Time-lapses were taken with the same parameters, position then lambda order and without analog gain, field and aperture diaphragms as closed as possible, DIC images were taken 9v DIA-lamp intensity with exposure of 200 ms, then we switched to red fluorescent channel (excitation <>, emission <> filter) and turning the DIA-lamp off with exposure of 600 ms, then to green channel (excitation <>, emission <> filter) with 200 ms exposure at the a light off image to avoid photo bleaching before the next time loop started.
Image Processing
To process and get information from the time-lapses we develop a tool called
\(\mu\)J (available
here) that consists mostly in scripts in ImageJ
\cite{Schneider_2012} macro language to process and organize the images, the we use DeepCell
\cite{Van_Valen_2016} a deep learning strategy to generate masks, and some other python [ref] scripts to visualize the data. The overall process could be generally described in the next series of steps: 1) Order the archives produced by the NIS Elements software. 2) Align traps. 3) Generate a segmentable image. 4) Perform segmentation with DeepCell. 5) Clean the masks. 6) Manual correction of the masks. 7) Substract information and generate data 8) Analyze and visualize data.