Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
>Search the CONFOCAL archive at
>http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
>
>There are two major factors beyond somewhat thicker optical sections at
>work here. First, light is being collected from each pixel for a much
>longer period of time than is the case in a point scanner, even if the
>point scanner is collecting data very slowly. For instance, collecting
>at a maximum rate of 60 images a second (512x512), each pixel has a
>dwell of ~ 32us on the 5-Live; a conventional point scanner collecting 1
>image every 2 seconds has a dwell of about 7us/pixel. Secondly, ccd
>based detectors are much less noisy than conventional uncooled PMT
>detection.
I think that the last statement is a bit oversimplified.
"Less Noisy" is a slippery concept. Images can look less noisy for a
number of reasons.
One reason is that more photons were detected to produce the image
and therefore the ratio between this signal and Poisson Noise
associated with it (i.e., the sqrt of the signal, expressed as the
number of photoelectrons/pixel actually detected) is higher. As this
noise becomes less obvious as the quantum efficiency of the detector
improves, CCDs with a QE up to 90%, do better with a given photon
signal than PMTs with an effective QE of 10%.
However, this is not the whole story.
If signal levels are very low, as they often are in confocal
fluorescence microscopy, then the "read noise" of the photodetector
also becomes important. Here the PMT is far superior to any CCD other
than an EM-CCD. PMT's and EM-CCDs are likely to have a noise level of
one photoelectron (PE) in every 250 pixels whereas normal SLOW-SCAN
CCDs, with a read noise of +/-10 electrons/pixel*, usually can't tell
the difference between 20 photons and none. Now, if the QE of the CCD
really is 10x greater, the photon signal that made these 20 PEs,
would only make 2 PEs in the PMT but there is still a major
difference, especially if the signal in most pixels is zero.
I capitalize slow scan because the read noise of a normal CCD
increases with the sqrt of the pixel clock.
So far, Zeiss has not divulged the RMS noise/pixel level of the
linear CCD in the LSM5 but as far as we can tell, it apparently is
NOT an EM-CCD. We do know that, at 120 frames/second, it apparently
reads out at 30 MHz and is is NOT slow scan. A read noise of +/- 30
electrons/pixel would be very good performance at this speed and it
might be much higher.
So, back to the noisiness of the CCD: The CCD is less noisy IF the
signal level is more than a few 10s of photons/pixel. In single-beam
scanning confocal, the signal level in each "bright" pixel is often
only 4-8 photoelectrons.
The the noisiness of most un-deconvolved confocal data stems not from
noise generated in the PMT, but because the PEAK detected signal
represents only a few photoelectrons/pixel.
Cheers,
Jim P.
* I know that the mfgrs claim lower noise levels but the cameras
usually don't seem to really provide this level of performance.
--
**********************************************
Prof. James B. Pawley, Ph. 608-263-3147
Room 223, Zoology Research Building,
FAX 608-265-5315
1117 Johnson Ave., Madison, WI, 53706
[log in to unmask]
3D Microscopy of Living Cells Course, June 11-23, 2005, UBC, Vancouver Canada
Info: http://www.3dcourse.ubc.ca/ Applications due by March 15, 2004
|
