Fig. 2

The different steps of the convolutive blind source separation algorithm. A The input signal is unfused tetanus. B The unfused tetanus is extended using the extension factor \(R\), where the extended signal has a dense covariance matrix. C Performing whitening such that the covariance matrix equals the identity matrix. D The fixed-point algorithms find the separation (projection) vector (red line), and by projecting it on the whitened extended signal in C, the resulting output is the source estimate (blue line). E The time instants of the local maxima of the source estimate were detected using peak detection with two parameters: the minimal peak distance (\(\mathrm{MPD}\)) and the number of mean absolute deviations (\(n\mathrm{MAD}\)). F The time instants of the local maxima are the estimated firing times of the spike train \({\varvec{s}}\) (red line)