advice on phosphorimager purchase

John Brunstein brunstei at UNIXG.UBC.CA
Mon Dec 5 02:55:42 EST 1994


	I beg to differ...16 bit D/A conversion DOES NOT in any way 
affect the image resolution as compared to 8 or 10 bit, as seems to be 
implied in some of the previous posts. What it does is split the recorded 
signal intensities into ~16384 'steps' while an 8-bit machine takes the 
same range of signal intensities and splits it into 256 'steps'.  This 
has no discernable effect on image quality.
	Where it DOES have an effect is in quantitation accuracy.  If one 
is attempting to compare the intensities of two signals which are close to 
one another, 16-bit machines (MD and Biorad) can distinguish a difference 
while the 8-bit (Fuji) may not.  How big of an effect is this?   Well, 
with the Fuji instrument the standard error in quantitation is about 4%, 
while with the 16-bit machines it is about 0.9%.  If your experimental 
techniques have a total uncertainty of <3% this is important; otherwise 
you won't be able to see a difference.
	There are, however, two differences in the image quality between 
the Fuji (BAS-1000) and the MD/Biorad machines.  The first, which has 
been mentioned to before on this group, is that the Fuji screen chemistry 
has a better signal/noise ratio and the machine uses a solid waveguide 
instead of optical fibres between the scanning assembly and the PMT which 
does the actual counting.  Both of these contribute to a better image.  
The second is that the BAS-1000 can ONLY scan at 200um resolution, 
whereas both the MD/Biorad machines can be adjusted to smaller pixel 
sizes.  This will in fact make a difference in image quality with very 
weak isotopes such as 3H, where little 'blurring' of the image from 
higher-energy emissions will occur.  A smaller effect will be noticed 
with 14C and even a little bit with 35S, but for 32P (and most 35S) 
applications increasing resolution beyond 200um does nothing except make 
rediculously large files.
	Finally, the 'resolution' really isn't that simple; making the 
stepper motor which positions the scan head go from 200um to 100 or 50um 
steps does not really give a 2x or 4x increase in resolution, due to the 
fact that the laser beam which excites the screen causes some decay of 
signal in the adjacent screen area; on the higher-resolution scans this 
is more of a problem, as each small pixel is closer to the center of the 
previous one than is the case for the larger pixels.
	I know that there are papers dealing with the 16 vs 8-bit 
conversion math and its accuracy if anyone is really interested; although 
I don't have the reference handy I am sure that Fuji would be more than 
happy to give anyone the particulars, as it is favourable to their 
system.  
	For what it's worth, we have now had our small MD imager for 
about 3 months, running on a Power Mac 7100/66 with image analysis on 
both the Power Mac and on a Mac IIci, with a 230MB MO drive for bulk 
storage.  The system has been getting a lot of use and everyone seems to 
be pretty happy with it; so far, no damaged screens or other problems.
	I hope this helps to clear up some of the apparent confusion 
between D/A conversion and signal quality.





 






More information about the Methods mailing list