Computational de-scattering for enabling high rate deep imaging of neural activity traces: Simulation study

Parallel excitation and `light-sheet' optical methods, such as multifocal multiphoton microscopy, selective plane illumination microscopy, and temporal focusing microscopy offer significant enhancement of data acquisition rates compared to the relabtively slow serial-scanning rate of two photon laser scanning microscopy (TPLSM). However, since these methods typically rely on detecting the fluorescence signal using a camera, they are sensitive to tissue scattering effects, which blur the images, reduce the spatial resolution, and eventually limit the maximal imaging depth. Therefore, an efficient method for de-scattering of the acquired data is needed in order to maximize parallel excitation methods' utility for imaging neuronal activity inside tissues. Here, we present a simple algorithmic approach for data extraction from blurred images acquired by a dual-modality microscope, which offers additional information regarding the cells' geometrical location. Simulations predict that the new approach is capable of extracting functional information for depths that may potentially exceed 700μm in cortical tissues, even in the presence of severe additive noise corruption.