At the
start of my fourth week, I was given an independent project to do along with an undergraduate student there, Sarah. Our job would be to
perform immunohistochemistry (IHC) on mice brains to delineate regions in the striatum,
a region in the brain. IHC utilizes the specificity of antibodies to attach to
epitopes on specific antigens. This allows certain cells of a tissue section
(with a certain antigen) to be “selected” by a certain antibody. By adding a secondary
antibody with a fluorescent tag, target cells can be selectively labeled, imaged,
and analyzed. We sought to mark regions in the striatum based on cellular count
intensity.
The
procedure begins with perfusing the mice.
This is a gory procedure that involves anaesthetizing a live mouse, pumping out the blood from its circulatory system, injecting a preservative
solution into it heart, pumping this solution throughout its circulatory
system, chopping its head off, and dissecting out the whole brain. This process
fixes and preserves the brain tissue in its current state so that it can be
sliced and later studied. Thankfully,
I was given a few mice to practice on as my post-doc made it look much easier
than it actually is.
Afterwards,
the brain is sliced into 50 micron slices using a vibratome. The slices are
then soaked in a series of antibody
baths, mounted onto slides, imaged under an epifluorescence microscope, and
then the cells are counted based on a grid formation. This allow us to create a
heat map of the concentration of certain cell types in the striatum. Counting
the often hundreds of cells (in each slice) can be a very strenuous and
monotonous task and this is furthered by the fact that this need to be repeated
for dozens of slices for each brain (and we need to do a lot of mice for
statistically significant data). As a result, I spent the time in between steps
of the IHC protocol developing a semi-automated pipeline to count the cells using
an imaging software called ImageJ. Much of the fifth week involved writing the
code, but the thing is – I don’t really know how to code so that was a bit of a
doozy. It was a lot of watching videos, reading forums, and having no clue what
to while debugging and troubleshooting. Eventually, it all worked out though.
Using machine learning, the software is “taught” to recognize cells from
the background, and this training data can be saved and used in the counting
pipeline. Afterwards, one only need to feed the pipeline an image of a slice of
the brain. After a short while, it will summarize the cell counts of the slice
based on an automatically generated grid box.
I still have to figure out how to write the code so it can be feed an
entire folder of images and sort the resulting data, but that’s one of my goals
for the upcoming week(s). Next week, we plan to do do IHC on numerous more
mice, and this pipeline will save a lot of time in the near future as well as
after I’m gone from the lab.
Comments
Post a Comment