Shift and Mean Algorithm for Functional Imaging with High Spatio-Temporal Resolution

Sylvain Rama

Understanding neuronal physiology requires to record electrical activity in many small and remote compartments such as dendrites, axon or dendritic spines. To do so, electrophysiology has long been the tool of choice, as it allows recording very subtle and fast changes in electrical activity. However, electrophysiological measurements are mostly limited to large neuronal compartments such as the neuronal soma. To overcome these limitations, optical methods have been developed, allowing the monitoring of changes in fluorescence of fluorescent reporter dyes inserted into the neuron, with a spatial resolution theoretically only limited by the dye wavelength and optical devices. However, the temporal and spatial resolutive power of functional fluorescence imaging of live neurons is often limited by a necessary trade-off between image resolution, signal to noise ratio (SNR) and speed of acquisition. Here, I propose to use a Super-Resolution Shift and Mean (S&M) algorithm previously used in image computing to improve the SNR, time sampling and spatial resolution of acquired fluorescent signals. I demonstrate the benefits of this methodology using two examples: voltage imaging of action potentials (APs) in soma and dendrites of CA3 pyramidal cells and calcium imaging in the dendritic shaft and spines of CA3 pyramidal cells. I show that this algorithm allows the recording of a broad area at low speed in order to achieve a high SNR, and then pick the signal in any small compartment and resample it at high speed. This method allows preserving both the SNR and the temporal resolution of the signal, while acquiring the original images at high spatial resolution.