PROJECT SUMMARY To understand the computations in the brain, we need to monitor the activity of neural circuits at high accuracy, which requires methodologies with high spatial and temporal resolution. Non-invasive and capable of resolving subcellular structures, optical microscopy has been extensively applied in the field of neuroscience, with a variety of methods developed to image neural activity at high speed, large depths, and/or over large spatial scales. For example, free-space angular-chirp-enhanced delay two-photon fluorescence microscopy was developed to record membrane voltage at kHz frame rate in the brain in vivo. Three-photon fluorescence microscopy, an emerging method that uses excitation light of longer wavelengths than two-photon fluorescence microscopy, has large penetration depths and is capable of imaging structures over 1-mm deep in the mouse brain. An alternative to the point-scanning multiphoton fluorescence microscopy above, single-photon widefield fluorescence microscopy has also been applied to in vivo monitoring of brain activity. Most commonly, the entire sample is illuminated and the emitted fluorescence collected by an objective lens and imaged with a camera, which enables fast activity imaging of superficial structures, sometimes over millimeters in lateral dimension. To obtain accurate measurements of neural activity in vivo, however, one has to combat the degradation of the resolving power of these microscopy methods when they are applied to brain tissue. The optical inhomogeneity of the biological tissue itself distorts the image-forming light and prevents all microscopy modalities from achieving their designed performance in vivo. When applied to activity imaging, such degradation can lead to erroneous conclusions. Here, we propose to optimize and apply adaptive optics methods developed in the Ji lab to select cutting-edge high- speed, large-depth, and large-scale activity recording modalities for high-accuracy measurements of neural activity in vivo.