Wavelet transform-based de-noising for two-photon imaging of synaptic Ca2+ transients.

Abstract

Postsynaptic Ca(2+) transients triggered by neurotransmission at excitatory synapses are a key signaling step for the induction of synaptic plasticity and are typically recorded in tissue slices using two-photon fluorescence imaging with Ca(2+)-sensitive dyes. The signals generated are small with very low peak signal/noise ratios (pSNRs) that make detailed analysis problematic. Here, we implement a wavelet-based de-noising algorithm (PURE-LET) to enhance signal/noise ratio for Ca(2+) fluorescence transients evoked by single synaptic events under physiological conditions. Using simulated Ca(2+) transients with defined noise levels, we analyzed the ability of the PURE-LET algorithm to retrieve the underlying signal. Fitting single Ca(2+) transients with an exponential rise and decay model revealed a distortion of τ(rise) but improved accuracy and reliability of τ(decay) and peak amplitude after PURE-LET de-noising compared to raw signals. The PURE-LET de-noising algorithm also provided a ∼30-dB gain in pSNR compared to ∼16-dB pSNR gain after an optimized binomial filter. The higher pSNR provided by PURE-LET de-noising increased discrimination accuracy between successes and failures of synaptic transmission as measured by the occurrence of synaptic Ca(2+) transients by ∼20% relative to an optimized binomial filter. Furthermore, in comparison to binomial filter, no optimization of PURE-LET de-noising was required for reducing arbitrary bias. In conclusion, the de-noising of fluorescent Ca(2+) transients using PURE-LET enhances detection and characterization of Ca(2+) responses at central excitatory synapses.