🚀 go-pugleaf

schwaigr@news.weeg.uiowa.edu (Jim Schwaiger) writes:

>I am seeking a flexible frame grabber which would allow me to grab
>the video from medical imaging systems (CT, MR, etc).  The closest
>I have come is a $4-5000 board from Imaging Technology.  Does
>anyone have alternative suggestions (Medical imaging systems use
>non standard line lengths and video parameters in an attempt to make
>this difficult).

This is not exactly true.  The requirements and underlying mathematics
for generating the image dictate the matrix size and produce square
pixels.  The monitor and associated electronics are designed to display
square pixels efficiently.  In addition, the images are usually displayed
on non-interlaced monitors to reduce flicker, eyestrain, and registration
problems inherent with RS-170.  RS-170 incorporates severe compromises
to make the signal fit in a limited bandwidth.
 
It is also not a particularly good idea to capture displayed video
using a frame grabber.  1) You are redigitizing a digital signal that has
been passed through a DAC and filtered.  2) There is the problem of
graphics characters that will create problems for even simple image
processing algorithms.  3) Much of the dynamic range of MR and CT images
is lost because the original 12-bit images have been passed through "window
and level" adjustments to suit a particular viewing requirement.  The
window and level adjustments remap the original gray-scale into another
and that will most likely invalidate any post processing you may have
in mind.  Images obtained in this manner are pretty to look at but
basically useless for any additional post processing.
 
A better acquisition strategy would be to get archive tapes from the
machines.  If you are only interested in the gray-scale image, tape
readers are not hard to build although some CT archive tapes have pixel
maps that have to be decoded to determine where the pixel should be
placed in the raster.  Compressed tapes are more trouble but you may be
able to get a specification from the vendor.  Uncompressed MR images
from some scanners can be extracted from tape using nothing more than
a 9-track tape drive on a UNIX machine, patience, and the UNIX "dd"
command.
 
Another strategy is to modify the scanner by placing a digital bridge
at the input of the output LUT.  This has been done in some locations
where it is desired to interface the scanners into a PAC System.  It has
the advantage of being relatively easy to do, it gives you images with
full dynamic range, and you can get images without graphics.  The down
side is that the vendor and/or owners of the machine might not want you
mucking about in their expensive toys.
 
If you are dead set on digitizing video, you may want to contact AT&T
PACS division (I don't know their address/phone).  They installed
equipment to do just what you want here at UW Medical Center as an
evaluation site for PACS hardware.   

--
Larry                                   |   Thunder makes all the noise;
                                        |   lightning gets the job done.
larryd@milton.washington.edu
I have my opinion; the university has its own.