AUTOMATED 3-D QUANTIFICATION OF BRAIN TISSUE AT THE CELLULAR SCALE FROM MULTI-PARAMETER CONFOCAL MICROSCOPY IMAGES

摘要

Brain tissue is the most complex tissue in the mammalian anatomy. It is structurally complex, containing diverse cell types and vasculature with complex spatial relationships. It is also complex in terms of molecular constituents, and their relationship to the structural constituents. There is a need for technologies to map normal and injured brain tissue and compute appropriate quantitative measurements in support of a large number of investigations. We present a systematic methodology named FARSIGHT (fluorescence association rules for image-based insight), combining multi-parameter immuno-histochemical labeling, three-dimensional confocal microscopy, linear unmixing, segmentation, and associative morphometry to map and quantify diverse aspects of brain tissue. In addition to morphometric measurements of individual structures from individual imaging channels, we propose the use of distance maps, distance-weighted integrals, and distributions as an efficient basis for computing a versatile set of such associative measurements. These features enable classification of all the major cell types, and computation of their inter-relationships.