MacIntosh Video RFLP Digitizing Analysis

Robert M. Straubinger rms at acsu.buffalo.edu
Thu Oct 24 19:56:11 EST 1991


>In article <9110230157.AA26288 at ccvr1.cc.ncsu.edu>, nmodena at unity.ncsu.edu writes:
>> Looking for advice ... Re: RFLP Video-imaging Setup and Use
>> We have ordered: ... MacIntosh FX computer ... 16 Mb RAM ...
>>         JVC TKH70 Still Image Video Camera
>>         RasterOps 24XLTV Quick Capture Card
>> Our purpose is to capture, store and transmit RFLP radiographs
>> for analysis here and in other labs ...
>> Looking for suggestions and experiences using this equipment ...
>> ... FTP sites with FAQs, appropriate software, etc.
>> And anything else I forgot to mention that you think it is worth 
>> knowing about.

Will you be getting enough resolution with this camera? Even if it gives
images of 1200x1200 pixels, that's far less than you can get out of flatbed
scanners, which can easily (ie cheaply!) give 300 pix/inch * 10 in = 3000 pix^2
How much resolution and dynamic range (eg. pixel depth) do you need for
RFLP analysis? Would a reasonably-priced scanner do?

In article <1991Oct24.112030.472 at gnv.ifas.ufl.edu> afc at gnv.ifas.ufl.edu (Andrew F. Cockburn) writes:
C>... image files produced
C> by scanning an autoradiogram are several MB each.  An ordinary PC
C> (even a 486) is just not going to be able to display and manipulate
C> these very fast.  I presume the same to be true of a Mac?

Macs such as the one mentioned, with large ram and *virtual* memory,
can handle several-meg images. We've done work with Scanning EM images
of 1.5 - 2.1 Meg on a mac cx or 2x (8 Meg ram and virtual memory).
Much of what determines 'acceptable speed' depends on what kind of processing
you do; 'Rolling Ball' flattening (of shading errors) can take minutes on
our macs. The FX is faster, but not by light-years. Simpler operations-
subtractions, thresholding, object analysis - take on the order of seconds
per step.

C>The two systems I mentioned use proprietary image analysis software
C>to locate bands and lanes automatically, calculate migration distances
C>and MWs automatically, and do fragment pattern matching.  ...
C>I don't know if equivalent ... software is available for other platforms. 

Much depends on the budget for the task. If effort will be substituted for $$,
(ie turn-key systems are out of the question) then NIH Image is
a versatile program to look at. It has a pascal-like macro language
that makes it increadibly easy to perform complex sequences of processing
steps that were not written into the original program.
There exist some example macros for gel analysis, and these perform
some of the operations mentioned above. Pattern matching is something
I've not noticed in any of the macros. Image is available by ftp from
sumex-aim.stanford.edu and other sites. I believe some dedicated gel-analysis
software is available from ftp.ncsa.uiuc.edu.

C>Finally, all of the systems and devices that I looked at can read and
C>write TIFF files.  As long as your system has this ability, ...

NIH Image exchanges tiff, pict, raw, and other formats.

Bob Straubinger, Pharmaceutics, SUNY Buffalo
rms at acsu.buffalo.edu




More information about the Bio-soft mailing list