Spinning vs fast scanner systems

[Charu Tanwar]

Dear All
Fast image acquisition can be done by resonant scanner (Nikon), Tandem
scanner (Leica), sim scanner (Olympus) and duo scanner (Zeiss).
Can anybody please let me know how comparable are the results in terms of
image quality. I may be highly wrong but can each one of them be compared
with the Yokogawa spinning disk system except the fact that these all are
single point scanning system and yokogawa spinning disk is a multi point
scanning system. How is the bleaching and phototoxicity effects???
I know that we are capturing signals through a high end camera system
(EMCCD) in Yokogawa and we use scanners in all these systems. But let us
talk about the end result i.e. an image with the required information. DO WE
REALLY GET COMPARABLE IMAGES because i think that these scanners are
evolved or generated as a requirement of fast imaging. It is quite obvious that
we ceratinly can not eliminate the requirement of fast imaging systems like
yokogawa spinning disk by introducing all these scanners keeping in mind the
benefits provided by yokogawa fast imaging system but what if we have these
scanner sytem and we want them to produce a quality image through fast
acquisition.
I have seen images acquired by Nikon's resonant scanner and Leica's tandem
scanner ( Calcium flux inside mammalian cell lines) and compared them with
the images acquired by yokogawa spinning disk, they seem to be quite OK (at
least in the presentation).

I have an idea of the speed of all these scanner systems in filter mode but
can any body let me know what is the speed of all these scanner systems
(recent models i.e. Nikon A1R, Zeiss 710/780, Olympus FV1000 and leica SP5)
in SPECTRAL MODE AT 512 x 512.

Thanks in advance

Charu Tanwar
Imaging Specialist
Advanced Instrumentation Research facility
Jawaharlal Nehru University
New Delhi
India

[Guy Cox]

There are key differences between the different methods of acquiring a
rapid confocal image.  Multipoint and line systems must have better
signal/noise at any given frame rate simply because they are are
acquiring many points at a time.  Multipoint and line systems will have
worse confocality since there is the possibility of crosstalk.  With
line scanners this is more or less independent of sample thickness, but
spinning disk systems confocality is pretty good on thin samples but
degrades when you reach a thickness at which the different cones of
illumination overlap.  This does not apply to the LaVision multipoint
multiphoton system since the path lengths to the different points are
different, so there can be no crosstalk.  But multiphoton does add a lot
to the cost.

                                         Guy

Optical Imaging Techniques in Cell Biology
by Guy Cox    CRC Press / Taylor & Francis
     http://www.guycox.com/optical.htm
______________________________________________
Associate Professor Guy Cox, MA, DPhil(Oxon)
Electron Microscope Unit, Madsen Building F09,
University of Sydney, NSW 2006
______________________________________________
Phone [phone]     Fax [phone]
Mobile [phone]
______________________________________________
      http://www.guycox.net
 

-----Original Message-----
From: Confocal Microscopy List [mailto:[hidden email]]
On Behalf Of Charu Tanwar
Sent: Wednesday, 6 January 2010 4:57 PM
To: [hidden email]
Subject: Spinning vs fast scanner systems

Dear All
Fast image acquisition can be done by resonant scanner (Nikon), Tandem
scanner (Leica), sim scanner (Olympus) and duo scanner (Zeiss).
Can anybody please let me know how comparable are the results in terms
of
image quality. I may be highly wrong but can each one of them be
compared
with the Yokogawa spinning disk system except the fact that these all
are
single point scanning system and yokogawa spinning disk is a multi point

scanning system. How is the bleaching and phototoxicity effects???
I know that we are capturing signals through a high end camera system
(EMCCD) in Yokogawa and we use scanners in all these systems. But let us

talk about the end result i.e. an image with the required information.
DO WE
REALLY GET COMPARABLE IMAGES because i think that these scanners are
evolved or generated as a requirement of fast imaging. It is quite
obvious that
we ceratinly can not eliminate the requirement of fast imaging systems
like
yokogawa spinning disk by introducing all these scanners keeping in mind
the
benefits provided by yokogawa fast imaging system but what if we have
these
scanner sytem and we want them to produce a quality image through fast
acquisition.
I have seen images acquired by Nikon's resonant scanner and Leica's
tandem
scanner ( Calcium flux inside mammalian cell lines) and compared them
with
the images acquired by yokogawa spinning disk, they seem to be quite OK
(at
least in the presentation).

I have an idea of the speed of all these scanner systems in filter mode
but
can any body let me know what is the speed of all these scanner systems
(recent models i.e. Nikon A1R, Zeiss 710/780, Olympus FV1000 and leica
SP5)
in SPECTRAL MODE AT 512 x 512.

Thanks in advance

Charu Tanwar
Imaging Specialist
Advanced Instrumentation Research facility
Jawaharlal Nehru University
New Delhi
India

No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 9.0.725 / Virus Database: 270.14.122/2590 - Release Date:
01/05/10 18:35:00

[James Pawley]

Dear Charu,

You bring up some good points. A lot of folk are
looking for the best way to image living
specimens fast but there are a lot of practical
variables to keep in mind.

Basically, it comes down the number of photons
involved. But the story is not a short one
(Please excuse length. At least it is shorter
than The Handbook)

Because Poisson statistics limits the accuracy of
any measurement of a photon signal, the
statistical accuracy of any image depends on the
number of photons detected. The "noisiness" of
most confocal images is caused by this quantum
mechanical conundrum.

Regardless of scanning system, if you want to
obtain an image of a given quality in 10x less
time, then you will need to detect photons at a
rate that is 10x higher. How can this be
achieved? Assuming that we have already chosen
the best objective (i.e., objective NA is fixed),
we can collect more signal by using more laser
power (up to a point), choosing a scan system
with more points and/or lower dead time, use a
larger effective pinhole size or employ a
photodetector with a higher quantum efficiency
(QE) (assuming that the photodetector used has a
very low read noise, which is true for PMTs and
EM-CCDs that are used properly).

Scan deadtime is the fraction of the scan time
when signal cannot be collected. On a single-beam
scanner, this is usually the time when the
horizontal scan mirror changes direction and
returns the spot to the "left" side to start a
new line. It can be about 30% of the line-scan
time. This wasted time can be reduced by scanning
in both directions but the signal still cannot be
collected at each end of the scan and also must
be digitally lined up to compensate for scan
non-linearity.

When a resonant scanner is used, the scan motion
is a sine wave rather than something approaching
a sawtooth (monodirectional) wave or pyramidal
(bidirectional) wave. Consequently the real-space
size of a fixed-time-length pixel in the center
of a scan will be (much) smaller than one near
either end of the line where the mirror slows
down to turn around. Although this geometrical
distortion can be linearized somewhat by varying
the pixel clock with the scan position (i.e.,
shorter pixel times in the center of the scan
where the mirror moves fastest), doing so does
not prevent the edges of the field from being
exposed to more light (and producing both more
signal photons and more damage). Scan systems in
which a mask of pinholes moves over a raster
pattern must have similar problems and these may
show up, for instance in the form bright lines
where the rasters from adjacent spots overlap (or
dark lines where they don't).

In terms of the number of scanning beams:

Clearly using 1,000 beams (about right for a
Yokogawa) should permit getting data at 1,000x
the rate of systems using a single beam
(especially as there is no dead time in a rotary
scan system, although the EM-CCD cannot be read
out continuously). However, in practice, using
1000 beams as bright as those used in a fast scan
single-beam instrument would cause almost
instantaneous bleaching of the specimen. In
practice, Yokos tend to be used with maybe 10x
more mW of power striking the specimen than with
a single-beam instrument. The latter is limited
to less than a mW in a 0.5µm spot by the need to
avoid singlet-state saturation. However, all
other things equal, more light at the excited
area of the specimen means either better looking
images or more of them in a given time period

Line illumination can be thought of as a row of
point illuminators: say 512, in a row. The Zeiss
LIVE-5 uses cylindrical optics to always create a
diffraction-limited line of illumination in the
specimen, but the detection slit is variable in
size. The larger slits (like larger pinholes) are
mostly useful to "spoil" the z-resolution, which
has the good effect of allowing signal from
near-focus planes to be added to that from
in-focus planes. More signal -> less apparent
noise.

Although the Olympus DSU (and perhaps the
CARV-2?) allows one to use disks having a variety
of patterns and slit-widths, on average the slits
tend to be larger and the overall "transparency"
greater than most other mask systems, probably
because the mercury arc sources they are used
with have much lower brightness than the lasers
used in other systems.

Although the Yokogawa uses fixed pinholes, their
effective size can be crudely adjusted by
utilizing objectives of differing magnification
(but the same NA. Usually, higher mag objectives
provide closer to "one Airy" pinhole sizes.).
There is however, some question as to the size
and shape of the illuminating spots in a the
Yokogawa. If we assume that the lenslets are
"perfect lenses", then they should form a
diffraction-limited spot centered on (and about
2x smaller than) each pinhole in the lower disk.
These spots will then be focused into
diffraction-limited spots in the specimen of
roughly the same size and shape as those produced
in a single-beam instrument. (I say roughly
because it is not quite clear how well the BFP of
the objective will be filled in either case.)
However, if the lenslet is not "perfect" (i.e,
has spherical aberration; is misfocused because
of chromatic aberration; or is somewhat
misaligned with its pinhole), then the light will
be focused into a larger "blob" at (or near) the
plane of the lower disk and this blob will be
first truncated by the pinhole in this disk and
then focused into a blob in the specimen that is
somewhat larger than the diffraction-limited spot
of the single-beam instrument.

Similar considerations apply to other swept-field
systems that use lenslets. These considerations
are worth mentioning because, depending on the
precision of the implementation, they give rise
to various forms of fixed-pattern noise. (i.e.,
as all lenslet-pinhole combinations may not be
identical, the part of the image scanned by one
lenselet may be brighter or dimmer than that of
another lenslet, something that is easily visible
as the faint, curved scan lines visible when
looking at a specimen of uniform fluorescence
with a disk-scanner. Likewise the scanned-line
systems can develop horizontal streaks if (even
out-of-focus) dust particles partially obscure a
segment of the detection slit.).

Although this artifact may be acceptable in the
Yoko if all 20,000 lenslets have a chance to scan
the field, it may be less acceptable as the
exposure becomes shorter and a each part of the
specimen is scanned only a few times. (i.e., For
instance, if it takes 1ms for a spot to move from
one side of the field to the other and that the
lenslet pattern repeats every "fields-worth",
then using an exposure of 1.1 ms will have the
effect that about 10% of the field has twice the
exposure of the rest [where the scan lines from
the first set of lenslets overlaps with that from
the next set], and there may be other differences
depending on the relative efficiency of the
individual lenslets, either in conveying light to
the specimen or back to the camera. Although the
Yoko disk rotates fast enough (1800 rpm, or about
3 ms, pattern) that this is a minor problem,
similar considerations may be important for some
of the other systems.).

So far however, the multi-beam/- systems are
broadly similar in terms of the total amount of
light "usually" striking the specimen. However,
as this amount can be adjusted, and more light
will give better looking images (less statistical
noise), it is always important to know how much
light was striking the specimen when a demo image
is provided. (i.e. you need to measure this.)

The fraction of the evoked fluorescent signal
that gets to the detector will depend on the
effective size of the detection slit/pinhole and
these too can be adjusted (with different
pinholes, different disks, or objectives with
different mags), at least to some extent. Again,
more photons will produce "better looking"
images. However, images with bigger pinhole/slits
will also have less depth resolution. The best
plan is to image something you know. Like a 0.2
µm fluorescent bead, and then look at the result
in 3D. (or as an xz image, to check the
z-resolution.)

Finally, the detectors: The LIVE_5 still uses a
conventional CCD and therefore has at least +/-
8-10 electrons of noise in each pixel
measurement. The best EM-CCDs, have essentially
no read noise but the multiplicative noise of the
electron multiplier register effectively reduces
the raw QE of the detector by 50% (compared to
that of the a normal CCD, such as that used in
the LIVE-5). Even so, the QE of the best EM-CCDs
is still at least 5x better than that of the best
PMTs in the green/yellow and probably 10-15x
better as you go into the red.

Things to watch out for.

During demos, the easiest way to "make the
picture look better" is to use more laser power.
Buy (or make
http://www.repairfaq.org/sam/laserioi.htm#ioisg1)
a cheap laser power meter and measure the beam
power coming out of the lens (Do this with the
same relatively low mag, low-NA objective in each
case and account for possible differences in the
field of view: Larger field = more pixels, will
require more light.) Remember, it is power
reaching the specimen (not that out of the laser)
that causes bleaching and phototoxicty.

In addition, one can always make a noisy image
look better by using time or spatial averaging,
or by binning the output of the various CCD and
EM-CCD sensors. But, assuming that the pixel size
was originally chosen to meet the Nyquist
criteria,  binning reduces the spatial
resolution. A better solution is a (so far only
imaginary) system that "deconvolves" the the data
(in 3D or Gaussian filters in 2D) as it comes in.
This can produce a MARKED improvement in the
apparent noisiness of the final data (i.e., Once
you adjust the contrast/brightness up to a useful
level, you can get the same image quality with
10-20x less signal.)

Don't forget the importance of looking at a
specimen that you know: my favorite is Nothing.
Obstruct the laser (or turn it off) and then see
what your detector records (probably the room
light). With an EM-CCD, it should be <10
single-photoelectron counts per 512-pixel line
(make sure that the gain is turned up enough to
easily see these peaks in a line scan through the
data). With a good PMT, you should get a similar
result (<10 spikes in a 2 mS line scan).
Although, because of multiplicative noise, the
spikes will NOT all be the same size, you should
still be able to discriminate them from the
electronic background noise by eye.

Finally, know your specimen. A specimen with 10x
more dye will produce about 10x more signal and
look a lot better. If you start viewing a thin
specimen mounted in oil and situated close to the
coverslip with a 1.4NA oil lens, you will get a
much better signal (and a nicer picture!) than
you get when you switch to a wet specimen in the
same set up (because the water will cause
spherical aberration (SA) and also reduce the
effective aperture of your objective). Even with
a water lens, the image will be worse (because
the signal goes with NA squared and you are now
at NA1.2, not 1.4), especially if you don't
adjust the correction ring very carefully to make
sure that you have no SA (+/- 2µm or
water-replaced-by-glass)

Tell us what you find.

Happy 2010,

Jim Pawley


               **********************************************
Prof. James B. Pawley,                          Ph.  608-263-3147
Room 223, Zoology Research
Building,                          FAX
608-265-5315
1117 Johnson Ave., Madison, WI, 53706  
[hidden email]
3D Microscopy of Living Cells Course, June 12-24, 2010, UBC, Vancouver Canada
Info: http://www.3dcourse.ubc.ca/             Applications due by March 15, 2010
               "If it ain't diffraction, it must be statistics." Anon.

>Dear All
>Fast image acquisition can be done by resonant scanner (Nikon), Tandem
>scanner (Leica), sim scanner (Olympus) and duo scanner (Zeiss).
>Can anybody please let me know how comparable are the results in terms of
>image quality. I may be highly wrong but can each one of them be compared
>with the Yokogawa spinning disk system except the fact that these all are
>single point scanning system and yokogawa spinning disk is a multi point
>scanning system. How is the bleaching and phototoxicity effects???
>I know that we are capturing signals through a high end camera system
>(EMCCD) in Yokogawa and we use scanners in all these systems. But let us
>talk about the end result i.e. an image with the required information. DO WE
>REALLY GET COMPARABLE IMAGES because i think that these scanners are
>evolved or generated as a requirement of fast
>imaging. It is quite obvious that
>we ceratinly can not eliminate the requirement of fast imaging systems like
>yokogawa spinning disk by introducing all these scanners keeping in mind the
>benefits provided by yokogawa fast imaging system but what if we have these
>scanner sytem and we want them to produce a quality image through fast
>acquisition.
>I have seen images acquired by Nikon's resonant scanner and Leica's tandem
>scanner ( Calcium flux inside mammalian cell lines) and compared them with
>the images acquired by yokogawa spinning disk, they seem to be quite OK (at
>least in the presentation).
>
>I have an idea of the speed of all these scanner systems in filter mode but
>can any body let me know what is the speed of all these scanner systems
>(recent models i.e. Nikon A1R, Zeiss 710/780, Olympus FV1000 and leica SP5)
>in SPECTRAL MODE AT 512 x 512.
>
>Thanks in advance
>
>Charu Tanwar
>Imaging Specialist
>Advanced Instrumentation Research facility
>Jawaharlal Nehru University
>New Delhi
>India

--
               **********************************************
Prof. James B. Pawley,                          Ph.  608-263-3147
Room 223, Zoology Research
Building,                          FAX
608-265-5315
1117 Johnson Ave., Madison, WI, 53706  
[hidden email]
3D Microscopy of Living Cells Course, June 12-24, 2010, UBC, Vancouver Canada
Info: http://www.3dcourse.ubc.ca/             Applications due by March 15, 2010
               "If it ain't diffraction, it must be statistics." Anon.

[Paul Maddox]

I'm a little late to the party (went underground for holidays), so I hope I
don't rehash a lot of points already made.

I will make a plug for the Nikon/Prairie swept field confocal (SFC, no
commercial interests).  I was a long time user of the yokogawa confocal for
many projects ranging from fixed deconvolution to low light level (several
fluors) live cell imaging.  Two years ago I purchased an SFC system to
replace the yokogawa in my lab.  The reason for this is not because the
yokogawa failed in any way, on the contrary, I fully endorse these systems
as excellent for live cell imaging of complex structures.  However, I
repeatedly found that I could not get enough light for high spatial res fast
imaging of dim signals (eg kinetochore proteins in C. elegans embryos).  So
I switched to the nikon SFC.  The ability to use a slit to vastly increase
throughput while only slightly reducing resolution has changed the way I can
approach imaging questions.  Likewise, using a pinhole setting I can
basically reproduce everything I was doing previously with spinning disk,
therefore I have one instrument with more flexibility.

Having said that, we also use wide field imaging (specifically a delta
vision, Nikon or Zeiss wide field system, again no commercial interests) for
many jobs (vastly under-appreciated in my opinion).  And I have recently
demoed yokogawa systems from PE and Quorum and found them both to be
excellent.  

As others have pointed out there are single-beam resonant scanners which are
quite good.  I have only tried the nikon A1r and was really surprised at the
performance (qualitative, remarkably nice) and have not tried others.

Basically, for maximum flexibility, I highly recommend the SFC.  I have not
recently demoed the Visitech system.

As a bit of a correction to what Jim said, EM cameras are not virtually
noise free, in fact they are quite noisy at high gain (±50e).  I recommend
that you demo any multipoint scanner with a none-EM CCD (Orca-AG, Roper HQ2,
or new Andor equivalent) to get a true sense of the system throughput.
These are low noise, high res cameras which can not hide throughput problems
that EM cameras often do.

Best of luck in purchasing!  Fortunately there are many really good options
out there and I doubt, after a proper demo etc, that you will be
disappointed by any one of them.

Paul

Paul S. Maddox, PhD
Assistant Professor
Institute for Research in Immunology and Cancer
Dept of Pathology and Cell Biol, U. de Montreal
P.O. Box 6128, Station Centre-Ville
Montréal QC  H3C 3J7
CANADA

Courier:
2900, boulevard Édouard-Montpetit
Pavillon Marcelle-Coutu, Quai 20
Montreal QC  H3T 1J4
CANADA

[hidden email]
Ph: 514-343-7894
Fax: [phone]



On 1/6/10 2:12 PM, "James Pawley" <[hidden email]> wrote:

> Dear Charu,
>
> You bring up some good points. A lot of folk are
> looking for the best way to image living
> specimens fast but there are a lot of practical
> variables to keep in mind.
>
> Basically, it comes down the number of photons
> involved. But the story is not a short one
> (Please excuse length. At least it is shorter
> than The Handbook)
>
> Because Poisson statistics limits the accuracy of
> any measurement of a photon signal, the
> statistical accuracy of any image depends on the
> number of photons detected. The "noisiness" of
> most confocal images is caused by this quantum
> mechanical conundrum.
>
> Regardless of scanning system, if you want to
> obtain an image of a given quality in 10x less
> time, then you will need to detect photons at a
> rate that is 10x higher. How can this be
> achieved? Assuming that we have already chosen
> the best objective (i.e., objective NA is fixed),
> we can collect more signal by using more laser
> power (up to a point), choosing a scan system
> with more points and/or lower dead time, use a
> larger effective pinhole size or employ a
> photodetector with a higher quantum efficiency
> (QE) (assuming that the photodetector used has a
> very low read noise, which is true for PMTs and
> EM-CCDs that are used properly).
>
> Scan deadtime is the fraction of the scan time
> when signal cannot be collected. On a single-beam
> scanner, this is usually the time when the
> horizontal scan mirror changes direction and
> returns the spot to the "left" side to start a
> new line. It can be about 30% of the line-scan
> time. This wasted time can be reduced by scanning
> in both directions but the signal still cannot be
> collected at each end of the scan and also must
> be digitally lined up to compensate for scan
> non-linearity.
>
> When a resonant scanner is used, the scan motion
> is a sine wave rather than something approaching
> a sawtooth (monodirectional) wave or pyramidal
> (bidirectional) wave. Consequently the real-space
> size of a fixed-time-length pixel in the center
> of a scan will be (much) smaller than one near
> either end of the line where the mirror slows
> down to turn around. Although this geometrical
> distortion can be linearized somewhat by varying
> the pixel clock with the scan position (i.e.,
> shorter pixel times in the center of the scan
> where the mirror moves fastest), doing so does
> not prevent the edges of the field from being
> exposed to more light (and producing both more
> signal photons and more damage). Scan systems in
> which a mask of pinholes moves over a raster
> pattern must have similar problems and these may
> show up, for instance in the form bright lines
> where the rasters from adjacent spots overlap (or
> dark lines where they don't).
>
> In terms of the number of scanning beams:
>
> Clearly using 1,000 beams (about right for a
> Yokogawa) should permit getting data at 1,000x
> the rate of systems using a single beam
> (especially as there is no dead time in a rotary
> scan system, although the EM-CCD cannot be read
> out continuously). However, in practice, using
> 1000 beams as bright as those used in a fast scan
> single-beam instrument would cause almost
> instantaneous bleaching of the specimen. In
> practice, Yokos tend to be used with maybe 10x
> more mW of power striking the specimen than with
> a single-beam instrument. The latter is limited
> to less than a mW in a 0.5µm spot by the need to
> avoid singlet-state saturation. However, all
> other things equal, more light at the excited
> area of the specimen means either better looking
> images or more of them in a given time period
>
> Line illumination can be thought of as a row of
> point illuminators: say 512, in a row. The Zeiss
> LIVE-5 uses cylindrical optics to always create a
> diffraction-limited line of illumination in the
> specimen, but the detection slit is variable in
> size. The larger slits (like larger pinholes) are
> mostly useful to "spoil" the z-resolution, which
> has the good effect of allowing signal from
> near-focus planes to be added to that from
> in-focus planes. More signal -> less apparent
> noise.
>
> Although the Olympus DSU (and perhaps the
> CARV-2?) allows one to use disks having a variety
> of patterns and slit-widths, on average the slits
> tend to be larger and the overall "transparency"
> greater than most other mask systems, probably
> because the mercury arc sources they are used
> with have much lower brightness than the lasers
> used in other systems.
>
> Although the Yokogawa uses fixed pinholes, their
> effective size can be crudely adjusted by
> utilizing objectives of differing magnification
> (but the same NA. Usually, higher mag objectives
> provide closer to "one Airy" pinhole sizes.).
> There is however, some question as to the size
> and shape of the illuminating spots in a the
> Yokogawa. If we assume that the lenslets are
> "perfect lenses", then they should form a
> diffraction-limited spot centered on (and about
> 2x smaller than) each pinhole in the lower disk.
> These spots will then be focused into
> diffraction-limited spots in the specimen of
> roughly the same size and shape as those produced
> in a single-beam instrument. (I say roughly
> because it is not quite clear how well the BFP of
> the objective will be filled in either case.)
> However, if the lenslet is not "perfect" (i.e,
> has spherical aberration; is misfocused because
> of chromatic aberration; or is somewhat
> misaligned with its pinhole), then the light will
> be focused into a larger "blob" at (or near) the
> plane of the lower disk and this blob will be
> first truncated by the pinhole in this disk and
> then focused into a blob in the specimen that is
> somewhat larger than the diffraction-limited spot
> of the single-beam instrument.
>
> Similar considerations apply to other swept-field
> systems that use lenslets. These considerations
> are worth mentioning because, depending on the
> precision of the implementation, they give rise
> to various forms of fixed-pattern noise. (i.e.,
> as all lenslet-pinhole combinations may not be
> identical, the part of the image scanned by one
> lenselet may be brighter or dimmer than that of
> another lenslet, something that is easily visible
> as the faint, curved scan lines visible when
> looking at a specimen of uniform fluorescence
> with a disk-scanner. Likewise the scanned-line
> systems can develop horizontal streaks if (even
> out-of-focus) dust particles partially obscure a
> segment of the detection slit.).
>
> Although this artifact may be acceptable in the
> Yoko if all 20,000 lenslets have a chance to scan
> the field, it may be less acceptable as the
> exposure becomes shorter and a each part of the
> specimen is scanned only a few times. (i.e., For
> instance, if it takes 1ms for a spot to move from
> one side of the field to the other and that the
> lenslet pattern repeats every "fields-worth",
> then using an exposure of 1.1 ms will have the
> effect that about 10% of the field has twice the
> exposure of the rest [where the scan lines from
> the first set of lenslets overlaps with that from
> the next set], and there may be other differences
> depending on the relative efficiency of the
> individual lenslets, either in conveying light to
> the specimen or back to the camera. Although the
> Yoko disk rotates fast enough (1800 rpm, or about
> 3 ms, pattern) that this is a minor problem,
> similar considerations may be important for some
> of the other systems.).
>
> So far however, the multi-beam/- systems are
> broadly similar in terms of the total amount of
> light "usually" striking the specimen. However,
> as this amount can be adjusted, and more light
> will give better looking images (less statistical
> noise), it is always important to know how much
> light was striking the specimen when a demo image
> is provided. (i.e. you need to measure this.)
>
> The fraction of the evoked fluorescent signal
> that gets to the detector will depend on the
> effective size of the detection slit/pinhole and
> these too can be adjusted (with different
> pinholes, different disks, or objectives with
> different mags), at least to some extent. Again,
> more photons will produce "better looking"
> images. However, images with bigger pinhole/slits
> will also have less depth resolution. The best
> plan is to image something you know. Like a 0.2
> µm fluorescent bead, and then look at the result
> in 3D. (or as an xz image, to check the
> z-resolution.)
>
> Finally, the detectors: The LIVE_5 still uses a
> conventional CCD and therefore has at least +/-
> 8-10 electrons of noise in each pixel
> measurement. The best EM-CCDs, have essentially
> no read noise but the multiplicative noise of the
> electron multiplier register effectively reduces
> the raw QE of the detector by 50% (compared to
> that of the a normal CCD, such as that used in
> the LIVE-5). Even so, the QE of the best EM-CCDs
> is still at least 5x better than that of the best
> PMTs in the green/yellow and probably 10-15x
> better as you go into the red.
>
> Things to watch out for.
>
> During demos, the easiest way to "make the
> picture look better" is to use more laser power.
> Buy (or make
> http://www.repairfaq.org/sam/laserioi.htm#ioisg1)
> a cheap laser power meter and measure the beam
> power coming out of the lens (Do this with the
> same relatively low mag, low-NA objective in each
> case and account for possible differences in the
> field of view: Larger field = more pixels, will
> require more light.) Remember, it is power
> reaching the specimen (not that out of the laser)
> that causes bleaching and phototoxicty.
>
> In addition, one can always make a noisy image
> look better by using time or spatial averaging,
> or by binning the output of the various CCD and
> EM-CCD sensors. But, assuming that the pixel size
> was originally chosen to meet the Nyquist
> criteria,  binning reduces the spatial
> resolution. A better solution is a (so far only
> imaginary) system that "deconvolves" the the data
> (in 3D or Gaussian filters in 2D) as it comes in.
> This can produce a MARKED improvement in the
> apparent noisiness of the final data (i.e., Once
> you adjust the contrast/brightness up to a useful
> level, you can get the same image quality with
> 10-20x less signal.)
>
> Don't forget the importance of looking at a
> specimen that you know: my favorite is Nothing.
> Obstruct the laser (or turn it off) and then see
> what your detector records (probably the room
> light). With an EM-CCD, it should be <10
> single-photoelectron counts per 512-pixel line
> (make sure that the gain is turned up enough to
> easily see these peaks in a line scan through the
> data). With a good PMT, you should get a similar
> result (<10 spikes in a 2 mS line scan).
> Although, because of multiplicative noise, the
> spikes will NOT all be the same size, you should
> still be able to discriminate them from the
> electronic background noise by eye.
>
> Finally, know your specimen. A specimen with 10x
> more dye will produce about 10x more signal and
> look a lot better. If you start viewing a thin
> specimen mounted in oil and situated close to the
> coverslip with a 1.4NA oil lens, you will get a
> much better signal (and a nicer picture!) than
> you get when you switch to a wet specimen in the
> same set up (because the water will cause
> spherical aberration (SA) and also reduce the
> effective aperture of your objective). Even with
> a water lens, the image will be worse (because
> the signal goes with NA squared and you are now
> at NA1.2, not 1.4), especially if you don't
> adjust the correction ring very carefully to make
> sure that you have no SA (+/- 2µm or
> water-replaced-by-glass)
>
> Tell us what you find.
>
> Happy 2010,
>
> Jim Pawley
>
>
>                **********************************************
> Prof. James B. Pawley,                             Ph.  608-263-3147
> Room 223, Zoology Research
> Building,                           FAX
> 608-265-5315
> 1117 Johnson Ave., Madison, WI, 53706
> [hidden email]
> 3D Microscopy of Living Cells Course, June 12-24, 2010, UBC, Vancouver Canada
> Info: http://www.3dcourse.ubc.ca/      Applications due by March 15, 2010
>       "If it ain't diffraction, it must be statistics." Anon.
>
>
>> Dear All
>> Fast image acquisition can be done by resonant scanner (Nikon), Tandem
>> scanner (Leica), sim scanner (Olympus) and duo scanner (Zeiss).
>> Can anybody please let me know how comparable are the results in terms of
>> image quality. I may be highly wrong but can each one of them be compared
>> with the Yokogawa spinning disk system except the fact that these all are
>> single point scanning system and yokogawa spinning disk is a multi point
>> scanning system. How is the bleaching and phototoxicity effects???
>> I know that we are capturing signals through a high end camera system
>> (EMCCD) in Yokogawa and we use scanners in all these systems. But let us
>> talk about the end result i.e. an image with the required information. DO WE
>> REALLY GET COMPARABLE IMAGES because i think that these scanners are
>> evolved or generated as a requirement of fast
>> imaging. It is quite obvious that
>> we ceratinly can not eliminate the requirement of fast imaging systems like
>> yokogawa spinning disk by introducing all these scanners keeping in mind the
>> benefits provided by yokogawa fast imaging system but what if we have these
>> scanner sytem and we want them to produce a quality image through fast
>> acquisition.
>> I have seen images acquired by Nikon's resonant scanner and Leica's tandem
>> scanner ( Calcium flux inside mammalian cell lines) and compared them with
>> the images acquired by yokogawa spinning disk, they seem to be quite OK (at
>> least in the presentation).
>>
>> I have an idea of the speed of all these scanner systems in filter mode but
>> can any body let me know what is the speed of all these scanner systems
>> (recent models i.e. Nikon A1R, Zeiss 710/780, Olympus FV1000 and leica SP5)
>> in SPECTRAL MODE AT 512 x 512.
>>
>> Thanks in advance
>>
>> Charu Tanwar
>> Imaging Specialist
>> Advanced Instrumentation Research facility
>> Jawaharlal Nehru University
>> New Delhi
>> India
>
>
> --
>                **********************************************
> Prof. James B. Pawley,                             Ph.  608-263-3147
> Room 223, Zoology Research
> Building,                           FAX
> 608-265-5315
> 1117 Johnson Ave., Madison, WI, 53706
> [hidden email]
> 3D Microscopy of Living Cells Course, June 12-24, 2010, UBC, Vancouver Canada
> Info: http://www.3dcourse.ubc.ca/      Applications due by March 15, 2010
>       "If it ain't diffraction, it must be statistics." Anon.