Optical and computational approaches for mapping brain activity and structure

Abstract

Neuroscience always has been a heavily technology-starved field that has been often revolutionized with the rise of a new technology; and arguably the future advancement of neuroscience will also depend largely on the development of new technologies that allow acquiring new data sets that can provide deeper insights into the brain. While there is no universal agreement as to what data sets are needed to fully reveal the underlying principles of brain computation, neurons will certainly serve as an important layer to study the brain considering their discreteness and electrical characteristics. In other words, seeing a brain as a circuitry that is made of the neurons, it is important to study the relation between the brain's structure and its function at the neuronal level. Unfortunately, a brain is a huge network that consists of a huge number of neurons which makes it difficult to see the "big picture" while retaining the single neuron resolution. The aim of this study is to the develop such technologies that allow to see and analyze the large network's structure and dynamics with the single neuron resolution. The core strategy is to use optical microscopy to acquire the raw data and to infer the information of our interests with computational techniques with designing both optical and computational parts with each other in mind to maximize the synergy. The first part of the thesis describes the development and application of computational imaging techniques to monitor the brain activity in 3-D which allowed us to see how the neurons interact at an unprecedented speed. The second part describes a computational approach to extract a wiring diagram of a brain from an optical image of the brain rather than an electron microscopy image, as optical microscopy is undergoing rapid development and is likely to outperform electron microscopy in terms of scalability which will be an important criterion to map the whole brain. These will be the tools of great utility in neuroscience that can generate rich data sets that will be of wide interests to system neuroscientists as well as cellular/molecular neuroscientists.