 Welcome everyone. Good morning. Thank you for joining us for the second celebration of Associate Professors event this semester, this fall semester. We have a lot of events that we organized from the college, right? Seemingly every day there's an email about some event, but honestly this one is amongst the most heartwarming ones that we organize here, because what we intend to do here is a celebrate the success of a recently promoted Associate Professors. Those who have been through, you know, that that stretch of career know how difficult it was. Perhaps some of us who have been through it have forgotten about it, but those who are in the process of going through it or those who hope to go through it in the future can learn a lot from some of the key insights of decisions that these colleagues have taken, what made them successful, etc. So that's the second part of what we try to do here is to share the wisdom from that experience. A third part is to try and get new collaborations. There are many faculty colleagues and I also like to welcome all those who are joining us virtually today. Thank you for joining us, but this is an opportunity for a wider cross-section of faculty to get to know about you and to know about your work, and in fact we do know that some of these discussions lead to future research collaborations as well, so hope that also happens. So without further ado, I'd like to call upon Yang Kim to introduce our first speaker, Feng Huang, for the day. So please, Yang. Good morning everyone. It is my great pleasure to introduce our rising star, Professor Feng Huang. Professor Feng is currently a Riley associate professor of biomedical engineering at Weldon School. Professor Feng received his bachelor's degree in physics at the University of Science and Technology of China in 2004 at his very early age, and his PhD in physics from the University of New Mexico in 2011, and his dissertation is about ERK1 dimerization modeling using fluorescence correlations spectroscopy and as well as single molecule super resolution method he's studying. Then he went on to post-doctoral training in cell biology at Yale University, and then Professor Feng joined Purdue Engineering in 2015, and he had a very interesting journey from physics, biology, and finally in engineering, so I want to actually ask your physicist or your engineer or your biologist, actually. I really want to know your identity. He received numerous awarding recognitions. In particular, he received the DARPA Young Faculty Award and received approximately $10 million in funding, including four or five large NIH grants. He has, I think, a few more at this moment. His research focuses on creation of novel microscopy method, including hardware and as well as algorithms that enable single molecule and super resolution imaging. Actually, he's known as a professor in our unit, publishes only nature papers, only nature papers actually. As a researcher, key aspect I truly respect is he's an excellent example of resilient researcher. One lesson, actually, including me, is that a bold and innovative mindset will be paid off if we are persistent. I really hope that we can really cultivate this culture in our unit as well. Without further ado, please join me in welcoming Professor Huang. Thank you so much. Thank you so much, Yang, for your introduction. It's a pleasure for me to present my journey or the student's journey, actually, to achieve all the success I'm going to show you today. My name is Fang. I'm an associate professor of Biomedical Engineering Department and my lab works on high resolution optical imaging technologies to review biological structure that is unresolvable using conventional methods. So for example, if you look at this picture of a confocal nuclear pore complex proteins on CO7 cells, all of these thoughts are nuclear pores, individual clusters you can say, we call it puncta. And in fact, that if you look at the super resolution image, you can see the picture is much better resolved. And in fact, if you look at cross or subsection of that, it is very apparent that high resolution in super resolution image give you much a clear view in term of resolution as well as it reveals that not 98 orbital positions potentially form this 40 to 60 nanometer ring like structure. So how do we solve that? And how do we get this high resolution image? The concept lies in to localize the individual isolated single molecule emitters precisely in their centers and reconstruct their centers into the super resolution image on the right over here. As you can see, as the more molecule being reconstructed, the higher resolution, you can see the better that are revealed such that these structures that are traditionally under diffraction limit can now be seen using super resolution technology. And this work is developed, this concept developed in 2006, won the Nobel Prize in 2014. Now, as you can see, you probably can imagine now cell biology processes or dynamics cannot be revealed in cells, tissues and animals with as high resolution as you want. And therefore, testing a biological hypothesis will be simply taking a photo or movie on the process you want to investigate with sufficient resolution. While this is not the case, at least yet, there's tremendous challenges that still exist before we can claim that. For example, taking this photo take a very long time. You have to isolate individual molecules, therefore taking thousands or even millions of images sometimes to construct one image. And this is very detrimental for life cell imaging where dynamics is what you are seeking. Single molecule emission patterns generate aberrations scattering also disrupts the features that we want to see in tissues because light propagate in tissue constitutes a different speed. At the same time, if you want really high resolution, one to five nanometers, you really need novel instrumentation and conventional method to detect photons, such that photons encode much more information than the current methods. So, MATLAB focused on three directions on these front or challenges that before we can claim really a revolutionary technology to solve biological hypothesis. We want to put super resolution in life cells, and we also want to transition them from cells to tissues to small animals. For example, over here in the middle, you can see an amyloid beta fibula structure in mouse frontal cortex in brain slices. We also developed novel technologies such that it will encode more information per photon count. Even with 500 photon count, we want to achieve one to five nanometer precision in localizing their centers. Mike and Paul spearheaded in the direction of high precision localization of single molecules through tissues. Here, they developed adaptive optics and also developed PSF engineering technologies using deformed mirrors really to move super resolution from cells to tissues. Here's a picture of amyloid beta fibula in 30 micron mouse brain sections in Alzheimer's disease model. And you can see the drastic contrast between diffraction limited image and super resolution image, for example, over here and here. And here's the other examples. Pei Yi and Sheng developed method they decided to use deep learning to tackle single molecule analysis. Single molecules can have really complex patterns, for example, this one over on the bottom left. And these complex patterns can encode molecular information in multiple classes. It can encode position of the molecule, orientation of the molecule, as well as a shared wave front of the molecule. It is very difficult to decode them using traditional statistics and inference method. So they developed deep neural network for a single molecule we call SMNAT, which are capable of inferring 3D location orientation and wave front distortion simultaneously from a complex in molecule pattern. This network does not only extract the information, but extract the information at the statistical theoretical limit which is demonstrated in these two figures. And here's a beautiful picture they reconstructed using their deep learning method showing the crispy structure of mitochondria membrane contour labeled by Tom 20 proteins. Sheng and Yi Lun spearheaded in the high speed acquisition. We want to use a high speed camera to collect high speed dynamics of cells. However, those cameras have caveats. Each individual pixel have different fluctuations. Therefore, if you want to look at the image, you're actually not only looking at the fluctuation of the biological signal, which is really what you want to get. You also have sensor fluctuations. And therefore, if you look at the picture in the middle on the red part, you can see all the dots over here showing the fluctuation of single pixels by themselves. It's not necessarily what we want to sense. So what they do is they combine the statistical knowledge and optic knowledge together and to separate the noise with a signal. Whereas the OTF boundary sets a boundary between signal and noise contribution towards noise only contribution. Therefore, they can minimize the noise without affecting the signal. So on the bottom, you can see each individual pixel give you fluctuations that we want to eliminate. And after their algorithm, noise correction algorithm or NCS for short, you can see the contribution from pixel sensor noise is already gone. And you do not have any effect on your signal. And Yi Lun actually further calculated the lower bound by statistics and also the constraint of OTF over here and demonstrate that NCS actually performed very close to the lower bound of the statistics. Sha and Fernan actually go ahead and invented another very unconventional Mexico. Instead of looking at the horizontal plane of the specimen, which you usually put the specimen in there and you look at the horizontal plane. But this Mexico look at vertical plane directly without scanning. So this allows you to image up to 20 micron depth with one shot without scan. And over here you can see a drastic difference between a diffraction limited image and super resolution image over here. And here's also knob 96, knob 98 labeled with the Alexa C47. And here's a super resolution image. Here's the contrast diffraction limited and super resolution diffraction limited and super resolution image. This is a very new setup and both of them work very hard to achieve this. Fernan Donghan furthermore are trying to tackle one of the long standing problem in single molecule imaging. As you know that our business depend on localizing centers of the molecule. However, those centers of the molecule, what model you use to model it? We take a bead image on the coverset. And these are never accurate. You're trying to image in the tissues, the tissue photons experience tissue structures while the bead never experience those. So what we are using is an in vitro model. And Fernan Donghan developed this method called insidio PSF retrieval. Directly retrieved points by function from single molecule blinking data. And therefore they can use their algorithm to accurately localize single molecules in spite of aberration introduced by tissues. Here's a picture of diffraction limited amyloid beta fibrils inserted micron brain sections. And you can see the wonderful reconstruction of high resolution image that you can trace the three dimensional fibril in a very high accurate manner. And Donghan Amayam now move forward to trying to get the system into a much higher resolution and depth. How John and Shun and Fan and Vermeira takes a lead on tackling one of the most difficult microscopy system in the field for high single molecule imaging system. Where we're trying to image the sample, the sample is sandwiched between two cover slips. You have two objective coherently detecting the fluorescence emitted by every single molecule. By coherently combining them together, you form a single molecule interference pattern that generates modulations that give you enormous improvement in the axial resolution of the system upon the super resolution system I already told you. So about five to seven times expected. As you can see here it's our initial result that Fan and Hao Zhang already demonstrate the possibility doing four pi single molecule imaging through tissue specimens. Most recently, Pei Yi developed a new method leveraging her previous invention on single molecule neural network to use a network controlling a default mirror, a device that can optimize your microscope's penetration depth and resolution. So as you can see over here, you have a fairly blurred image of single molecules. Every single molecule is a little bit distorted. Now we're going to perform deep learning driven adaptive optics. After even one cycle of compensation, you can see the aberration being drastically compensated. And over time, you can see the compensation result in a PSF response that as if it never went through the specimen. And we have Yue Zhang over here, which is our organic chemist developed organic diet for lifestyle imaging. Maddie is our ER expert, is our link to diabetes. Li Fang already generated tremendous insight for our understanding on how aberration affects resolution and how we aberrations affect how we correct aberrations, which is a very interesting topic. And Cheng is our self allergist and an expert on single molecule dynamic analysis. And their work is ongoing. I hope to share with you in future. All of that is not possible without support for my mentors, collaborators, our heads and our colleagues. So we'd like to thank David for his support and advocate for our research and also collaboration. And Dave Kish, who helped us every single time almost within two or four hours, our need in research. Eugenio, who actually taught us what is deep learning after all. And Mao Ban, Charlie, who helped us to recruit me here to Purdue, which is a wonderful place and the best decision in my life probably, other than my marriage. I have to say. And he also, they also helped me a lot on initial grant applications and also research ideas. Kenan Park for continuous support and advocate for our research and nominating me for multiple awards. Richard and Ram's and for their help during a very difficult time in 2019 for the lab. Alex, Tron Lee, Daniel and Gary for being a wonderful collaborator. We look forward to working with you further to advance technologies as well as to probably solve questions and George for their support. And we'd like to send Keith Slicky and York Beaver of my PhD and postdoc mentors and also Tom Pollard, who we have been working with for more than 10 years and we're still learning how to do research from him. And thank you very much. Wonderful for your research. I think it's a time for questions or comments. We do have questions. So anyone participating on the Zoom, I think you can type in your questions so that we can read. I have two questions on the ratings. Number one is more technical question and number two is like a general in life. Number one is do you have any kind of target objective that the resolution that the best resolution you want to achieve with optical microscopy? I think what's the best resolution so far you have like 10 nanometers, five nanometers? I don't know. It depends on the depth and et cetera. That is true. So there's two directions in the lab as I mentioned potentially deeper into the specimen. But the deeper you go, the worst resolution you will definitely get. And we want to mitigate that. D4 mirror is one of the ways to do that, adaptive optics and other ways potential advanced illumination technologies. We are also working on in the other direction of high resolution. And in term of exact resolution, I will be comfortable from one to five or one to three nanometers. That will be our goal. Certainly we have not achieved it. And we also are investigating possibility to find those structures in biology such that we can really take advantage of that resolution to review new processes or dynamics. So it's ongoing and it's evolving all the time, I will say. So one to three nanometers are using optical microscope. Today I will say yes. Wow, it's impressive. Yes. Oh yeah. And the second one actually is that are you a physicist? Which title you're most proud of? Physicist, biologist, or engineer? I'm trying to be a physicist. But a lot of my work requires a lot of engineering which I really enjoy. And I also enjoy working with biologists to solve biological questions. One of my dream course to teach is cell biology. If I can teach that course, can I call myself biologist? So I really, as you can see in the lab, we have chemistry in the lab. We have cell biologists in the lab. We have physicists, engineers, all these disciplines. And that's all I can say. Yes, I do. Okay, sure. Jan Alibak asks, how do you get ground truth information for your deep learning based methods? That's a very interesting question. So for single molecules, many of our ground truths can come in from a fiduciary marker. That will be a bead on the cover set. We can potentially localize. And we have very high precision stages like one to three nanometer or zero point three nanometer resolution. So we can move the stage and see whether our deep learning give the exact result. But do we have ground truths in biology? We don't. The only thing we can validate is we can know, we can image some known biology structures and see whether these structure looks like they are supposed to be, for example, in an electron microscope when it's shallow depths. So two questions, Feng. Sure. By the way, congratulations. I love the results every time I see them. But you spoke of two important decisions, right? One is life partner. The other is, you know, coming to Purdue. What about decisions you've taken after that? What are some key decisions after you joined Purdue that helped guide you to this success? Can you share some thoughts on that? And maybe they're not as important as the first two, but they're still important for many others here. So I think the most important decision is having a whole lab member joining the lab. And I've been very lucky to have very productive and lab members that work together with the whole team and they collaborate. And different people have different mentality, but they all willing to work together and contribute to the knowledge growth of the entire lab. I think that's probably the number one careful thing that I'm trying to let whom join the lab. And the other decision would be collaborations. And single molecule is not easy technology to implement. And yearly is not that confocal. And collaboration is extremely important. And our collaboration so far has been very persistent in using our technology and optimizing the manner. Every single picture that I showed you is not within obtainable within three months. It's always like almost half a year or a year optimization before you can see the picture over here. But during the process student and postdoc learn a lot, our contribution actually synergize together. And we produce something that is difficult to produce on the first shot, which is actually a very rewarding experience. And again maybe going to Kinam's, well Kinam is not online here. Is he online? He's okay. Yeah, I know. But what would be the grand challenge problem in biophysics or in biochemistry that some of these techniques will be able to unlock? Very interesting question. The technology we're trying to pursue has a broad application area I would say. And in term of which question is more important than the others, I will let the biologists consider. Very interesting talk. So I enjoyed it. For those fantastic final images, you take a lot of time to get from very initially very blurred original data. But to trust, put trust on the final optimized image. Is there a standard case that you can calibrate your process that the one you obtained finally actually is the physical image rather than numerical artifacts? Yes. So it's important to calibrate all these methods. And we have different methods to calibrate. We have DNA origami that we can manufacture. And while in the end you still don't know what the size of DNA origami is. And it varies from different particle to particle. But at least they give you an average effect of how much you are expecting. On average about two sites. For example, 50 nanometers. And we also have model system used and typically using molecule imaging. And we have a biological system called nuclear pore complex systems that I showed you in the beginning. But there's another subunit you can label that which have very well measured EM units. So they know the copy number as well as the distance between those. And these are developed at EMBL. A wonderful technology that we can use to calibrate. And these are very important and calibration methods we use. Yes. Thank you. I'll have the last question actually. So I know you're at the forefront of highly, highly competitive this optical microscopy community. And how do you kind of manage your stress or to have any secret to be successful in this highly competitive scientific community? First of all, I cannot say myself successful. But looking at all these pictures, I can say our projects, many of them are successful. So how do we achieve that? And we have students and we also have study conversation and journal clubs we often hold and on weekly basis. We meet every two twice every week to dissect all the important papers in the field. And during that time we dive pretty deep into that. And overall I think the field also has many, many publications. And we can cluster them into different knowledge bases. And therefore evolve and the knowledge also evolve in students and postdocs. So that they have gradually, not very fast, but gradually they have a cohesive understanding of how, why this thing goes certain direction, why this thing goes not. Therefore it's effectively reduce the number of readings or number of careful studies we have to perform. Because as you said there's many publications on this field and we have to filter through this. Great. Thank you so much and congratulations again. And I hope to, I'm looking forward to seeing you more on progress and more nice images. All right. Thank you. Thank you so much.