Issue link: https://iconnect007.uberflip.com/i/1183414
48 DESIGN007 MAGAZINE I NOVEMBER 2019 Also, attend conferences and lectures like this if you can. Connect with people from the industry and ask what their tips and tricks are because then you can use them and share them with others as well. It's awesome when engi- neers share their secrets. I wish I had more ex- perience so that I could share it and help more people because it's much easier when some- one shows you one on one rather than having to go through the book and do it step-by-step by yourself. Shaughnessy: Do you ever get the chance to talk to young people and explain how cool your job is? Jovanovic: Not really. I keep in touch with some of the younger engineering students at my school. We had a close community because we were a smaller program; we all knew each other. Shaughnessy: I think you'll find that designers usually want to help other designers. Jovanovic: I've discovered that, too. When you're working on a complicated design, and you can't find any documentation online, oth- er designers will want to help you. It's nice to know that there's camaraderie between de- signers and that we're not alone; we're all go- ing through this. Shaughnessy: Thanks for the update, Tamara. It's always a pleasure talking with you. Jovanovic: Thank you. Likewise! DESIGN007 A UCLA research team has devised a technique that ex- tends the capabilities of fluorescence microscopy, which allows scientists to precisely label parts of living cells and tissue with dyes that glow under special lighting. In a study published in Nature Methods, the scien- tists also reported that their framework, called "Deep-Z," was able to fix errors or aberrations in images, such as when a sample is tilted or curved. Further, they demon- strated that the system could take 2D images from one type of microscope and virtually create 3D images of the sample as if they were obtained by another, more advanced microscope. "This is a very powerful new method that is enabled by deep learning to perform 3D imaging of live speci- mens, with the least exposure to light, which can be tox- ic to samples," said senior author Aydogan Ozcan, UCLA chancellor's professor of electrical and computer engi- neering and associate director of the California NanoSys- tems Institute at UCLA. Deep-Z was taught using experimental images from a scanning fluorescence microscope. In thou- sands of training runs, the neural network learned how to take a 2D image and infer ac- curate 3D slices at different depths within a sample. Ozcan and his team showed that their frame- work could then use 2D wide-field microscope images of samples to produce 3D images near- ly identical to ones taken with a confocal micro- scope. This conversion is valuable because the confocal microscope creates images that are sharper, with more contrast, compared to the wide field. (Source: UCLA Newsroom) Researchers Convert 2D Images Into 3D Using Deep Learning