*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****
Jim,
This just takes us back to my original calculation. If we are sampling a million points we need a million times as much light in each spot. We are not illuminating a ccd, we are collecting a signal from a PMT and putting ALL that signal in one pixel, whatever the size of the area that signal is actually coming from. So the number of pixels that area would represent is irrelevant - we should not multiply or divide. I got led astray by trying to follow your argument!
As the number of pixels in an image of a single sub-resolution dot at Nyquist we both know that the answer is infinity. So it's purely a matter of semantics. I have no quibble with 4 pixels across but we can't map a circle of diameter 4 pixels easily on a square grid, so I prefer my version. There is no light in the dark ring! As to the lost light from the rings - it's just background noise in the widefield image and losing it in the confocal version is only going to improve S/N, which in the end is what we are after. And of course if I'm actually taking a resolution test on sub-resolution beads I'd sample at twice Nyquist or even higher.
Guy
Guy Cox, Honorary Associate Professor
School of Medical Sciences
Australian Centre for Microscopy and Microanalysis,
Madsen, F09, University of Sydney, NSW 2006
-----Original Message-----
From: Confocal Microscopy List [mailto:[log in to unmask]] On Behalf Of James Pawley
Sent: Monday, 30 September 2013 1:19 AM
To: [log in to unmask]
Subject: Re: Photobleaching of confocal microscopy
*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****
>*****
>To join, leave or search the confocal microscopy listserv, go to:
>http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>*****
>
>Well, I guess I'd better explain mine. I was simply taking the idea
>that to get 'equivalent'
>(whatever that means) images between confocal and wide field in the
>same time one needs ~1 million x brighter illumination (for 1024x1024)
>since one is only collecting one pixel at a
>time. Jim is right to point out that there
>will be collateral damage, which I hadn't taken into account, but he
>should be multiplying rather than dividing. I assume that sampling at
>Nyquist one is irradiating 9 pixels, while collecting one:
> x
> xx
>xxx
> xx
> x
>
>(however that may come out on your computer).
>That gives us 2 pixels in the Airy disk radius.
>So our effective dwell time for bleaching purposes is 9x the pixel
>dwell time. Actually that isn't correct either, if you think about
>FWHM, it's really only about 5 times. In other words, to estimate the
>likely bleaching we need to multiply by 5, not divide by 10.
>
>Of course nobody does run a confocal under those conditions - we
>normally collect for a much longer frame time than in wide-field, so we
>don't need such extreme intensities. We also often accept a worse
>signal to noise ratio, not least because in a 3D projection we
>typically average multiple frames.
>
>
>Guy
>
>Guy Cox, Honorary Associate Professor
>School of Medical Sciences
>
>Australian Centre for Microscopy and Microanalysis, Madsen, F09,
>University of Sydney, NSW 2006
Surely, how much light you collect depends on your pinhole size. Set it to the size of the first dark ring and, assuming diffraction-limited optics, you will collect 80% of 80% (i.e., 64%) of the light emitted by any dye molecules that have been excited near the in-focus-plane, not 10%.
This is why we almost never use a truly "confocal" pinhole (i.e., one that is less than 1/10th Airy), even though the use of such a pinhole would provide a 40% increase in x-y resolution. But that is a whole other argument and I am sure that Guy isn't recommending the use of such a pinhole. It just doesn't let enough photons through.
And then I suppose that we could argue about whether this 36% of light loss is a bad thing or a good thing. True, it is never nice to waste signal, but is light that is excited by the outer rings of the excitation, signal or noise? Surely, if this emitted light were collected, it would be recorded as though the dye were located where it was actually not located. Maybe we are better not collecting it. (Besides, a similar displacement occurs when recording (though not exciting) light from a point object in widefield.)
Apart from this I stick to my claim that "if you want to stick to Nyquist, the image of an isolated point object should show significant counts in a spot at least 4 pixels wide" not 3.
(The "at least" is to cover the likely situation that the center of the "point" object emitting the light isn't located exactly at the corner where 4 adjacent pixels come together.)
Of course, it is extremely unlikely that we will collect enough photons to actually "see"
Rayleigh/Abbe resolution (a maximum of 25% drop in signal between Airy peaks) because this contrast is reduced to something less than 10% when you compartmentalize the nice continuous mathematical curves into measurements taken at only 3-4 pixels. Therefore, one is free to assume that the resolution one is aiming at is lower than Rayleigh/Abbe and consequently, one is justified in using fewer, larger pixels to improve the Poisson statistics.
On the other hand, apart from wasting storage space, I don't really see having slightly too many pixels as a problem. True, if we assume that a constant number of photons are emitted, more pixels means fewer photons per pixel and worse effects of Poisson Noise. However, this will all be reclaimed if the confocal data is properly deconvolved.
JP
>-----Original Message-----
>From: Confocal Microscopy List
>[mailto:[log in to unmask]] On Behalf Of James Pawley
>Sent: Sunday, 29 September 2013 1:57 AM
>To: [log in to unmask]
>Subject: Re: Photobleaching of confocal microscopy
>
>*****
>To join, leave or search the confocal microscopy listserv, go to:
>http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>*****
>
>>*****
>>To join, leave or search the confocal microscopy listserv, go to:
>>http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>*****
>>
>>The difference between 3, 4 or 5 orders of magnitude is simply the
>>pixel size one chooses for calculation. 1024 x 1024 gives 1,048,576
>>(actually that's 6 orders of magnitude),
>>512x512 gives a factor of 262,144, and 256 x 256
>>gives 65,536. I'm not trying to labour a
>>trivial point, I'm making the important one that you should NOT use
>>more pixels than you need to capture your information. One of my more
>>spectacular 3D images was of an extremely beam-sensitive specimen
>>taken
> >at 256x256x97.
>>Of course if you want to enter one of Mike Davidson's competitions
>>you'll need a high pixel count (and a stable sample) but for cutting
>>edge science the moral is just that - cut.
>>
>> Guy
>
>
>Perhaps I should explain my arithmetic.
>
>The Airy Disk covers an a larger area of the specimen than that which
>is later represented by one, Nyquist-sampled pixel.
>
>I estimate that Nyquist-sampled pixel is about 10x smaller than the
>area covered by "those pixels in the Airy disk that have 'significant
>counts'".
>
>The argument is as follows:
>
> If we accept the distance between the centre of the disk and the
>first dark ring as a reasonable guess of the "resolution," this
>distance should be represented by at least 2 pixels, and the distance
>across the central maximum (which is doing most of the damage) will be
>4 pixels. If the pixel boundaries were set up symmetrically around the
>centre, one might make a rough estimate of the disk as 4x4 pixels ,
>however, there would be very few counts in the 4 pixels on the corners,
>so maybe 12 pixels total.
>
>On the other hand, the distance from one dark-ring to the dark-ring on
>the other side should really be closer to 4.5 of or 5 pixels or a
>"gross first estimate of area" of 20 or 25 pixels. Against this we
>must realize that, in general, the disk won't be centered and, if even
>the centre pixel produces a recorded signal of say 20 counts, you don't
>have to get far from the centre for counts (and therefore damage to the
>specimen which is what we are talking about) to become fairly
>negligible.
>
>So I am happy to guesstimate that the
>Nyquist-sampled PDF has about 10 pixels "with significant counts" hence
>10 pixels in which "significant damage" will occur.
>
>Every assumption could be disputed, but the "true number (if there is
>one) is not 1 and is not 100.
>
>This leads to 1000 x1000/10 = 100,000.
>
>Of course, if we actually have a real AIry disk, (Usually we don't. The
>peak intensity is seldom ~30x higher than that of the first bright
>ring.) almost 20% of the power is still in the outer rings, and above
>arithmetic ignores the effect of this relatively low intensity
>light/damage.
>Real "Airy disk" spots focused inside real specimens probably have
>considerably more of the total light in the area of these "outer
>rings".
>
>Regards,
>
>Jim Pawley
>
>>Guy Cox, Honorary Associate Professor
>>School of Medical Sciences
>>
>>Australian Centre for Microscopy and Microanalysis, Madsen, F09,
>>University of Sydney, NSW 2006
>>
>>-----Original Message-----
>>From: Confocal Microscopy List
>>[mailto:[log in to unmask]] On Behalf Of James Pawley
>>Sent: Saturday, 28 September 2013 12:09 AM
>>To: [log in to unmask]
>>Subject: Re: Photobleaching of confocal microscopy
>>
>>*****
>>To join, leave or search the confocal microscopy listserv, go to:
>>http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>*****
>>
>>>*****
>>>To join, leave or search the confocal microscopy listserv, go to:
>>>http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
>>>*****
> >>
>>>Dear Listers,
>>>
>>>I have a basic question about the relationship between
>>>photobleaching, pixel dwell time and intensity of excitation beam in confocal microscopy.
>>>
>>>Review paper "Optical Sectioning microscopy" in Nature Methods 2005
>>>said about bleaching of confocal microscopy.
>> >
>>>"The probability that a molecule bleaches depends on its exposure to
>>>the excitation light. This is the product of the irradiance a
>>>molecule receives and the time it receives it.
>>>Thus, a molecule that receives irradiance I1
>>> of duration t1 is as likely to bleach as one that receives twice
>>>the irradiance (2I1) for half the time (t 1/2)."
>>>
>>>Although this is well accepted among confocal microscopy user, I am
>>>wondering if there are any data or paper to support this notion.
>>>I also would like to know how exactly same the photobleaching of
>>>these two cases are.
>>>
>>>If someone knows any information, we appreciate comments.
>>>
>>>Best Regards,
>>>Shigeo Watanabe, HPK
>>
>>Dear Shigeo,
>>
>>Dave piston did a lot of work on bleaching rate for 1- and 2-photon
> >excitation about a decade ago, using dye dissolved in oil and
> squished
>>into a very thin layer by prolonged pressure. He found that above a
>>certain power density in the spot, the bleach rate was greater than
>>what one would assume from a linear extrapolation. Of course, such
>>experiments are complicated by the fact that the intensity varied
>>markedly across the focused spot (so maybe it was being "super
>>bleached" in the centre of the spot but that this was masked by
>>emissions from the outer parts of the spot where bleaching was only
>>linear) , but because his dye was uniform and thin, at least he could
>>avoid the complication that dye might diffuse in or out of the
>>exposure area or that, although the dye might bleach in the focus
>>plane, signal might still be detected from so-far unbleached dye
>>located above or below this plane where the intensity was lower
>>(because the spot was not yet fully focused and hence larger in
>>diameter and so less intense).
>>
>>On the other hand, bleaching behavior in oil is likely to be different
>>than that in water, especially if the water contains dissolved oxygen.
>>(Bleaching is seldom a matter of photon absorbed -> molecule damaged,
>>it is more commonly, photon absorbed, excited molecule interacts with
>>oxygen to produce reactive oxygen species which may diffuse some
>>distance before interacting to produce some permanent chemical change.
>>One should note that this chemical change will probably only
>>occasionally involve the original dye molecule.)
>>
>>Guy speaks of the peak intensity as being 3-4 orders of magnitude
>>greater in confocal than in wide field. I usually guess 5 orders
>>(assuming something close to Nyquist sampling, you are illuminating
>>about 10 pixels out of a raster that I assume to be 1,000x1,000). The
>>early editions of The Handbook had chapters by the Watt Webb group
>>that showed that with most dyes, you reached very close to saturation
>>at about 1mW of laser power in a high-NA spot.
>>
>>In other words, I disagree with the "short version of bleaching" in
>>the Nature article that you quote. It may be OK when the laser power
>>in the spot of around one µW, but not at one mW or when one is working
>>at low NA and therefore the spot is bigger and so less intense.
>>
>>There are several chapters (particularly those by Karsten Koenig,
>>Alberto Diaspro and Jurek
>>Dobrucki) devoted to bleaching in the third edition of the Handbook.
>>
>>Regards,
>>
>>Jim Pawley
>>--
>>James and Christine Pawley, 5446 Burley Place (PO Box 2348), Sechelt,
>>BC, Canada, V0N3A0, Phone 604-885-0840, email <[log in to unmask]> NEW!
>>NEW! AND DIFFERENT Cell (when I remember to turn it on!)
>>1-604-989-6146
>
>
>--
>James and Christine Pawley, 5446 Burley Place (PO Box 2348), Sechelt,
>BC, Canada, V0N3A0, Phone 604-885-0840, email <[log in to unmask]> NEW!
>NEW! AND DIFFERENT Cell (when I remember to turn it on!) 1-604-989-6146
--
James and Christine Pawley, 5446 Burley Place (PO Box 2348), Sechelt, BC, Canada, V0N3A0, Phone 604-885-0840, email <[log in to unmask]> NEW! NEW! AND DIFFERENT Cell (when I remember to turn it on!) 1-604-989-6146
|
