Resolution of a Microscope

00:00:12.16 My goal in this discussion
00:00:14.29 is to leverage
00:00:17.27 the insights you have in diffraction
00:00:21.16 and the point spread function
00:00:24.02 to understand how to practically
00:00:27.22 take advantage of this
00:00:29.07 as one does optical imaging.
00:00:31.25 So, resolution of course
00:00:34.11 is the question of
00:00:36.18 how much one can resolve
00:00:41.06 detail in an image,
00:00:42.05 and it relates basically to the grain of the image
00:00:44.21 -- we've talked about this in previous discussions.
00:00:48.18 If we take a pair of points of objects,
00:00:54.23 resolution could be asked,
00:00:56.08 how close could those points get
00:00:58.05 and you would still have confidence
00:01:00.05 in saying that they're two separate points
00:01:02.00 and not one?
00:01:03.07 And you can think about this in many ways,
00:01:05.10 but you can see that there's no absolute way,
00:01:09.28 no absolute criterion
00:01:12.23 that would allow you to say that,
00:01:14.14 oh yes, they are definitely two points
00:01:16.05 or they're definitely one,
00:01:17.26 and so the criterion that has been used
00:01:19.24 is somewhat arbitrary,
00:01:21.18 but it is one that is used by microscopists,
00:01:25.06 generally, it's called the Raleigh criterion,
00:01:26.26 and maybe to get started here
00:01:28.24 let me just show you a pair of points
00:01:30.26 that are moved closer and closer together
00:01:33.00 and we'll talk about
00:01:35.07 why the Raleigh criterion is a valuable way of looking at this.
00:01:37.23 Raleigh was a physicist
00:01:40.17 who came up with this as an idea.
00:01:42.21 There's nothing pure about it
00:01:45.07 but it is a very nice and straightforward way
00:01:48.17 of thinking about resolution.
00:01:49.20 So, how close can two point sources be
00:01:52.26 and their images still be seen
00:01:55.03 as two points.
00:01:56.12 So, this isn't about how close they are
00:01:58.10 in real space,
00:01:59.28 it's how close in image space they are
00:02:03.05 and still, can we resolve them as separate objects.
00:02:06.20 So, here are a series of points
00:02:08.06 and, now, if we start with the one up there,
00:02:10.25 all the way here,
00:02:13.00 no doubt those are two separate points
00:02:15.01 -- they're separated, there's a dark area between them.
00:02:17.13 If we look all the way down here
00:02:19.15 where they're sitting on top of each other,
00:02:21.14 then it would be very hard for anyone
00:02:23.23 to say that they're two points,
00:02:24.29 although this is probably
00:02:27.09 much brighter than each of those points,
00:02:28.18 and one might even use
00:02:30.08 other kinds of information
00:02:31.27 to infer that there may be more than one point of light
00:02:34.24 on top of each other.
00:02:36.20 But the Raleigh criterion,
00:02:38.05 which is somewhere in the middle of this range,
00:02:41.11 it's this one up here...
00:02:43.14 there are other criteria known as the Sparrow
00:02:46.29 or the half-width, half-maximum,
00:02:48.11 which are not used much in microscopy,
00:02:49.24 so we'll just focus on the Raleigh criterion,
00:02:53.03 because that's the one that turns out to be valuable.
00:02:54.27 So, there's not generally accepted criterion,
00:02:57.01 but most microscopists use the Raleigh criterion
00:03:00.09 for their resolution limit,
00:03:02.11 and partly because this distance
00:03:04.08 between these two points
00:03:06.01 is 0.61λ/NA.
00:03:09.12 That is why that criterion
00:03:11.03 is called the Raleigh criterion.
00:03:12.16 And so what it means,
00:03:14.08 if you remember from previous discussions,
00:03:16.18 the distance 0.61λ/NA
00:03:18.21 is the distance from a point
00:03:20.26 to the middle of the first dark ring,
00:03:23.15 so if the second spot's
00:03:27.02 peak intensity is sitting right on top
00:03:29.28 of the dark ring of the first point,
00:03:31.17 then they are at the Raleigh criterion distance apart,
00:03:34.22 which is 0.61λ/NA,
00:03:36.14 and that is typically used as the resolution limit.
00:03:38.27 Even though these points that are closer together,
00:03:40.15 you might still say are separate points,
00:03:44.09 the Raleigh criterion is 0.61λ/NA.
00:03:47.10 If we look at the intensity profiles of these points
00:03:50.10 with a line going through them,
00:03:52.28 we can see that if we put them
00:03:55.25 at 0.61λ/NA,
00:03:57.17 the peak of each one
00:03:59.21 is sitting at the first dark ring,
00:04:01.29 the first dark area,
00:04:03.25 of the other,
00:04:05.22 and where their intensities add,
00:04:07.10 right between,
00:04:09.02 you drop the intensity by about 26%
00:04:11.13 from the intensity at the first one.
00:04:14.18 So, that's the lateral resolution Raleigh criterion.
00:04:17.19 You can also have an xz or yz, in depth,
00:04:22.08 Raleigh criterion,
00:04:23.24 which would now be 2nλ/NA^2,
00:04:25.23 based on something I told you in a previous discussion,
00:04:28.20 and that is also a drop of around 20%, 19%,
00:04:32.18 between the two.
00:04:34.02 So those would be the lateral, over here,
00:04:35.28 and the depth, over here,
00:04:37.26 Raleigh criterion.
00:04:40.16 The Raleigh criterion gives you a hint
00:04:43.07 that if objects are closer than that
00:04:46.15 they will be hard to resolve,
00:04:47.25 and based on what we've discussed previously
00:04:50.09 about the point spread function,
00:04:52.02 you know as the numerical aperture goes up,
00:04:54.09 0.61λ/NA goes down,
00:04:57.00 because NA is is in the denominator,
00:04:59.10 so it should be possible
00:05:01.28 to resolve objects that would,
00:05:03.25 at the same distance apart,
00:05:05.12 be blurred with a low numerical aperture lens,
00:05:07.17 would be sharply in focus
00:05:10.00 with a high numerical aperture lens.
00:05:11.23 And this following picture I think
00:05:14.20 makes that point pretty clearly.
00:05:16.13 So, here's a bunch of beads imaged,
00:05:18.10 at the same magnification, mind you,
00:05:20.05 but just with different numerical apertures,
00:05:22.10 and you can see with a low numerical aperture lens,
00:05:24.23 you just have this big blob,
00:05:26.20 you probably infer that there's more than one thing in there,
00:05:28.26 but you have a hard time counting;
00:05:31.12 with a medium NA,
00:05:33.07 you begin to get the sense that there are individual objects,
00:05:34.28 but the exact number would be hard to see;
00:05:37.24 but with a high numerical aperture lens,
00:05:39.25 now you see this remarkable set of beads,
00:05:42.23 they have lots of rings around them,
00:05:44.06 but you can discriminate
00:05:47.00 each and every one of them.
00:05:48.11 So, the higher the NA,
00:05:49.28 the smaller the Raleigh criterion
00:05:51.25 and the better the resolution.
00:05:54.12 And important point about this is, however,
00:05:58.22 that if the beads are smaller
00:06:00.28 than the resolution limit of light,
00:06:03.10 the point spread function says that
00:06:05.07 they're basically going to look the same,
00:06:07.00 even though they get smaller and smaller,
00:06:08.16 the resolution limit
00:06:10.14 limits how well you can resolve a small point.
00:06:12.13 So, if we look at a bead
00:06:14.06 that's below the diffraction limit
00:06:16.18 or the point spread function limit
00:06:19.00 of a light microscope,
00:06:20.09 which is...
00:06:21.19 you know, with even the best objectives,
00:06:23.00 it's not gonna be much better than around 200 nm,
00:06:25.11 so 138 nm bead will be
00:06:28.13 still a little fuzzier than 138 nm,
00:06:32.05 even with a good oil objective
00:06:35.09 like a 1.4 NA objective with 480 nm light.
00:06:38.12 As we look at beads that are smaller,
00:06:41.06 like 114 nm,
00:06:43.22 and do the computation of the point spread function,
00:06:47.10 there is a slight shrinkage of the size of the point spread function,
00:06:49.11 but it's pretty subtle.
00:06:51.14 And as we go to, now, 90 nm,
00:06:55.00 there's still a little bit of shrinkage,
00:06:57.20 but it's not much,
00:06:59.13 and then you see as we go to 66 nm, 42 nm,
00:07:03.06 and finally 18 nm,
00:07:05.12 those images are all rather similar.
00:07:08.22 So there is no ability,
00:07:11.25 because of the diffraction limit,
00:07:14.10 to get any better resolution than 0.61λ/NA.
00:07:18.03 So, image size does not change at all,
00:07:21.14 computationally,
00:07:23.27 once one is below about 60 nm.
00:07:26.29 These objects will get dimmer because there's less light coming out,
00:07:29.25 so you might be able to infer from a diffraction-limited point
00:07:31.27 that it's a small object if it's very dim
00:07:34.05 and must be a slightly bigger object
00:07:35.27 if it's brighter,
00:07:37.19 but in fact you can't infer it
00:07:39.18 from the dimensions.
00:07:41.20 So, this is a fundamental limit of light microscopy
00:07:44.25 based on diffraction.
00:07:47.22 As you probably know,
00:07:49.07 there are tools that kind of break through this limitation.
00:07:51.20 They're not something I'm going to talk about today,
00:07:54.02 but they are worth keeping in mind
00:07:55.25 if one wants to get more resolution.
00:07:57.15 What I would like to talk about
00:07:59.14 is that if you have a resolution limit
00:08:01.06 and you want to get
00:08:03.24 the most out of that objective,
00:08:05.10 you have to not only pay attention
00:08:06.29 to the quality of the objective
00:08:08.13 but the image itself has to have a grain
00:08:11.03 that's capable of seeing the finest details,
00:08:14.13 and that often is
00:08:16.13 an obvious missed point
00:08:18.29 and opportunity in a lot of imaging,
00:08:20.14 where people have very expensive objectives,
00:08:22.26 but they don't take images
00:08:24.27 at sufficient resolution in the imaging plane
00:08:27.25 to see all the details
00:08:30.02 that actually exist there.
00:08:31.06 So I want to talk a little bit about
00:08:33.18 sampling an image
00:08:35.01 and why one has to think about this
00:08:36.18 a little more deeply
00:08:38.24 than perhaps you might immediately think.
00:08:40.06 You might say, well,
00:08:41.26 if the resolution limit is 0.61λ/NA,
00:08:43.24 I should just have a resolution in my image plane
00:08:46.24 of 0.61λ/NA,
00:08:48.23 that should be the distance between the pixels
00:08:51.22 in the magnified image,
00:08:52.23 and you'll see that that is way insufficient.
00:08:55.15 So, let's start with an image, and here it is,
00:08:59.21 of something,
00:09:01.12 and I would say most of you would probably say,
00:09:01.25 I don't know what that is.
00:09:03.12 It's something,
00:09:05.22 but it's hard to tell what it is.
00:09:07.15 And you would say the reason you can't tell
00:09:09.06 is that there's not enough resolution in the image
00:09:11.20 to tell.
00:09:13.06 In fact, it is a burry image --
00:09:14.23 it's not been blurred,
00:09:16.08 it's just that it's been sampled with pixels
00:09:18.08 that are way too big to see the details.
00:09:20.02 So, let's move up by sampling
00:09:22.22 with a higher grain in our imaging plane.
00:09:27.00 Same camera, same image,
00:09:28.20 but now we're using better paper
00:09:30.13 that has a finer resolution.
00:09:32.18 And now you probably would say,
00:09:34.04 ah, that's probably a zebra.
00:09:36.03 But it doesn't look like a pretty good zebra.
00:09:38.03 I mean, you know where the head is,
00:09:39.16 but the details seem to be a little funny.
00:09:42.19 And if we move up a little further,
00:09:44.25 you now say,
00:09:47.00 oh yes this is a zebra,
00:09:48.16 but it has some unusual features,
00:09:50.02 and one of the features I like to point out
00:09:52.01 is this checkerboard of stripes, here,
00:09:56.09 running in two different directions,
00:09:58.29 and you may know enough about zebras to know that
00:10:01.12 zebras don't have checkerboards,
00:10:02.28 they really are striped,
00:10:04.11 and yet this image has this
00:10:06.09 artifact of stripes
00:10:09.23 running in two different directions.
00:10:11.18 This is known as aliasing
00:10:13.17 and it is a very obvious
00:10:16.07 and problematic aspect of images
00:10:18.00 because, especially as a researcher,
00:10:19.25 you're looking at things that may not have been looked at before,
00:10:22.11 so if you see objects with some periodicity,
00:10:24.27 you may say, oh,
00:10:26.22 that really is there,
00:10:27.27 because you don't know it's a zebra,
00:10:29.09 you don't know what it is.
00:10:30.14 But it is an artifact of sampling.
00:10:33.08 So, let's sample this same image
00:10:35.16 at higher resolution
00:10:37.23 as you see, as I go back and forth, here,
00:10:41.24 that those checkerboards
00:10:45.00 have now turned into vertical stripes...
00:10:46.28 well, not vertical, but diagonal stripes.
00:10:49.29 You may also notice on the neck...
00:10:52.15 sorry, on the nose,
00:10:54.21 I think this is the nose of a zebra,
00:10:56.12 I don't know what you call that part of a zebra,
00:10:58.17 but headed towards the nostrils,
00:11:00.20 there are a couple of stripes
00:11:02.13 that are running diagonally.
00:11:05.01 Is that real?
00:11:07.00 Do zebras have stripes like that?
00:11:09.02 Well, let's resolve this image
00:11:11.12 by using even better paper
00:11:13.17 or a better camera with higher...
00:11:15.22 not a better lens, just a better resolution,
00:11:19.04 more megapixels, let's say,
00:11:20.23 if it's a CCD or a solid state camera,
00:11:24.01 and you see that those stripes have also disappeared, now.
00:11:27.26 So, now, do we have the real zebra yet?
00:11:29.28 Well, it's hard to know.
00:11:31.22 You may noticed on the other side of the nose
00:11:34.18 there are still some horizontal...
00:11:36.22 or, sorry, some diagonal lines
00:11:39.17 that seem not symmetric with the other side.
00:11:41.20 Well, let's image at even higher resolution
00:11:43.29 and you see those have gone away.
00:11:46.08 So, this looks like a pretty good image of a zebra.
00:11:48.23 There is still...
00:11:50.04 I'm not sure you're going to be able
00:11:52.05 to see on your monitors...
00:11:53.11 there are still some bands
00:11:54.26 that are running horizontal across the front of the nose
00:11:58.06 at the finest resolution.
00:11:59.19 If we go one higher,
00:12:01.09 and for most of you
00:12:02.29 you probably won't even detect the difference,
00:12:05.16 now finally even the vertical stripes there
00:12:09.15 are obvious, whereas in this previous image
00:12:11.18 they are slightly diagonal.
00:12:13.21 So, aliasing,
00:12:16.11 this kind of problem of lower resolution,
00:12:19.29 is not just that things are fuzzy,
00:12:22.02 but it generates
00:12:24.01 areas of light and areas of dark
00:12:25.25 that are artifacts,
00:12:27.13 and this can severely
00:12:30.02 crimp your ability to interpret an image,
00:12:32.18 so that's why I'm spending so much time with this.
00:12:35.10 So, how do you know how much resolution you need?
00:12:38.17 The person who has helped us best with this
00:12:44.11 is a guy named Harry Nyquist
00:12:46.04 and, you know, some of you may have heard of the Nyquist limit.
00:12:48.13 It's a law that was named after him in 1933.
00:12:51.20 He was a scientist interested in sound
00:12:53.13 and he stated that
00:12:55.21 sound must be sampled
00:12:57.10 at at least twice the highest frequency
00:12:59.25 in order to extract all the information
00:13:02.15 from the bandwidth
00:13:04.15 and accurately represent the original acoustic energy.
00:13:07.20 So, his idea was, for example,
00:13:10.07 if the human ear can hear frequencies
00:13:12.19 up to 20,000 Hz, 20 kHz,
00:13:14.23 then if you want to hear those frequencies without artifacts,
00:13:19.07 you're going to have to sample that sound
00:13:21.18 at at least twice that frequency,
00:13:24.01 and so a modern CD
00:13:26.15 that is a high fidelity recording is at 44 kHz.
00:13:32.10 Not because you can hear 44 kHz,
00:13:33.23 but in order to hear 20 kHz accurately,
00:13:35.12 you have to be sampling
00:13:37.14 twice the highest frequency.
00:13:39.06 Phone lines pass frequencies up to 4 kHz,
00:13:42.13 so the phone company samples sounds
00:13:45.22 at 8 kHz.
00:13:47.05 So, this is true in sound
00:13:48.20 but it's also true
00:13:51.19 in all forms of signals that have oscillations,
00:13:55.24 that you need to sample
00:13:58.05 twice the highest frequency
00:14:00.07 you're interested in.
00:14:01.10 So, the sampling theory, based on Nyquist,
00:14:03.10 is a continuous function
00:14:05.25 that can be completely represented
00:14:07.14 by a set of equally space samples
00:14:09.04 if the samples occur
00:14:11.10 at more than twice the frequency
00:14:13.18 of the highest frequency component of the function.
00:14:15.22 Now, with our imaging,
00:14:17.06 the highest frequency component
00:14:19.10 is 0.61λ/NA,
00:14:21.08 so if you want to see 0.61λ/NA details,
00:14:24.00 you're going to have to sample that kind of oscillation
00:14:28.20 with at least two points between each point.
00:14:32.03 So, to capture a function with maximum frequency,
00:14:35.03 the sample... it has to be sampled at twice that frequency
00:14:39.15 -- this is called the Nyquist limit.
00:14:40.19 And the reason aliasing occurs
00:14:42.15 is shown in this bottom diagram.
00:14:44.01 So, if you have a sinusoid,
00:14:45.16 let's say zebra stripes
00:14:47.15 that are bright and dark, bright and dark,
00:14:48.24 that's that red line,
00:14:50.24 so it's an oscillation between bright and dark, bright and dark,
00:14:53.26 but you're only sampling
00:14:56.09 once per wavelength, here,
00:14:59.03 so you only get one sample
00:15:01.05 for every peak and trough,
00:15:02.22 you're going to end up with a sinusoid, in this case,
00:15:06.02 that's got half its...
00:15:07.21 it goes from light to dark,
00:15:09.21 but it has half the actual frequency of the original data.
00:15:11.25 And the aliasing generates these kind of nice patterns
00:15:15.03 that are just total artifacts of subsampling the data.
00:15:20.03 So, in the case of the resolution limit of light,
00:15:23.03 which is 0.61λ/NA,
00:15:25.13 there are those two spots,
00:15:27.00 they're separated by 0.61λ/NA.
00:15:29.19 If you want to resolve them,
00:15:31.25 you're going to have to have two pixels between them.
00:15:34.28 So, the Nyquist limit is 0.3λ/NA.
00:15:38.14 That is, your imaging plane
00:15:40.15 is going to have to be able to see
00:15:42.24 something that's half as big
00:15:45.15 as the limit of resolution.
00:15:48.12 For example,
00:15:51.01 if you're imaging and putting your image
00:15:52.21 onto a charge-coupled device (CCD),
00:15:54.16 in the spec sheet for the charge-coupled device
00:15:57.04 is a datum which tells you
00:15:59.13 the size of the pixels on the charge-coupled device.
00:16:02.13 So, a charge-coupled device
00:16:04.28 might have pixels that are 12x12 microns
00:16:07.19 -- square pixels --
00:16:10.20 and you can figure out what the size of your image
00:16:14.08 will be in your microscope.
00:16:15.22 So, for example, if you're using
00:16:17.16 a 1.4 numerical aperture lens at 480 nm,
00:16:20.26 the resolution limit, 0.61λ/NA,
00:16:23.27 is 0.22 microns.
00:16:26.08 The 60X magnification
00:16:28.07 is going to blow that 0.22 microns up
00:16:30.26 into a bigger object,
00:16:32.16 that is, it's going to go 0.22 x 60,
00:16:35.07 so now your diffraction-limited size
00:16:38.28 is 13.2 microns,
00:16:40.26 which is the distance of the resolution limit
00:16:45.01 on the face plate.
00:16:47.26 Your pixels on your face plate are 12 microns apart,
00:16:50.08 so this is matching very well the resolution limit,
00:16:53.22 but that's not enough to see the finest details
00:16:56.16 -- as I said, you have to oversample by a factor of two.
00:16:59.07 So, what do you do?
00:17:00.22 Well, you're using a 60X lens that's 1.4 numerical aperture.
00:17:03.01 You've gotta get more magnification.
00:17:05.12 One of the reasons companies sell
00:17:07.24 100X 1.4 NA lenses
00:17:10.09 is not because they have better resolving power
00:17:12.18 than 60X,
00:17:13.29 but they add that empty, extra magnification
00:17:16.14 that then allows you to better match
00:17:19.08 the size of the image
00:17:21.03 with the resolution of the CCD.
00:17:23.12 Another thing is some people have mag changers,
00:17:25.13 they're called optivars,
00:17:27.09 right in the microscope, that blow up the image,
00:17:29.24 just add magnification,
00:17:31.23 and so those are the kinds of things you can do.
00:17:34.11 So, here, dramatically,
00:17:38.00 a mag changer to zoom by 2X
00:17:40.29 or use 100X objective.
00:17:42.20 Then the resolution limit is now 26.4 microns on the CCD,
00:17:47.14 if you do a mag change of 2,
00:17:49.26 which is greater than twice the sampling resolution of 12 microns,
00:17:54.25 so now you will get the full value
00:17:56.15 of that high numerical aperture lens.
00:17:58.11 But if you need more resolution,
00:18:00.21 what are the things you can do.
00:18:02.06 So, I'll just review some that we've already talked about:
00:18:04.07 one is that you can go
00:18:06.04 from red light to blue light
00:18:08.21 or ultraviolet light
00:18:10.13 to get higher resolution,
00:18:12.08 because 0.61λ/NA,
00:18:14.24 as the λ gets smaller,
00:18:17.10 the wavelength gets smaller,
00:18:18.28 the resolution, the point spread function gets tighter,
00:18:20.29 and you get higher resolution.
00:18:22.20 A second thing is,
00:18:24.16 if you're doing your work in air, you can go to water,
00:18:27.11 if you're working in water, you can go to glycerin,
00:18:28.25 if you're doing it in glycerin, you can go to oil...
00:18:30.24 each of those increases the numerical aperture
00:18:32.20 because the numerical aperture is n,
00:18:36.15 the index of refraction,
00:18:38.06 times the half-angle of acceptance,
00:18:40.16 so higher index immersion fluid
00:18:44.16 will give you better resolution.
00:18:46.18 Of course, you've got to use an objective
00:18:48.06 that takes those different immersion fluids.
00:18:52.03 As I've already mentioned, the confocal...
00:18:53.25 if you stop that pinhole down
00:18:55.26 to the size of an Airy disk or smaller,
00:18:57.04 you get an increase
00:18:59.15 of about the square root of 2 -- 1.4 --
00:19:01.23 improvement in resolution,
00:19:03.22 so that always can be used.
00:19:05.13 If you are studying
00:19:07.26 just single spots of light, single molecules,
00:19:12.09 in fact, then you can know their position in space
00:19:15.29 to much greater resolution
00:19:17.26 because the center of the spot
00:19:20.06 is a very accurate rendering of where that object is,
00:19:22.23 so people who track the movement of objects
00:19:24.25 can go well beyond the diffraction limit,
00:19:27.08 because the movement of the object from one place to another
00:19:29.23 can be tracked by looking at the center of that spot.
00:19:32.01 Indeed, some of the super-resolution techniques
00:19:35.08 take advantage of just that property
00:19:37.20 to give you super-resolution.
00:19:39.17 And some of you may know that there's
00:19:41.21 this whole set of other techniques available
00:19:45.13 called super-resolution.
00:19:46.18 Before mentioning those,
00:19:48.05 I'll also tell you that if you have two different colored objects,
00:19:49.29 they can become much closer together,
00:19:52.00 because they're not interfering with each other,
00:19:54.03 they're different colors,
00:19:55.20 so they're disambiguated,
00:19:57.08 so one can get the distance
00:19:59.16 from one object to another
00:20:00.29 that's much closer than the diffraction limit.
00:20:02.22 For example, in a synapse,
00:20:05.21 the presynaptic terminal can be labeled with one color
00:20:06.29 and the postsynaptic with another color,
00:20:08.20 and you can see distances apart
00:20:10.09 that are well below the diffraction limit
00:20:12.11 by virtue of different colors.
00:20:13.29 But that last thing is
00:20:16.27 super-resolution nanoscopy
00:20:18.16 and I'll think there will be other discussions of this
00:20:20.10 in other lectures.
00:20:22.13 Thank you.

This Talk

Speaker: Jeff Lichtman

Audience:

Researcher

Recorded: December 2012

Talk Overview

The resolution of a microscope can be defined as the smallest distance at which two small objects can still be seen as separate objects. This lecture discusses various criteria for resolution, the factors that influence resolution in the lateral and axial planes, and how to sample an image adequately using a camera or confocal microscope, such that the full optical resolution is retained.

Questions

You have a 100×1.4na objective, no additional magnification in your microscope and can choose between 4 different cameras, varying in pixel size. Which camera should you choose to fulfill the Nyquist sampling criterium, assuming that you resolution limit is described by the Rayleigh criterium and you work with light of 500nm?
1. 16 micron pixels
2. 10 micron pixels
3. 6.5 micron pixels
4. 4.0 micron pixels
You use a 40×1.4na objective on your confocal microscope and are imaging a red fluorescent protein emitting at ~575nm. You are using an image size of 512×512 pixels. To satisfy Nyquist sampling (using the Rayleigh criterium for resolution) you should scan an area no larger than:
1. 64×64 mm
2. 128×128 microns
3. 64×64 microns
4. 40×40 microns
Which change(s) in your imaging setup will increase the resolution of your sample for GFP fluorescence?
1. Higher wavelength
2. Higher refractive index of immersion medium
3. Lens with higher numerical aperture
4. Lens with lower numerical aperture
5. Lens with better chromatic correction

Answers

Speaker Bio

Jeff Lichtman

Jeff Lichtman

Jeff Lichtman’s interest in how specific neuronal connections are made and maintained began while he was a MD-PhD student at Washington University in Saint Louis. Lichtman remained at Washington University for nearly 30 years. In 2004, he moved to Harvard University where he is Professor of Molecular and Cellular Biology and a member of the… Continue Reading