Unraveling SARS-CoV-2 Host-Response Heterogeneity through Longitudinal Molecular Subtyping

Transcript

• 00:00It's my our colleague, Hui
• 00:02Kao from,
• 00:04applied physics and physics in
• 00:05in the school of, engineering.
• 00:07So, Hui, it's, the John
• 00:09Malone professor of applied physics,
• 00:12and, her research focuses on
• 00:14mesoscopic
• 00:14physics, complex nanophotonics,
• 00:17she got her PhD in
• 00:17applied physics from Stanford, and
• 00:18then,
• 00:20she was on the faculty
• 00:20for Northwestern before she joined
• 00:20Yale,
• 00:21lucky for us. And, she's
• 00:23been a a a pioneer
• 00:26and leader in
• 00:30on random, working on random
• 00:32lasers. And and recently, she's
• 00:34been, working on,
• 00:36trying to see do deep
• 00:38tissue imaging, and I think
• 00:39that's what she's gonna talk
• 00:40about. And doctor Kao is
• 00:41an elected member of the
• 00:42National Academy of Sciences and
• 00:44also the, American Academy of
• 00:46Arts and Sciences. So, Huey?
• 00:54Let me try to see
• 00:55if I can find it.
• 00:56Okay. So first, it's a
• 00:58great pleasure for me to
• 00:59give this talk. I'd like
• 01:00to thank John for this
• 01:01invitation.
• 01:03I'm really trained as a
• 01:04physicist, and I'm doing applied
• 01:05physics engineering.
• 01:07I know nothing about the
• 01:08immune system,
• 01:09but I would love to,
• 01:10you know, see whether I
• 01:11can, you know, collaborate with
• 01:12people here at your medical
• 01:14school.
• 01:15So my talk, I'm gonna
• 01:16start with the introduction about
• 01:18my my previous work as
• 01:20I'm trying to show what
• 01:21we're trying to do right
• 01:22now. Hopefully, you know, this
• 01:24will just to give you
• 01:25some idea what I'm doing,
• 01:26and I'm very much looking
• 01:27forward to, you know, talking
• 01:29with you and see what
• 01:30are the potential collaboration we
• 01:31could have.
• 01:33So,
• 01:34okay. So this is, actually,
• 01:36my title. But before that,
• 01:37as I said, I want
• 01:38to talk a little bit
• 01:39of what we have what
• 01:40I've been doing, especially kind
• 01:42of you can collaborate with
• 01:43people at your medical school.
• 01:45So I'm, doing optics. Right?
• 01:47So I think for optical
• 01:49imaging,
• 01:49what is really important, it
• 01:51is the light source for
• 01:52illumination.
• 01:53And especially for wide field
• 01:55imaging like a microscope,
• 01:57we all know that mostly
• 01:59people using, you know, a
• 02:00lamp or LUD. Right?
• 02:03But, those are has a
• 02:04limited intensity. If you want
• 02:06to,
• 02:08tissue or get into some
• 02:10material has some absorption,
• 02:12this may not be sufficient,
• 02:13you know, enough intensity.
• 02:15So of of course, you
• 02:16know, people to say, how
• 02:17about we go to lasers?
• 02:18Because laser is much, you
• 02:20know, brighter.
• 02:21Right? So that is, you
• 02:23know, really or superluminescent
• 02:25LED. You have a, you
• 02:26know, you know, much stronger
• 02:28signal or much stronger excitation.
• 02:31But I want to show
• 02:32there's another thing important in
• 02:33addition to brightness.
• 02:35That is the coherence.
• 02:37So we know that, you
• 02:38know, you know, laser, it
• 02:40is coherent, where there's a
• 02:41LED or lamp is incoherent.
• 02:43And this coherence actually here
• 02:45is really hurting us because
• 02:47that actually can introduce incoherent
• 02:49artifact, which I'm gonna show
• 02:50you in particular like a
• 02:52speckle noise. Maybe some of
• 02:53you have already seen that.
• 02:55So what we really ideal
• 02:57light source
• 02:58is to, you know, as
• 03:00bright as a laser, a
• 03:01standard laser,
• 03:02but it has a low
• 03:03coherence. So we don't have
• 03:05all this coherent artifact
• 03:06if we we try to
• 03:08use this as a illumination
• 03:09source.
• 03:10So we have been trying
• 03:11to study how we can
• 03:12find, you know, a light
• 03:14source that combine the advantage
• 03:15of a laser and also
• 03:17a lamp. Right? So that's
• 03:19actually we I got into
• 03:20this random laser,
• 03:22which can really, produce a,
• 03:23you know, a speckle free
• 03:25image.
• 03:26So here shows example. You
• 03:28see this kind of a
• 03:29a speckled image. That's when
• 03:30we try to image a
• 03:32a sample, and then there's
• 03:33some scattering, you know, you
• 03:35know, layer in between the
• 03:36sample and the objective lens
• 03:38and also to the camera.
• 03:40We are using a laser
• 03:42as illumination.
• 03:44You see that we see
• 03:45all the speckle pattern. We
• 03:46don't see anything else. That's
• 03:47really a strong coherent artifact
• 03:49coming from this, you know,
• 03:51interference
• 03:52of this laser.
• 03:55So we actually develop a
• 03:56random laser
• 03:57and that using that as
• 03:59illumination,
• 03:59you see that actually, we
• 04:01can really even we have
• 04:02a strong scattering we have
• 04:04significant scattering here, we can
• 04:05still see this kind of
• 04:07features.
• 04:07And this kind of random
• 04:09laser can be as bright
• 04:10as a conventional
• 04:11laser. So that is really,
• 04:13give us a more powerful
• 04:14source for speckle free imaging,
• 04:17especially for full field imaging.
• 04:20So that is what I
• 04:21did the first part, try
• 04:22to get rid of speckle.
• 04:23But it turns out speckle
• 04:25is not always bad. Because
• 04:26even though they cannot take
• 04:27a allow us to see
• 04:28the structure, they can allow
• 04:30allow us to see the
• 04:31motion.
• 04:32Because if you see, for
• 04:33example, if there's something was
• 04:35moving when they generate speckle
• 04:36pattern, the speckle pattern will
• 04:37change.
• 04:38If you integrate over a
• 04:39certain time, the speckle pattern
• 04:41will average out so you
• 04:43have a low contrast.
• 04:44So speckle contrast tell us
• 04:46the motion.
• 04:47So what would be ideal
• 04:49is that if we can
• 04:49switch the laser source, the
• 04:51coherence,
• 04:52I can first get into
• 04:53low spatial coherence. I can
• 04:55see the structure.
• 04:56Then I can switch to
• 04:57high spatial coherence. I can
• 04:59see the speckle and how
• 05:00they're moving, how they're changing.
• 05:01Then I can see the
• 05:02flow.
• 05:03So here shows one example,
• 05:05which I I should collaborate
• 05:06with professor, Michael Choma and
• 05:08Mustafa
• 05:10Koha here at the medical
• 05:11school. We try to imaging
• 05:12the, you know, this, you
• 05:14know, heartbeat
• 05:15of a a leading tadpole.
• 05:17So for this kind of
• 05:18tadpole here, so what we
• 05:20want, we want to see
• 05:21how this, you know, the
• 05:22this heartbeat, what can happen
• 05:24to the structure.
• 05:26We also want to see
• 05:27what is the flow
• 05:28because this is the animal
• 05:29mode of heart disease.
• 05:32So here, this basically show
• 05:33that we develop a laser.
• 05:35We can switch very quickly
• 05:36the spatial coherence back and
• 05:37forth.
• 05:38So if I can,
• 05:41oh, why I cannot,
• 05:43play the
• 05:44can somebody help me to
• 05:45play the video or
• 05:48I have to play this
• 05:49okay. Anyway, this video is
• 05:50not working.
• 05:53I get okay. No. I
• 05:54think okay. Let me try
• 05:55to see what I can
• 05:56just use in this thing
• 05:57here too.
• 05:58Right. Okay. Folder.
• 06:01Oh,
• 06:03okay. So
• 06:06okay. Anyway
• 06:08so
• 06:09the unit the the this
• 06:10thing is not playing.
• 06:13I have a unit here.
• 06:21Oh, somehow, I think this
• 06:22cannot play. I don't think
• 06:23it's gone. Okay. Anyway, so
• 06:25that's fine. I think it's
• 06:27the same.
• 06:28No problem. Okay.
• 06:30Yeah.
• 06:32Yeah. I see maybe when
• 06:33we,
• 06:34translate this file, there's some
• 06:36issue there. So, anyway, you're
• 06:37supposed to see this heart
• 06:38is beating. As I can
• 06:40see, there's a flow of
• 06:41the blood from one chamber
• 06:43to another chamber simultaneously.
• 06:45So this allows to see
• 06:46simultaneously the structure,
• 06:48change and also there's a
• 06:49shape change there. So this
• 06:51is basically,
• 06:52allow, you know, them to
• 06:53really study what is this,
• 06:55you know, if there's some
• 06:56disease there, what happens to
• 06:57the deformation and what, I
• 06:59mean, what happened to the
• 07:00flow of blood. Because the
• 07:02contrast shown as the yellow
• 07:04here tells you how much
• 07:04blood in each location.
• 07:07You can say, well, maybe
• 07:08I can also do this
• 07:09with ultrasound. Why you want
• 07:10to do this with, you
• 07:11know, light? Because light can
• 07:13give us much better spatial
• 07:14resolution.
• 07:15So, you know, for some
• 07:16application, this could be useful.
• 07:20Alright. So that is the
• 07:21first part we developed given
• 07:23a light source. We tune
• 07:24the coherence with different applications
• 07:26because the reason I have
• 07:27to say I really, enjoy
• 07:28collaboration with Michael,
• 07:30the most offer because they
• 07:31really tell me what they
• 07:33need, what kind of light
• 07:34source they need, and then
• 07:35this will help me to
• 07:36really develop this kind of
• 07:37light source for the application.
• 07:39So this is very important
• 07:40for engineer because we have
• 07:41tools, but we don't know
• 07:42what this can be useful
• 07:43for. Right?
• 07:45So that is the first
• 07:46part. And now the second
• 07:47part is that, we also
• 07:49try to do some computational
• 07:51imaging.
• 07:52Meaning that, you know, we
• 07:53try to do structure illumination.
• 07:55We try to illuminate, you
• 07:56know, the sample with some
• 07:57particular patterns so that, for
• 07:59example, can give us some
• 08:01better resolution or give us
• 08:02some, you know, additional feature
• 08:04which we can now do
• 08:04for homogeneous illumination.
• 08:07So for this, we actually
• 08:08are using speckle pattern because,
• 08:10you know, compared to periodic
• 08:12structure people use for this
• 08:13kind of structure illumination,
• 08:15actually, speckle pattern has some
• 08:17additional advantage.
• 08:18For example,
• 08:20vertical,
• 08:20axial sectioning.
• 08:22I will not get into
• 08:23those detail, but what we
• 08:25did is that we actually
• 08:26can design the speckle pattern,
• 08:28not just using, like, a
• 08:30a a common speckle pattern.
• 08:31We can design the speckle
• 08:33pattern for particular, you know,
• 08:35application,
• 08:36and then we can really,
• 08:37use that to enhance the
• 08:38imaging performance.
• 08:40So this, again, we actually
• 08:42collaborate with, professor Jorga Viverstov
• 08:44here at your medical school
• 08:46to using this, design spec
• 08:48pattern
• 08:49for parallelized nonlinear pattern illumination
• 08:52microscopy.
• 08:53This is based on fluorescence
• 08:55photo switching.
• 08:57So just show you a
• 08:58simple example here. So what
• 09:00we did is, for example,
• 09:01outside of this region, this
• 09:03is just a I mean,
• 09:04illustration with a standard speckle
• 09:06pattern people use. And in
• 09:08the central box here, it
• 09:09is a particular speckle pattern
• 09:11we designed.
• 09:12And then, if we're using
• 09:14this to illuminate this, you
• 09:16know, protein,
• 09:17and then for the bright
• 09:18regime, the protein will switch.
• 09:20And in the dark regime,
• 09:22they will not switch, so
• 09:22they can still fluorescent when
• 09:24we excite them.
• 09:25So now you can see
• 09:26the fluorescence here. Now you
• 09:28can see a very different
• 09:29things here. You know, outside
• 09:31using standard
• 09:32speckle, illumination,
• 09:34you can see this kind
• 09:35of, you know, like, this
• 09:37kind of,
• 09:38elongated or this kind of
• 09:40a network of fluorescence, you
• 09:42know, from the sample, which
• 09:44is hard to tell what's
• 09:45going on. But inside here,
• 09:47you see the fluorescence coming
• 09:48from isolated
• 09:50individual, you know, bright, you
• 09:52know, I mean, region there.
• 09:54So this actually coming from
• 09:55those dark region which we
• 09:56design in this kind of
• 09:57speckle pattern, and this actually
• 09:59can give us a better
• 10:00resolution.
• 10:01So we actually show that,
• 10:03we can break the optical
• 10:04diffraction limit three times, which
• 10:06allow us get a three
• 10:07times better
• 10:09spatial
• 10:10resolution.
• 10:11So this is basically to
• 10:12you know, we are actually
• 10:13continue this design in the
• 10:15speckle pattern for particular applications
• 10:17so that, you know, to
• 10:18optimize so that we can
• 10:20combine with the, you know,
• 10:22reconstruction
• 10:23algorithm, which people have developed
• 10:25so that we can simultaneously
• 10:27optimize hardware, which is illumination,
• 10:29and the software, which is
• 10:31this algorithm to reconstruct the
• 10:32information.
• 10:34So, again, we would like
• 10:35to see what potential application
• 10:36this could be,
• 10:38for medical application.
• 10:40So
• 10:40the next one, which we
• 10:43have been studying is, actually,
• 10:45this is a a deep
• 10:46tissue imaging.
• 10:47Like, John just mentioned, we
• 10:49actually recently got, also,
• 10:52a phase two grant from
• 10:53this Charles Zuckerberg initiative
• 10:55on deep tissue imaging.
• 10:57And, so our goal is
• 10:59try to utilize the correlation
• 11:01engineering
• 11:02to achieve deep multi photon
• 11:04microscopy.
• 11:05And,
• 11:06our, again, our, you know,
• 11:09you know, approach
• 11:10is, again, on the light
• 11:12source because light source is
• 11:13really the engine for optical
• 11:14microscopy.
• 11:15If you see the last,
• 11:16I don't know, twenty, thirty
• 11:17years, all there's a breakthrough
• 11:19in optical, you know, imaging
• 11:21is coming from
• 11:22the drum or the light
• 11:23source, ultra short, you know,
• 11:25pulses and other things. So
• 11:26we want to build a
• 11:27next generation of the laser
• 11:29source for better, deeper, and
• 11:31a gentle microscopy.
• 11:33So for this one here,
• 11:34we actually have a team.
• 11:36We actually have a a
• 11:38collaborations
• 11:39from professor Tian Yu Wang
• 11:40from Boston University and and
• 11:42also professor from Cornell University
• 11:45who really is a, you
• 11:46know, world leader on, you
• 11:48know, not two photon, three
• 11:49photon microscopy.
• 11:51And, also, we are fortunate
• 11:52to have,
• 11:53Logan Wright, who's a young
• 11:54assistant professor, join Yale applied
• 11:56physics department,
• 11:58working together to develop this
• 12:00kind of light source. So
• 12:01we are engineers and those
• 12:03you know? So,
• 12:04what particularly we want to
• 12:06do, I will not get
• 12:07into much detail,
• 12:08is that what we want
• 12:10to is really, to precisely
• 12:12tailor the coherent multimodal laser
• 12:14so that we can maximize
• 12:16the multi photon absorption efficiency,
• 12:19in this complex fluorescent molecules.
• 12:21So, basically, we want to
• 12:23say what kind of fluorescent
• 12:24molecules we want to target,
• 12:26and so we want to
• 12:27see how we can maximize
• 12:28this multi photon absorption
• 12:30by tailoring this illumination source.
• 12:33Okay? So in some way,
• 12:34we want to co design
• 12:35the light source also with
• 12:36the molecule so that we
• 12:38can enhance this kind of
• 12:39three photon,
• 12:40microscopy
• 12:41that is mostly for the
• 12:42tissue imaging, like go to
• 12:44the brain imaging.
• 12:46So,
• 12:47this is, like, a overview
• 12:49what we have been what
• 12:50I've been doing so far.
• 12:51And also, I will now
• 12:52I want to go to
• 12:53this, you know, wavefront shaping.
• 12:55So this is another technique
• 12:57I wish to introduce to
• 12:58you, to see whether that
• 12:59could be applied for immune
• 13:01system.
• 13:02So
• 13:03we know that, you usually,
• 13:05you know, why we cannot
• 13:06see deep into a tissue?
• 13:07Because of the light scattering,
• 13:08not because the absorption.
• 13:10So you can see this
• 13:11actually, you know, if you,
• 13:13you know, in a foggy
• 13:14day. Right? If you see,
• 13:15you know, what's going on,
• 13:17you know, actually, if you
• 13:18see something, you know, close
• 13:20by, that is easy to
• 13:21see. That's coming from, you
• 13:22know, single scattering with single
• 13:24reflection.
• 13:25Now if you go a
• 13:26little bit, you know, deeper
• 13:28in there, you still can
• 13:30sort of see some, you
• 13:31know, some something there, but
• 13:33it's not so easy.
• 13:34And this kind of actually
• 13:36signal can be reconstructed,
• 13:38or, you know, extracted from
• 13:40optical coherence tomography,
• 13:41from multifocal, you know, microscopy,
• 13:44from confocal, you know, microscopy.
• 13:47So those you you you
• 13:48can do that, you know,
• 13:49if you do this in
• 13:50a tissue.
• 13:51But if you go even
• 13:52deeper in,
• 13:54that become a mission impossible
• 13:56because you just cannot see
• 13:57anything. Right? So that is
• 13:59really what we want. We
• 14:00wanna go really deeper in,
• 14:01to see what's going on
• 14:03there. So, we want to
• 14:04push this frontier.
• 14:06So that's why I think
• 14:07from the imaging to biological
• 14:09tissue, there's two regimes. Of
• 14:11course, if there's nothing there,
• 14:12that's perfect. You just have
• 14:13a lens you can image
• 14:14it perfectly. Right? For example,
• 14:16if you can make your
• 14:17brain become transparent, then you
• 14:18can just do that. But
• 14:20if you, have some, you
• 14:21know, like a slight aberrations
• 14:24or, like, the scatterings, you
• 14:25see this kind of distorted
• 14:27image, but you still can
• 14:28guess something out of that.
• 14:30But what we are really
• 14:31interested in, you go really
• 14:32deeper in and then you
• 14:33see all this kind of
• 14:34scattering, give you a scattering,
• 14:36like a speckle pattern. We
• 14:38just cannot see anything anymore.
• 14:39Right? But we still want
• 14:40to see something. So how
• 14:41can we get you know,
• 14:42how can we do that?
• 14:44So, you know, you're not
• 14:44to really see something deep
• 14:46inside the way to first
• 14:47get the light deeper in.
• 14:49Right? Then we can probe
• 14:50something.
• 14:50So the first,
• 14:52task is whether we can
• 14:54focus in light through this
• 14:55strongly scattering, you know, medium.
• 14:59So because I said, if
• 15:00you have light going through
• 15:01a scattering medium, it's become
• 15:02a really a speckle. Right?
• 15:03You don't see anything. But
• 15:05I want still focusing light
• 15:07to a spot after this
• 15:09strongly scattered medium. You say,
• 15:11how can I do that?
• 15:12Actually, this is basically we
• 15:13can use in this so
• 15:14called waveform shaping,
• 15:16or we can precompensate
• 15:18the effect of scattering.
• 15:20So in other words, if
• 15:21I think about I have
• 15:22a laser beam, I go
• 15:23to a specialized modulator.
• 15:25If I can modulate this
• 15:27wavefront
• 15:28in a smart way, and
• 15:29then when the light is
• 15:31scattered from different, you know,
• 15:32paths into this particular
• 15:35position, they can interfere constructively,
• 15:37and some they can greatly
• 15:38enhance the intensity at this
• 15:40position with which.
• 15:41So this, you can really
• 15:43get a light deeper in,
• 15:44and people have already shown
• 15:45that.
• 15:46Actually, this enhancement originally,
• 15:49become getting to a thousand
• 15:50times higher. Now, actually, people
• 15:52can go to one hundred
• 15:53thousand times stronger
• 15:55intensity
• 15:56enhancement. So that you can
• 15:57really enhance the intensity at
• 15:58some location.
• 16:00So now if you have
• 16:01a way to do that,
• 16:02to develop to get the
• 16:04light in there to excite
• 16:05some molecules,
• 16:06to do the imaging, we
• 16:07need to scan these positions.
• 16:09If we can scan these
• 16:10positions, now we can get,
• 16:12we can get the image.
• 16:13And this one, actually, people
• 16:14have tried to work using
• 16:16this for the memory effect.
• 16:17I don't have time to
• 16:18get into that. But I
• 16:19want to show one example
• 16:21which people try to, not
• 16:23my group,
• 16:24but actually, people try to
• 16:26show that you can, do
• 16:27this optogenetic control of cell
• 16:30signaling pathway so there's a
• 16:32highly scarring, skeleton mouse skull.
• 16:35So, basically, if you have
• 16:36a, you know, skull, it
• 16:37is highly scattering. If you
• 16:39just send a light through
• 16:40it, using objective lens, you
• 16:43just don't get anything. You're
• 16:44diffused away.
• 16:46Of course, you can say
• 16:46we can't use an optical
• 16:47fiber, which is still really
• 16:49important, you know, technology,
• 16:51but the question is this
• 16:52will be somehow invasive.
• 16:55So how we can do
• 16:56it noninvasively,
• 16:57and that's that's what people
• 16:59have shown that by shaping
• 17:00this wavefront,
• 17:02we can really, focus, you
• 17:03know, deep inside,
• 17:05you know, through this scattering
• 17:07skull. And then by scanning
• 17:09this spot, you know, you
• 17:10can get the image there.
• 17:13But then you say, how
• 17:14can I know where I
• 17:15focus the inside? I I
• 17:16don't even, you know, drill
• 17:17a hole. How can I
• 17:18see what is inside?
• 17:20To do that, we need
• 17:21a guide star.
• 17:22A guide star, you know,
• 17:23can be fluorescence markers,
• 17:26can be nonlinear optical particles,
• 17:28can be, you know, some
• 17:29some people using photo acoustic
• 17:31feedback
• 17:32or ultrasound or kinetic guide
• 17:34stars. So there's many different
• 17:35type of guide stars people
• 17:36have been developed
• 17:38over the last, decade,
• 17:40try to really help to
• 17:42focus in light deep into
• 17:44the system there.
• 17:46So there has been a
• 17:47lot of work in there,
• 17:48so I don't have time
• 17:49to get into that. But
• 17:50I think this is basically
• 17:51a wavefront shaping maybe allow
• 17:53us to see something deeper
• 17:54into that.
• 17:56Then you say, well, fine.
• 17:57You probably can go somewhere
• 17:59deeper in, but how but,
• 18:00you still need to put
• 18:01something into your system like
• 18:02a guide star to help
• 18:03you. Right? But how about
• 18:05complete noninvasive?
• 18:07If we really have complete
• 18:08noninvasive,
• 18:09there's another technology called a
• 18:11called a diffusive optical tomography.
• 18:14So this is basically,
• 18:16has been widely
• 18:17used in brain imaging.
• 18:19Basically, you know, you can
• 18:21see array of, you know,
• 18:23laser diodes or LEDs,
• 18:26placed on, you know, together
• 18:27with the, you know, these
• 18:29photo detectors, they are interleaved
• 18:31and and placed on the
• 18:33skull.
• 18:34So, I mean, what happened
• 18:35to that when the you
• 18:37know, from a light source,
• 18:38you can the light can
• 18:39be injected. This usually is
• 18:40a near infrared light, which
• 18:42has less absorption and less
• 18:44scattering so it can go
• 18:45through the skull.
• 18:46And then
• 18:47get into this brain, and
• 18:50after all this kind of
• 18:51scattering inside, a small component
• 18:53can be really remitted
• 18:55at some distance away from
• 18:57the original injection location.
• 19:00And,
• 19:01usually,
• 19:02those kind of remitted light
• 19:03will carry information deep inside
• 19:06the brain.
• 19:07You can see that if
• 19:08you further increase the distance
• 19:09between source and the detector,
• 19:11the the the light had
• 19:12to go deeper even deeper
• 19:14in would take information
• 19:16even from a deeper regime.
• 19:17That is better.
• 19:19But what is the, you
• 19:20know,
• 19:21difficulty?
• 19:22The difficulty is the signal
• 19:23is very weak. If you
• 19:25want to increase the distance
• 19:26between source and the detector,
• 19:28the signal scales as one
• 19:30over this distance squared.
• 19:32So if you double the
• 19:33distance, that means you are
• 19:34going to, you know, have
• 19:35your intestine four times weaker.
• 19:38So what is yeah. So
• 19:40yeah. So what is our
• 19:41approach?
• 19:42Our approach is to shape
• 19:43the input wave front so
• 19:45that we can really enhance
• 19:47the light injection
• 19:48into this brain,
• 19:50or the diffuse of media
• 19:51so so that we can
• 19:52enhance this remitted signal.
• 19:54So, basically, we're showing here
• 19:56that indeed we can actually
• 19:58enhance this, okay, remission signal,
• 20:00the, remission signal enhancement
• 20:02can approach about ten times.
• 20:05And meanwhile, the sensitivity
• 20:06of this remitted signal to
• 20:08the internal absorption change can
• 20:10also be increased ten times.
• 20:12So I think my time
• 20:14is, I mean, getting close.
• 20:15So let me try to
• 20:15finish with last thing, and
• 20:17then I'll be I'll be
• 20:18done. So, of course, I
• 20:20said this one, we want
• 20:20to do non invasive. Non
• 20:21invasive can only go you
• 20:22so far. You can know
• 20:24eventually, if you really wanna
• 20:25go very deeply in, we
• 20:26want to go through the,
• 20:27you know, the fiber using
• 20:29the endoscope. Right? You you
• 20:30want to dip in there.
• 20:32And the the best is
• 20:33to use multimode fiber because
• 20:34they have many more spatial
• 20:36channels, carry more information.
• 20:37But the multimode fiber also
• 20:39give you speckle pattern because
• 20:40of this random mode mixing.
• 20:42So, again, we can use
• 20:43a wavefront shaping to focusing
• 20:45light,
• 20:46you know, at the distal
• 20:48end of the fiber by,
• 20:49you know, using this kind
• 20:50of same technique, specialized modulator.
• 20:53In fact, using this kind
• 20:54of, you know, specialized modulator,
• 20:56we can control not only
• 20:57the output, you know, the
• 20:59at a distal end of
• 21:00the fiber, what's a spatial
• 21:01pattern, but also what is
• 21:02a temporal pulse shape and
• 21:04also what is a polarization
• 21:05state. And all this can
• 21:07be used to probe the,
• 21:08you know, the molecules or
• 21:09cells at the other end
• 21:11of this, monte mode fiber.
• 21:13So with that, I think
• 21:14I'll stop here, and thank
• 21:15you for your attention.
• 21:22Thank you.
• 21:23We actually have time for
• 21:24some questions.
• 21:37Thank you. So I was
• 21:38wondering,
• 21:39in the deep tissue imaging,
• 21:41how is it calculated
• 21:42what the correction would be
• 21:44for the wavefront going in?
• 21:45I know you don't you
• 21:46don't want to get too
• 21:46too much of details, but
• 21:47how how would you know
• 21:49how to apply that correction?
• 21:50So the question is how
• 21:51do we know what the
• 21:52front end to send in
• 21:53to the that's what I
• 21:54said. We need a guide
• 21:54star.
• 21:55If you have a guide
• 21:56star, it is a fluorescent
• 21:57marker with some, you know,
• 21:59a nonlinear particles or even
• 22:00people using some bubble bursting,
• 22:02whatever, to give you acoustic
• 22:04signal. If you have something
• 22:05in there, then that will
• 22:06tells you what is the
• 22:07input wavefront. Indeed, you need
• 22:08something inside.
• 22:11On that note, like, do
• 22:12you actually do you think
• 22:13you can so back back
• 22:14to the machine learning aspect
• 22:15also, can you generate sort
• 22:17of, like, reasonably random inputs
• 22:19and actually learn how you
• 22:20can map it with various,
• 22:22like, outputs in a in
• 22:23a control system. Yeah. I
• 22:24think that's a very good
• 22:25question. Whether we can use
• 22:26a machine learning to learn
• 22:27what is wavefront we need
• 22:28to do. Right? Yes. I
• 22:28think people have been trying
• 22:30that. If you have enough,
• 22:32you know, data, then you
• 22:33can really use this to
• 22:34find out what is a
• 22:36new wavefront to do focusing
• 22:37afterwards. But, the thing is
• 22:39that usually the biological tissue
• 22:41is moving, so you you
• 22:42need to learn this very
• 22:43fast. Mhmm. So that is
• 22:44usually the typical challenge. Right.
• 22:46So good feedback is important.
• 22:47Good feedback is important and
• 22:48a quick one. Yep. Yep.
• 22:51Okay. Great. Thanks, Wei.