Fig. 3
Flowchart of the data processing workflow. The data for each sample undergoes the six data transformation steps outlined. First, 2 d FFT transforms the holograms into reconstructions on the image plane. Next, the data are condensed by cropping around the region of interest (one of the two first-order reconstructions). Next, temporal FFT produces the fluctuation spectra for every pixel. The spectra of the foreground pixels are then averaged. Finally, the log-difference of every spectrum is computed with respect to the averaged baseline spectrum. The numbers in the parentheses in each box indicates the shape of the data, in the order of (number of measurements, pixels in x, pixels in y, number of frames).
