I've been using bTrack to track cells over long time courses and visualize them using napari, but I've been running into two (related) issues. They are as follows:
bTrack labels single pixels as cells. This appears to be a consequence of discontinuous cell masks at the periphery of cells (masks generated by CellPose). I want to be able to discount them when using the btrack.utils.segmentation_to_objects() function (perhaps imposing a minimum area requirement for cell labeling?), but this doesn't appear to be an option. How might I be able to remedy this issue without altering the cell masks I've generated?
bTrack fails to keep track of mitosis events for cells of expected size, however fluctuations in the falsely detected single pixel "cells" are detected as mitosis events. For cells of expected size, the parent cell is tracked and after division, one of the daughter cells keeps the same track as the parent while a new track is generated for the other daughter cell. As a consequence, lineage is indeterminate and absent in the data for relevant tracks. Conversely for the false cells, if a false cell is detected a frame or two before multiple false cells are detected in the nearby area, a lineage is established by bTrack. Why is bTrack generating lineages for these false cells while the expected cells that undergo mitosis do not get any? And what can I do to fix the situation?
I've attached my configuration file, but I'm happy to provide any additional materials from my work. Thank you!
Hi,
I've been using bTrack to track cells over long time courses and visualize them using napari, but I've been running into two (related) issues. They are as follows:
bTrack labels single pixels as cells. This appears to be a consequence of discontinuous cell masks at the periphery of cells (masks generated by CellPose). I want to be able to discount them when using the btrack.utils.segmentation_to_objects() function (perhaps imposing a minimum area requirement for cell labeling?), but this doesn't appear to be an option. How might I be able to remedy this issue without altering the cell masks I've generated?
bTrack fails to keep track of mitosis events for cells of expected size, however fluctuations in the falsely detected single pixel "cells" are detected as mitosis events. For cells of expected size, the parent cell is tracked and after division, one of the daughter cells keeps the same track as the parent while a new track is generated for the other daughter cell. As a consequence, lineage is indeterminate and absent in the data for relevant tracks. Conversely for the false cells, if a false cell is detected a frame or two before multiple false cells are detected in the nearby area, a lineage is established by bTrack. Why is bTrack generating lineages for these false cells while the expected cells that undergo mitosis do not get any? And what can I do to fix the situation?
I've attached my configuration file, but I'm happy to provide any additional materials from my work. Thank you!
cell_config -test 3.txt