Deep Learning Neuronal Mapping in Fluorescence Microscopy Imaging of the Human Brain

Quantitative studies of the human brain at the microscopic scale require both high-resolution imaging of large brain tissue samples, and ways of automatically analyzing the massive quantities of data created in the process. Fluorescence Microscopy has the imaging capability needed to cover the scale gap between the micron scale, on which individual neurons are defined, and the centimeter scale of brain functional areas. This potential is still partially unexpressed due to engineering challenges in automated processing and analysis of such massive data-streams that, to this day, have not been entirely solved. This work proposes a methodological framework for the exploration of extended areas of Human Brain, by means of Two-Photon Confocal Microscopy and Light-Sheet Fluores- cence Microscopy Imaging, exploiting Deep Learning models and careful data-flow design to map large quantities of individual neurons across vast volumetric extensions.