Scaling up multiphoton neural scanning: the SSA algorithm.

In order to reverse-engineer the information processing capabilities of the cortical circuit, we need to densely sample neural circuit; it may be necessary to sample the activity of thousands of neurons simultaneously. Frame scanning techniques do not scale well in this regard, due to the time "wasted" scanning extracellular space. For scanners in which inertia can be neglected, path length minimization strategies enable large populations to be imaged at relatively high sampling rates. However, in a standard multiphoton microscope, the scanners responsible for beam deflection are inertial, indicating that an optimal solution should take rotor and mirror momentum into account. We therefore characterized the galvanometric scanners of a commercial multiphoton microscope, in order to develop and validate a MATLAB model of microscope scanning dynamics. We tested the model by simulating scan paths across pseudo-randomly positioned neuronal populations of differing neuronal density and field of view. This model motivated the development of a novel scanning algorithm, Adaptive Spiral Scanning (SSA), in which the radius of a circular trajectory is constantly updated such that it follows a spiral trajectory scanning all the cells. Due to the kinematic efficiency of near-circular trajectories, this algorithm achieves higher sampling rates than shortest path approaches, while retaining a relatively efficient coverage fraction in comparison to raster or resonance based frame-scanning approaches.