How to choose an optical imaging
system for recording fast brain activity.
Optical imaging technologies are
becoming more widely used in the fields of Neuroscience and Physiology.
To choose an appropriate system for one's application the following
aspects should be considered:
Dynamic Range and
Signal-to-Noise Ratio
An important figure of merit for an
optical recording system is dynamic range. Dynamic range can be
specified in db, in bits, or as an exponent (e.g. 100 db = 17 bits = 105
). The dynamic range determines the size of smallest fractional
intensity change that can be measured. For example, a dynamic range of
100 db would allow one to measure a signal with a fractional change,  I/I, of 10-5. The smallest signal that can be
measured with a dynamic range of 60 db is 10-3. Quite often,
optical signals of biological interest are small (10-2 - 10-4).
A large dynamic range ensures that the apparatus can measure small
signals with an optimal signal-to-noise ratio.
The signal-to-noise ratio is linearly related to the light intensity
when dark noise dominates (low light) and related to the square root of
the light intensity when shot noise dominates (higher light). At low
light levels, as in fluorescence measurements from a single neuron or
cardiac cell,
the performance of a cooled CCD camera is better than a photodiode
array (low spatial resolution) or a CMOS camera (high spatial
resolution). But unlike a CCD camera, the diodes or a CMOS camera won't
saturate at high light levels, as in fluorescence or absorption
measurements from an intact heart, vertebrate brain, brain slice, or a
ganglion. As a result, CMOS cameras and photodiode systems with
parallel amplification generally have a larger dynamic range, thereby
yielding a better signal-to-noise ratio at high light levels. While the
dynamic range cannot be larger than the effective resolution of the
analogue-to-digital converter, it can be considerably smaller if, for
example, saturation limits the number of measured photons. Dynamic
range can also be reduced by extraneous noise.
The following is a listing of
the dynamic range required for several voltage-sensitive-dye
applications:
· Population signals from
intact brains and brain slices -- 12-15 bits.
· Multiple individual neurons in invertebrates -- 15-17 bits.
· Multiple sites on a single cell -- 12 bits.
· Tissue cultured neurons -- 12-17 bits.
· Intact hearts -- 10 bits.
Dark Noise
The dark noise (the noise of the
system with no light) affects the signal-to-noise ratio at low light
levels. The dark noise is lowest in a cooled CCD. CMOS cameras and
Photodiode arrays have a larger dark noise.
Temporal Resolution (Frame Rate)
A frame rate of at least 1 kHz is
required for recording action potentials and other fast neuronal or
cardiac signals.
Spatial Resolution
Good spatial resolution is important
for obtaining a high quality image of activity. In general, increasing
the number of pixels will increase the spatial resolution. However, in
many preparations, light scattering or signals from out-of-focus light
are substantial. Thus, a large number of pixels may not actually
increase the spatial resolution, but rather will only reduce the amount
of photons each pixel receives, thereby reducing the signal-to-noise
ratio.
Software
The software should be comprehensive
and easy to use.
For additional reading, see:
- Wu, J-Y. and Cohen, L.B. (1993). Fast multisite
optical measurement of membrane potential. In: Fluorescent and
Luminescent Probes for Biological Activity, W.T. Mason eds, Academic
Press, London. pp 389-404.
- Grinvald, A., Frostig, R.D., Lieke E. and Hildesheim, R. (1988).
Optical imaging of neuronal activity. Physiological Reviews, 68(4):
1285-1365.
Comparing technologies for
improving S/N of CCD cameras:
Two different approaches could be taken to improve
the signal-to-noise-ratio of a measurement at extremely low light
levels, where the read noise of the chip is usually higher than the
shot noise:
One could increase the signal (EMCCDs), or alternately reduce the noise
(MIT/LL CCD).
EMCCDs use an electron bombardment technique to
amplify the signal above the
readnoise floor of the CCD. The penalties for that are:
|
1. Spurious noise events -
approximately one corrupt pixel in 1000, randomly
in space and time is the very best that has been done. |
|
2. Sqrt(2) penalty in shot
noise. i.e. shot noise is sqrt(2N) instead of
sqrt(N). |
|
3. Variable gain - gain
can vary pixel to pixel, line to line, frame to
frame. Impossible to calibrate in an absolute sense, even in the same
frame. |
|
4. Electron bombardment
process is destructive and the CCD degrades over time. |
On the other hand, the low read noise MIT/LL CCD does not have any of
these
issues. The chip is using an ultra-low readnoise output while boosting
the responsivity to 40 uV/e-. This process is simple, reliable, linear,
non-destructive and constant.
Comparing signals and images from a 2-photon
microscope and an ordinary fluorescence microscope (equipped with a
CardioCCD-SMQ camera)
A. Calcium Green-1 signals measured
with a 2-photon microscope and an ordinary fluorescence microscope with
a CardioCCD-SMQ camera
B. Images
Olfactory receptor neuron nerve
terminals in the mouse olfactory bulb were stained with Calcium
Green-1.
Figure A. at top shows odorant
elicited signals in in vivo preparations using the two imaging systems;
in both cases the signals were the spatial average of the light from
one glomerulus. The signal-to-noise ratio for the CardioCCD-SMQ
recording is much larger. This results from a larger number of measured
photons in the CardioCCD-SMQ recording. Similar results were obtained
in comparisons made on five mice.
Moreover, four trials were
averaged in the 2-photon measurement shown in the figure while the
result from CardioCCD was from a single trial. In addition, the
numerical aperture (NA) of the lens used for the 2-photon measurement
was 0.8 while that used in the ordinary microscope was only 0.5. If a
correction for these two factors is applied, the 2-photon measurement
would have a signal-to-noise ratio six times smaller than that shown.
Factors that contribute to the
relatively small number of photons in the 2-photon measurement are:
1. The incident light in the
2-photon microscope interacts with many fewer dye molecules because
only a thin section receives high intensity illumination.
2. Calcium Green-1 has a 2-photon cross section which results in a low
optical efficiency. This low efficiency can not be overcome by
increasing the incident intensity because higher intensity will heat
the preparation.
The images formed by the two kinds of microscope are shown in the
bottom figure B. (the image made with the ordinary microscope covers a
2x larger area of the bulb). The 2-photon image is the total intensity;
the ordinary microscope image is the image of the signal. The
advantages of 2-photon microscopy are clear; rejection of scattered
light and very shallow depth of focus results in much better x-y and
z-axis resolution. Clearly the two kinds of imaging systems are optimal
for different niches in the parameter space of imaging.
The olfactory receptor neuron staining procedure in the mouse
(Wachowiak and Cohen, 2001) followed one developed by Friedrich and
Korsching (1997) for zebrafish. (Data provided by Rainer Friedrich,
Matt Wachowiak and Larry Cohen, Max Planck Institute for Medical
Research, Heidelberg, and Yale University, New Haven.)
|
