Autofocus blues on microplates
Peter Van Osta
pvosta_ROMMEL at rommel_cs.com
Mon Dec 1 07:39:34 EST 2003
I have some questions regarding the quality of microplates for microscopy,
mainly 96- and 384 well SBS-standard layout formats (1536 and 6144 layouts
would be interesting too of course).
I have been doing some quality tests for certain types of multiwell plates
which are very popular for use in "readers" with our 40x 0.75 N.A.
objective and a 63x 0.8 objective in brightfield and fluorescence. These
objectives have a very good spatial resolution and capture a lot of light
for weak fluorescence (I ~ N.A.^4/M^2). However these objectives have a
short working distance and as such the quality of the plate bottoms is
Our autofocus system only needs about 5 videoframes to gather enough
information to find a focus level, so with a 40 millisec. frametime (PAL
25 fps.) and some time for mechanical movement, it does a complete image
content based autofocus in 0.3 or 0.4 seconds (300 or 400 milliseconds).
But this only is the case, when a focuslevel is found within a certain
range, i.e. the travel range needed for 5 frames, which is in principle
determined by the Nyquist sampling theorem for Z-slicing.
We also do a complete range check, as we want to focus in those microscopy
modes which can cause multiple maxima in a focus algorithm (patented) in
which case we focus on the highest peak/score found ( Geusebroek J.M.,
Cornelissen F., Smeulders A. W. M., and Geerts H., Robust autofocusing in
microscopy, Cytometry, 36(1):1-9, 2000 ).
Using an image content based autofocus system, allows us to use very fine
Z-level focusing into cells with a varying offset to the multiwell plate
I found out that when you plot the bottom profile of standard multiwell
plates as a 3D landscape, these plates look more than a mountaneous region
than having a flat profile.
I also found some microplate types with very thin plastic bottoms and a
flat plate bottom, but they are of course a bit more expensive. Glass
bottom plates are an alternative in some cases, but I heard that the glue
which is used for these plate-bottoms is not compatible with some of the
solvents used in farmaceutical research (DMSO, etc), so the bottoms tend
to fall off the plate after 24 h. incubation, which would be rather
inconvenient for our time-lapse work [:-(] ? So using good quality plastic
bottom plates seems to be one possible solution ?
A possible solution for the variability of the plate bottoms is to do a
prescan at lower mag./N.A., either with a plate bottom finding
laser-system or a content-based autofocus and model the plate with a
mathematical model, but this takes time.
Using Long Distance (L.D.) objectives is another possible solution, but
these objectives have a lower N.A. and as such less spatial resolution,
which hampers subcellular imaging and analysis. Also the reduced N.A. has
a very dramatic efect on the amount of light captured by the objective (I
~ N.A.^4/M2) and we always do our image capturing in real-time, we just
use very sensitive cameras which allow imaging without the need to
integrate frames. The image capturing must be done at 40 msec. (PAL 25
fps.), which is important when doing large volumes of multiwell plates in
our screenings. Using the 0.75 and 0.8 N.A. objective in combination with
the ultrasensitive camera allows us to dilute reagents up to 1000x, which
is very good for our assay costs when doing large screens of subcellular
I would like to know what is the experience of other researchers with the
quality of multiwell plates on microscopes or similar (automated) readers
? Does anyone else uses high N.A. objectives on multiwell plates for
screening ? As far as I understand the SBS-standards define XY-measures of
multiwell plates, but leave the bottom offset and bottom thickness to the
manufactureres of multiwell plates ?
Peter Van Osta
More information about the Cellbiol