High resolution imaging in awake behaving mice: Motion correction and virtual reality.

摘要

Two photon laser scanning microscopy (TPLSM) is a powerful tool for the examination of neural circuits. This thesis discusses some advancements in methodology necessary for applying its use to awake behaving mice. Chapters 1 and 2 discuss the challenge of motion artifact in TPLSM as it relates to imaging in head fixed mice atop a spherical treadmill. Chapter 1 describes a hidden Markov model (HMM) based motion correction algorithm which allowed our laboratory to be the first to image activity related fluorescence transients from calcium sensitive dyes in an awake mammal. Chapter 2 discusses measurements of the dynamics of brain motion with high spatial (50nm) and temporal (500Hz) resolution, which characterize brain motion at an unprecedented scale and serves as a ground truth measurement with which to compare the predictions of the HMM algorithm. We find that the spectral properties are such that most brain motion is in the 0-30 Hz range, that the maximal brain speed is ∼25 nm/ms, and that velocity distributions have exponentially distributed tails with speed constants ∼3nm/ms. Because these measurements are done simultaneously with TPLSM, they also directly demonstrate that the HMM model accurately predicts brain motion, avoiding over-fitting errors. Chapter 3 describes the construction of a virtual reality apparatus which can facilitate the study of complex navigation behaviors in head fixed animals. The technical considerations in designing a projection system using a toroidal screen and an angular amplification mirror are discussed. We find that ray tracing simulations suggest that the system is adequate for presenting realistic visual stimuli to mice. This system has facilitated the study of intracellular dynamics of hippocampal place cells, as well as imaging of the same phenomenon. Chapter 4 discusses the first biological feasibility tests of a theoretical idea called rate specific synchrony, in which a common noisy oscillation to a group of neurons causes their spike times to be synchronized when their firing rates are equal, but rapidly desynchronized when their rates are different. Using whole cell patch recordings from layer 2/3 cortical neurons, we injected complex noisy oscillations with varying amounts of steady depolarization, measured the resulting spike times, and quantified the synchronization observed between successive trials as a function of rate.