What Paul Goodwin has said is correct.  The SNR for a true deconvolution
  system equipped with a cooled CCD camera is much higher than a confocal
  microscope equipped with a photomultiplier tube.  For a Scanalytics
  system, the cooled CCD camera recommended has a dynamic range of 12 bits
  or higher-- providing 4095 grey levels or more -- whereas a PMT is
  usually 8 bit, with 255 grey levels
  .
  For most problems where spatial and temporal resolution is important,
  wide-field capture of images followed by rigorous deconvolution is a
  better solution than the signal rejection technique of confocal
  microscopy.  In an independent comparison between several confocal
  microscope manufacturers and the Exhaustive Photon Reassignment (EPR)
  algorithm by Scanalytics (at the AQLM course held at Woods Hole each
  spring), the z-axis resolution (as measured by FWHM) of the Scanalytics
  system produced superior results to any of the other confocal microscopes
  (0.4 um for Scanalytics, 0.8 um for other confocal manufacturers).  For
  x-y resolution (also FWHM), resolution was equivalent between the two
  systems, both systems ultimately limited by the diffraction limitations
  of the microscope.  The specimen was composed of 0.1 um latex microbeads
  with a fluorescein-like spectrum.
 
  However, with ratio imaging, there are other issues to consider.  Most
  researchers looking at calcium flux are looking at millisecond (or
  faster!) events.  Some of the best cooled CCD cameras can only read out a
  few frames per second, and that over a fairly small area.  In addition,
  deconvolution usually requires imaging in the z-dimension over several
  optical depths (the Scanalytics system requires at least 5 planes for a
  restoration).  If events are occuring very rapidly, by the time the 5th
  plane is imaged, the first plane is probably no longer the same.  As a
  true quantitative deconvolution (eg. EPR) requires that that all signal
  present in a through-focus volume retain correct inter-plane
  registration, it becomes necessary to acquire the 5 planes in a shorter
  period of time than the event being studied (actually, in half the time
  according to Nyquist).  As the quantum efficiency and read-out rates of
  affordable CCD cameras continues to improve, it is slowly becoming
  possible to perform this level of imaging.  A word to the wise is to be
  extremely skeptical of claims made many manufacturers about the
  "resolution" capabilities of their instruments.  We are all limited by
  photon statistics which, in practice, ultimately govern the
  spatial/temporal resolution tradeoffs we must make.