Bryan Douglas Crawford (bryanc at fraser.sfu.ca) wrote:
> I'd like to solicit some net.advice on what sort of apparatus to
> purchace for doing computer-aided digital light microscopy. We're
> thinking of buying something along the lines of a color CCD, a
> computer of some kind (preferably PC, as that's what the rest of
> the machines here are) with a high-res frame grabber, a tape or worm
> drive for storage of large image files (possibly time-lapse series),
> a video printer, and software to drive it all.
First, I do not think you should get a _color_ CCD, as those have less
resolution. We recently got a system, and settled for the Sony XC-77,
which is good quality; we paid the equivalent of $3000, but this is
transatlantic, and we always pay too much. At about 2/3 price, some
other acceptable cameras are available. We would have liked the Dage
CCD-72, but it costs about three times as much as the XC-77 (both give a
_video_ signal, best resolution about 768*512 px in CCIR standard,
640*480 in yours).
The computer and the grabber card depend entirely on which program you
would like to use. This again depende on your purpose for the
microscopy. In our case, we wanted to measure grain counts and area
brightness values, mainly from in situ hybdirized slides. For
brightness, the NIH Image program is PD, available by anonymous ftp from
zippy.nimh.nih.gov, and exists only for the Mac. This program is highly
satisfactory, and works e.g. with the Scion LG-3 grabber card, which at
1 Mb memory is US$895, and we upgraded that for speed and storage
to 16 Mb at a cost you can guess at, normal RAM prices. An alternative
would have been the Data Translation DT2255 card, cost about US$1300.
-- One then needs the computer, which must have a full size NuBus slot -
means a Mac LCIII or something.
I emphasize that you need to choose first the program which does what
you want; then the hardware between camera and program.
Our solution was influenced by the fact that another lab wanted to
purchase a cooled CCD camera system at the same time, to achieve
resolutions of 1300*1000 pixels, and also do fluorescence microscopy.
For their purpose, we eventually decided for the Photometrics 250 camera
with a Kodak KAF1400 chip. This is expensive stuff, about US$20,000 with
a _digital_ controller card and drivers. Why I mention this is because the
possible program solutions were: 1) (most cost-effective) IPLab with
extensions, from Signal Analytics, for Mac II:s up; 2) BDS Image, from
Biological Detection Systems, for the Mac; 3) MCID, from Imaging
Research in Canada, for AT/386/486. Because the MCID M4 (cheaper) does
not give more resolution than the video signal-based XC-77, and we could
not afford the top-of-the-line M2, we rejected the PC solution, and went
for IPLab. We bought an extra IPLab license to go with our XC-77 camera,
and use this for grain counting, where it beats the NIH Image. For
manipulating images, IPLab is also superior to NIH Image. Currently, we
have three IPLab licenses in the department, one digital (Photometrics),
and two video.
However, in your case a PC-based possibility would be Optimas, from
BioSoft. This is a versatile program for biological use, and while it is
much more difficult to learn than NIH Image, it sure performs. I am not
sure about the price, but it could be like $3000. This is a very good
program, and I would not look twice at products like Jandel's JAVA
(which we have) or Global Lab subroutines. Have a look at Image Pro, and
Jandel's Mocha, while you are at it. -- Remains grabber and PC. As you
do not need more than the 640*480 resolution (RS-170; the CCIR is
768*512) anyway, a DT2855 might do, you might get it for about US$1300;
Perceptics have nice cards in their Pixel... series, but they cost more,
the PixelGrabber is OK, the PixelBuffer has a lot of memory etc. For the
PC, I think a 486 is reasonable, but you want some hard disk. Here, 170
Mb:s are still cheap, and the question is - do you _really_ want to
store the digitized files - why not keep only the slides, the
measurement results, a photo, and maybe a hard copy of the view. You
hardly need a video printer - export your view as a TIFF file, and print
it on a laser printer with e.g. CorelDraw or whatever you have.
(There the PC beats the Mac in printing speed...)
> Can I do this for around $10kCDN?
Apparently you have the trinocular microscope already.
If you go for the NIH Image solution, you need the XC-77, say US$2000,
plus the Scion LG-3, $895, plus a Mac LCIII with a big-big HD plus an
optomagnetic drive. The Image program is free. There are other boards
available, and they usually come with a special version of NIH Image.
If you absolutely want a PC solution (in my mind not reasonable, you
want the XC-77 at the $2000, a DT2855 at least at $1300, the Optimas
program (guessing at $3000), and a 486 with a big HD plus the
optomagnetic drive. You lose on the program cost, compared to the PD
Mac solution. You also lose some flexibility in use of mouse, etc.,
which is easier to implement on Macs.
Image analysis made me a Macintosh user, despite hard resistance.
Dag Stenberg stenberg at cc.helsinki.fi
Department of Physiology stenberg at finuh.bitnet
P.O.Box 9 tlx: 1002125 finuh sf
(Siltavuorenpenger 20 J) fax: int.+358-0-1918681
FIN-00014 University of Helsinki,Finland tel: int.+358-0-1918532