Partha P. Mitra
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2802576194
