Fig. 2: Parallel cluster wavelet analysis (PCWA) for single-peak analysis.

a Cropped window of fluorescence signal taken from 200 nm fluorescent beads excited by single-mode (SM) waveguide (inset: Ricker wavelet used with PCWA algorithm). b CWT coefficients in time-Δt space (where a scaled and dilated version of the mother wavelet is convolved with raw data) with square markers indicating selected local maxima points found by the PCWA event detector algorithm. c Zoomed-in window of three events with circle markers showing the adjusted location of peaks. d CWT map of (c) including local maxima points (black dots). The clustering algorithm utilizes Euclidean distance and adjusted ellipses around each local maximum to search for links. The overlap of an ellipse with the centroid point defines a link. e Macro and micro clusters (ΜC and μC): local maxima are first grouped into ΜC highlighted by blue circles by simplified 1D overlap calculation. The clustering algorithm finds μC for each ΜC in parallel. A μC is a star graph containing a minimum of links with the largest CWT coefficient maximum as the centroid (red-filled circles). f Flowchart of the clustering algorithm. g Run time comparison of clustering algorithm with established CWT peak finders, showing orders of magnitude faster speed and run times below the real-time limit (gray dashed line).