DIGRESSION:
If you want better discrimination in Z, there are a few options. First, you can take your image slices in Z closer together. Typically we do 0.2 micron spacing in Z and thus we're stuck with an error that's not much better than 0.2 microns. Second, if you have a dilute system of particles, then you can increase your diameter in Z and get better subpixel resolution. (This works in X and Y as well). However, this only works if you have particles spaced far apart, otherwise you'll grab information from other particles. This will result in less apparent pixel-biasing, but it's misleading, and won't help your particle location.
I hope to write more about this later. A couple quick comments,
though. First, the better your resolution, the sharper your pair
correlation function will look. In general, if you do
something to poorly pretrack your data, it will only make
the pair correlation function look worse, in the sense
that the height of the first peak will be diminished.
Second, if you track your data, you can calculate the mean
square displacement (MSD). The better your resolution,
the lower the MSD will be. Again, hard to do a bad job
pretracking without having it increase the MSD. Be sure to
check all three directions
One last comment: sometimes you can do extremely well at
pre-tracking, as defined by these measures, but only by
throwing out any and all even slightly ambigious particles.
This is up to you. If you want a nice looking mean square
displacement curve, maybe this is OK; but for the pair
correlation function, it seems like one ought to consider as
many particles as you think are real, even if their positions
aren't terribly accurate. As always, tracking is a bit of an art
and you can make a judgement call on the tradeoff: track every
particle, or fewer particles better. I usually prefer the
former.
|
Click here to go back to the 3D tutorial