Posted by
George McNamara on
URL: http://confocal-microscopy-list.275.s1.nabble.com/confocal-detectors-and-deconvolution-tp7588223p7588229.html
*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopyPost images on
http://www.imgur.com and include the link in your posting.
*****
Hi Mike,
My comments are in addition to Vitaly's.
* EVERY confocal microscope and every widefield microscope should have
GPU deconvolution immediately downstream, routinely, with 'instant
gratification'. Both Leica HyVolution2 (with SVI Huygens under the hood,
mostly seamless) and new Olympus are checkbox or dropdown to turn on
deconvolution (Olympus GPU deconvolution just launched).
* Any Deconvolution artifacts are present in the raw data, just not so
obvious. Fluorescent beads of very high R.I. under water (R.I. 1.33) or
air (1.0) produce excellent artifacts. If you write an NIH grant
specifically to study these artifacts, don't expect to get funding. Near
perfect R.I. matching is now readily doable for both live cells
(
https://www.ncbi.nlm.nih.gov/pubmed/?term=boothe+iodixanol ... if you
have a Silicone oil objective lens) or fixed cells (R.I. ~1.518 if need
oil immersion using thermoFisher/Mol Probes Prolong Glass, can also
check out Marker Gene's Opti-Bryt, and check EMSdiasum's web site for
other mounting media).
* We just ordered our Olympus FV3000RS on IX83 stand, several bullet
points related to this purchase:
-- thanks to NIH 'confocal' study section and U.S. taxpayers for
approving our S10 shared instrumentation grant proposal.
-- We are getting four GaAsP detectors inside, plus two external
GaAs "NIR" PMTs (see Jim Pawley's handbook chp 2 fig 2.10 for QE curves:
GaAs is superior in >700 nm, though I don't know the entire electronics
chain ... and yes, I wish Olympus sold 90% Q.E. APDs for this, and APDs
could have fit in budget, we'll be ok). Seven laser lines, including 730
nm for the NIR stuff.
- RS = resonant scanner, yes, standard high res and RS fit in
budget. Ever since Jonathan Boyd (who moved from Leica to Medimmune)
showed me RS is superior when matching pixel dwell time (8000 Hx * 100
scans vs 800 Hz * 10 scans vs 80 Hz one scan), likely due to less
photobleaching, RS with a little averaging (or summing for HyD photon
counters) is the way to go, if field of view, zoom, ok.
- The Olympus software released in mid-April (2018) now is able to
use NVidia GPU. Negotiate with your salespeople (i.e. Leica [SVI Huygens
for HyVolution2, don't know if they also do their own thing for FALCON),
Olympus, Nikon may also be GPU, Microvolution, AutoQuant) on price of
the deconvolution AND what GPU card(s) is(are) going into the
workstation (might end up being sold a cheap GPU and then going out and
buying the right one ... make sure the case and power supply can handle
it/them). Also important to know if the workstation can handle say 3
Quadro GV100's (three double width PCIe cards) or say 4 V100's on the
motherboard, see for example,
https://www.supermicro.com/products/system/4u/7048/sys-7048gr-tr.cfm
(sure, one FV100 or one V100 is $9000, but one or more would enable
'instant gratification' on a possibly $500K confocal or $900K light
sheet rig ... I also note that 100Gbe Ethernet cards are $800 so if your
new PC quote is still 1gig, tell the salesperson to fix it ... sure,
also need more cards, and probably a switch and 100Gbe NAS to send the
data to).
- Olympus "SSU" ultrasonic stage is very nice. If you go with
Olympus, get it (or at least demo it and decide for yourself). You can
ask other companies if they have the same thing.
* I manage a Leica SP8 confocal microscope with 2 HyD detectors and
HyVolution2, works great (I see I have more edits),
http://confocal.jhu.edu/current-equipment/leica-sp8-confocal-microscope* looks like Nikon's spectral confocal unmixes in by fluorophore
(finally a big confocal company gets it right), whereas Leica (LAS X on
SP8, same as earlier LAS on SP5) and ZEN Black (which might or not be
improved in 'zen one') unmix brightest channel first (bad idea if ever
try to do controls, especially "fluorescence minus one" controls).
* Geoff Daniels (Leica confocal sales) explained FALCON ... very
impressive sounding. If you have the money, four FALCONhyD's should be
great. If you have only money for one or two of them, start with that
(the pulsed white light laser likely costs more than four HyDs). Ideally
get an X1 port to enable more detectors (i.e. 2 or more APDs for NIR).
* Zeiss LSM880 with Airyscan (I'm not sure why anyone would buy an 880
without Airyscan ... seems like a 780 and marketing added 100).
* There are a lot of very interesting "other scopes" out there, mention
three:
- Abberior Instruments - see
http://www.abberior-instruments.com/site/assets/files/1057/abberior_instruments_adaptive_optics_easy3d_sted_deep_imaging.png
- ISS Alba and Alba-STED.
- confocal.nl "rescan".
- whatever AO LLS thingy Eric Betzig recently published ... hopefully
whatever company eventually distributes it does not botch the product
rollout.
Note quite comprehensive table of microscope vendors is one of the
tables in my (open access) Unit
https://currentprotocols.onlinelibrary.wiley.com/doi/abs/10.1002/cphg.42which is slightly newer than my online table
http://www.geomcnamara.com/light-microscopy-websitesenjoy,
George
p.s. if making "channel unmixing" matrix, I think photon counting is the
way to go (HyD's), and do a lot of line and frame accumulation to get
the best data for computing the coefficients (from single stained
controls). For example, 16 line * 16 frame accumulations = 256 (the
current max in leica LAS X), which produces bigger numbers than my
experimental scanning 'standard of care' of 10 line accumulation (no
frame accum). Sure, I am assuming no photobleaching (If bleaching was a
concern, I would repeat, compare, and if no bleaching, add the two).
On 5/10/2018 11:19 AM, Vitaly Boyko 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.
> *****
>
> Hi Mike,
> there is no big difference between HyDs and GaAsP detectors. Leica SMD HyDs are cooled to 18 C, thus having ca. 400 dark counts versus 2K or more for the GaAsP.I am not aware of linear deconvolution algorithms for confocal image data, thus it seems to be "cosmetic" and/or "aesthetic", and not suitable for accurate intensity based image analysis.Recent integration of FALCON combined with FCS/RICS to the Leica SP8 may bring new era to biology letting biologists quantify their data and image biological molecules at their physiological concentrations. However, most of biologists seems to be "scared" of basic equations...If you are located in the US, Leica provides much better Service and Application support compared to Zeiss. I do not have much experience with Olympus and Nikon on the confocal side.
> If you have any specific questions, please let me know.
> VitalyMCCF/MSKCC
> On Thursday, May 10, 2018, 10:01:36 AM EDT, MODEL, MICHAEL <
[hidden email]> 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.
> *****
>
> Dear colleagues,
>
>
>
> we are looking for a new confocal microscope, and I have two questions so far.
>
> 1. Are Leica'a hybrid detectors (I am talking about SP8) in any way better than the standard GaAsP? I found an old discussion of this topic on this list, but technology may have improved since then.
>
> 2. Both Leica and Olympus 3000 use deconvolution to boost resolution. We haven't had enough time during the demos to test it thoroughly, but my impression is that some features "revealed" by deconvolution might be artefactual, so I don't know if it is worth paying extra.
>
>
>
> Thank you
>
>
>
> Mike Model
>
> Kent State University
--
George McNamara, PhD
Baltimore, MD 21231
[hidden email]
https://www.linkedin.com/in/georgemcnamarahttps://works.bepress.com/gmcnamara/75 (may need to use Microsoft Edge or Firefox, rather than Google Chrome)
http://www.ncbi.nlm.nih.gov/myncbi/browse/collection/44962650http://confocal.jhu.eduJuly 2017 Current Protocols article, open access:
UNIT 4.4 Microscopy and Image Analysis
http://onlinelibrary.wiley.com/doi/10.1002/cphg.42/abstractsupporting materials direct link is
http://onlinelibrary.wiley.com/doi/10.1002/cphg.42/full#hg0404-sec-0023figures at
http://onlinelibrary.wiley.com/doi/10.1002/cphg.42/figures