I made 2 prototypes for decrosstalk.
(1) FOV-based decrosstalk, using a fixed filtering parameter.
(2) Time series-based decrosstalk, similar to the legacy code with some updates (paired plane registration, constrained ICA).
Some more updates are needed, including developing QC metrics and visualizations, but these should be enough to give us starting points (at least a little bit of improvement from the legacy protocol).
I made 2 prototypes for decrosstalk. (1) FOV-based decrosstalk, using a fixed filtering parameter. (2) Time series-based decrosstalk, similar to the legacy code with some updates (paired plane registration, constrained ICA).
Some more updates are needed, including developing QC metrics and visualizations, but these should be enough to give us starting points (at least a little bit of improvement from the legacy protocol).