Workflow from Scientific Research

Pipeline for longitudinal tracking of fluorescently labeled SEP-GluA2 synapses in vivo. Daily imaging volumes were aligned using pairwise affine registration, followed by slice-by-slice pairwise affine registration to compensate for depth-dependent local tissue shift. Registered volumes were restored with XTC. Individual synapses were segmented with an ilastik-trained random forest model, followed by watershed to separate adjacent objects. Finally, a tracker trained through structured learning was used to longitudinally track synapses. t indicates current timepoint; n indicates number of subsequent timepoints.

DOI