Search

Recording the activity of neural populations at high sampling rates is a fundamental requirement for understanding computation in neural circuits. Two photon microscopy provides one promising approach towards this. However, neural circuits are three dimensional, and functional imaging in two dimensions fails to capture the 3D nature of neural dynamics. Electrically tunable lenses (ETLs) provide a simple and cheap method to extend laser scanning microscopy into the relatively unexploited third dimension. We have therefore incorporated them into our Adaptive Spiral Scanning (SSA) algorithm, which calculates kinematically efficient scanning strategies using radially modulated spiral paths. We characterised the response of the ETL, incorporated its dynamics using MATLAB models of the SSA algorithm and tested the models on populations of Izhikevich neurons of varying size and density. From this, we show that our algorithms can theoretically at least achieve sampling rates of 36.2Hz compared to 21.6Hz previously reported for 3D scanning techniques.

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.