Please use this forum to discuss our data analysis package. Details can be seen at Analysis Package. Here we have compiled some Frequently Asked Questions about analysis:
Q: I want to record the same cells day after day for weeks. When looking at your example segmentation I was concerned that the cell contour shapes were not visually very similar day after day - from a visual perspective I had trouble matching contours to contours on the maps from day 1, day 14 and day 30. One possible theory for some source of the change is change in Z.
A: We see, as well as others, about an 80% turnover rate of active cells on a linear track over the course of just 7 days. So we might only expect about 50 segments to match across days 1 to 14 or days 14 to 30. Visually inspecting the figure to find some of the ~50 out ~950 segments that match is difficult. That being said, the segment shape, or contour, will vary depending on the segmentation algorithm used. Our old algorithm would generate very similar looking contours across sessions. Our newer segmentation approach produces a bit more variability of a cell's contour across recording sessions. This is because the contour is not only affected by the cell shape being segmented but also by which surrounding cells were active in each session.
Our segment matching algorithm supports both centroid closeness and ROI overlap but, from our experience, centroid closeness produces much more consistent results. This again is mainly due to the ROI shape being more variable between recording sessions than ROI centroid. I should also mention that the centroid closeness criteria we use is extremely conservative (a little larger than half the average radius of a segment).
We think the z drift is pretty minimal for us, at least relative to the depth of view of our system. A reasonably sized shift in z doesn't cause us to lose cells, it just changes their size and scattering/focus a bit. There are 2 ways we could have drift in z: 1. The distance between the lenses and imaging sensor changed; 2. The cell layer moved relative to the implanted GRIN lens. We lock the imaging sensor in place relative to the optics and this doesn't change for the length of our experiments so #1 is unlikely. Movement of the cell layer relative to the GRIN lens is possible but might be unlikely after giving the brain a few weeks to recover after GRIN lens surgery (which we do).
Posted by Tristanshuman (administrator) on 14 January 2016 at 16:44.
Edited by Tristanshuman (administrator) on 18 January 2016 at 05:56. |
|
Write a reply