Re: CCD vs sCMOS

[Kenneth Chen]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hello,

I am in the same situation. Our lab is purchasing a new microscope and deciding between
EMCCD, sCMOS, and CCD cameras. However, our applications are a bit more specific. In
particular, we will be using the scope to do widefield fluorescence live-cell imaging of yeast
cells, with time courses that cover the lifespan of the cells (approx 3-4 days if not damaged
by phototoxicity). We will be quantifying protein levels (multiple fluorophores) to answer
questions regarding cellular noise and information transmission. We will also be looking at
structure/localization of organelles. With this particular application, does anybody have any
thoughts/experience?

From my own tests, it seems like the EM gain function is great for boosting the signal over
background of homogenously dispersed proteins in the cell. However, it seems to be
detrimental to viewing organelle structure/localization where the background is the
cytoplasm of the cell, rather than the space between cells. However, I have limited
experience in microscopy and this may be more the result of user error.

Thanks,
Ken Chen

[George McNamara]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi Ken,

get a demo or demos ... should be between EMCCD and sCMOS ... and
especially since you are a microscopy newbie, microscope rep vs
microscope rep.

You may also want to ask yeast microscopy experts what they are finding
best - and ask your local colleagues! For example, you could read Sandy
Lemmon's recent papers (ex. http://www.ncbi.nlm.nih.gov/pubmed/25557545)
and/or ask her (I used to work at UMiami).

More suggestions:

* consider going "state of the art" on FPs, for example, mNeonGreen
(even if it costs $ for a license), 3x brighter than EGFP (latter is
soooo 1996 ... are you still driving a 19 year old car?). If they
(Allele Biotech) don't have a yeast codon/expression optimized version
that is documented to work, offer to 'evaluate' but at a steep discount.
* if not critical to answering your primary question, multimerize and
localize the FPs (ex. nucleolus localized-3x-mNeonGreen). ... more on
this at
http://works.bepress.com/gmcnamara/42/
http://works.bepress.com/gmcnamara/65/
(there is additional content on the site) and
http://www.ncbi.nlm.nih.gov/pubmed/25307933 (if you are not using Cas9,
good entry point).
* multiplex, a nice example is  http://www.ncbi.nlm.nih.gov/pubmed/24949979
* more multiplex: consider buying multiple cameras and device(s) to
split the light path, and generate high performance FP's that all excite
at the same wavelength ... this will get you the most "bang for your
excitation photon". One potential device,
http://www.cairn-research.co.uk/catalogue/detection/image-splitters/product/multicam 
(if your rep is an "old school" microscopy sales rep, they may not have
even heard of Cairn ... if so, try another rep). For examples,
CyPet-YPet and CY11.5 (http://www.ncbi.nlm.nih.gov/pubmed/16700538) are
both over 90% FRET efficient.
* consider www.microvolution.com GPU deconvolution and (hopefully
microvolution will add it "soon" - and if you talk with Marc or
Cassandra, mention I encouraged you to ask about it) what I call "joint
spatial deconvolution and spectral unmixing" - see
http://www.ncbi.nlm.nih.gov/pubmed/18339754  ... 10x improvement in
signal-to-noise ratio (and not just for FRET).

Enjoy,

George
p.s. if you find all these tips useful, consider downloading
Tiki_Goddess from http://works.bepress.com/gmcnamara/70/  and printing a
poster (on fabric, non-fading ink, 44x84 inches is good) and putting in
your microscopy room. You can admire classic Tiki_Goddess (only 8 Mb
download) at http://home.earthlink.net/~tiki_goddess/TikiGoddess.jpg   
(click to zoom). Was also the May 2015 Biotechniques cover, as discussed
at http://works.bepress.com/gmcnamara/71/   (by the way #72 is a cool
microscope tip ... too bad you work on yeast that hardly move - #73 is
even cooler).
For that matter, Halloween is not that far away - you can download and
print http://works.bepress.com/gmcnamara/25/  (this can be on a desktop
printer).





On 9/1/2015 3:54 PM, Kenneth Chen wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Hello,
>
> I am in the same situation. Our lab is purchasing a new microscope and deciding between
> EMCCD, sCMOS, and CCD cameras. However, our applications are a bit more specific. In
> particular, we will be using the scope to do widefield fluorescence live-cell imaging of yeast
> cells, with time courses that cover the lifespan of the cells (approx 3-4 days if not damaged
> by phototoxicity). We will be quantifying protein levels (multiple fluorophores) to answer
> questions regarding cellular noise and information transmission. We will also be looking at
> structure/localization of organelles. With this particular application, does anybody have any
> thoughts/experience?
>
>  From my own tests, it seems like the EM gain function is great for boosting the signal over
> background of homogenously dispersed proteins in the cell. However, it seems to be
> detrimental to viewing organelle structure/localization where the background is the
> cytoplasm of the cell, rather than the space between cells. However, I have limited
> experience in microscopy and this may be more the result of user error.
>
> Thanks,
> Ken Chen
>
>    

--



George McNamara, Ph.D.
Single Cells Analyst
L.J.N. Cooper Lab
University of Texas M.D. Anderson Cancer Center
Houston, TX 77054
Tattletales http://works.bepress.com/gmcnamara/42

[Gerhard Holst]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hello Ken,

Despite the fact that I could be biased, because we don't make emCCD cameras any longer, I try to answer. If the weak signals consist of more than 20 photons per pixel, then the signal-to-noise of sCMOS cameras should be better. The so called intra-scene dynamic, that's also describes the ability of the camera to separate and distinguish light levels is far better in sCMOS, since it is based on the calculation of fullwell or saturation capacity of the pixel of the image sensor divided by the readout noise (the smallest portion that could be distinguished). The emCCD usually have a larger fullwell but also a larger readout noise. I know that the readout noise of the emCCD usually is given below 1 electron due to the gain, but the gain also minimizes the usable fullwell, because at a much lower signal the image sensor reaches fullwell, therefore this does not improve the dynamic.
Therefore, your prospected separation of details on a bright background might benefit from a higher dynamic. If most of your measurements are so weak with respect to the signal, that you expect very long exposures times, then a cooled CCD might still do good (but like always with the emCCD at a poor frame rate).

Unfortunately the CCDs are on their way down the hill.

with best regards,

Gerhard
_______________________________

Dr. Gerhard Holst
Science & Research
PCO AG
Donaupark 11
93309 Kelheim, Germany
fon [phone]
fax [phone]
mob [phone]
[hidden email]
www.pco.de


-----Ursprüngliche Nachricht-----
Von: Confocal Microscopy List [mailto:[hidden email]] Im Auftrag von Kenneth Chen
Gesendet: Dienstag, 1. September 2015 22:54
An: [hidden email]
Betreff: Re: CCD vs sCMOS

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hello,

I am in the same situation. Our lab is purchasing a new microscope and deciding between
EMCCD, sCMOS, and CCD cameras. However, our applications are a bit more specific. In
particular, we will be using the scope to do widefield fluorescence live-cell imaging of yeast
cells, with time courses that cover the lifespan of the cells (approx 3-4 days if not damaged
by phototoxicity). We will be quantifying protein levels (multiple fluorophores) to answer
questions regarding cellular noise and information transmission. We will also be looking at
structure/localization of organelles. With this particular application, does anybody have any
thoughts/experience?

From my own tests, it seems like the EM gain function is great for boosting the signal over
background of homogenously dispersed proteins in the cell. However, it seems to be
detrimental to viewing organelle structure/localization where the background is the
cytoplasm of the cell, rather than the space between cells. However, I have limited
experience in microscopy and this may be more the result of user error.

Thanks,
Ken Chen

[Colin Coates]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

*commercial response*
Hi Ken,
I largely agree with Gerhard on this one, except in our experience over
thousands of EMCCD users, people in practice rarely have an issue with
EMCCD dynamic range, especially as it can be fine tuned by careful control
of EM Gain. The max dynamic range of an EMCCD can be accessed by
matching to the EM Gain to approx. the readout noise at the selected
readout speed (that value accessible from the datasheet).
Our company sells a healthy balance of both EMCCD and sCMOS cameras
and the technology decision is usually down to the need to access the very
low light regime to which EMCCDs remain more suited. This isn't just a
theoretical statement, this is a reflection of many user's experiences.
Sometimes this is driven by use of an imaging modality that is designed to
limit the background, such as spinning disk confocal or TIRFM, which can in
turn be complimented by use of a 'negligible' read noise EMCCD. Some of
our users make use of EMCCDs to enable them to reduce the illumination
intensity and exposure times as much as possible, thus reducing rate of
phototoxicity. Some like to use EMCCDs to reduce fluorophore
concentrations as much as possible in order to minimally affect cell
physiology.
However, I suspect that in your application the intracellular background
level is reasonably high and your own initial tests are reflecting a
legitimate  conclusion, i.e. that you are in a photon regime where sCMOS
could do an acceptable job.    

All the Best,
Colin

Dr Colin Coates
Product Manager
Andor Technology