BTrack

This algorithm was developed to obtain growth rates of mouse mammary gland tissue imaged in 2D using transmitted light microscopy as a timelapse sequence. The algorithm required a segmentation method followed by a tracking methodology to track the growing end points of the tissue. To do so we developed a U-Net segmentation model to obtain a binary image, we then extended the
skeletonization implementation of imglib2 to obtain the skeleton end points. The end points of a skeleton always end before the actual tissue branch end, in order to ensure that the end point always ended at the actual tissue branch end we used a line drawn from the skeleton split point to the end point and obtained the intersection of that line with the boundary point extracted from the segmented image. Once we have such skeleton end points we repeat the process over time and use LAP/Kalman Tracker to link these points to obtain end point tracks for each growing tissue branch. Using these tracks we obtain growth rate of each track and finally average the growth rates of all the obtained tracks to obtain an average growth rate of the tissue. The developed Fiji plugin is a TrackMate detector called BTrack and is available after activating the Fiji update site BTrack from the update menu. TrackMate is fully interactive allowing for the users to remove or create new skeleton end points. The output of the Fiji plugin serves as an input to a jupyter notebook which takes in several of such experimental analysis and outputs a distribution plot of the growth rates coming from several such experiments.
We also created a T-test notebook using scipy stats  library functions of performing the T-test to compare the statistical significant difference between the results from several experiments.