Lei Tian: Augmenting Optics with Algorithms
When Lei Tian, PhD, joined Boston University’s Electrical and Computer Engineering department in 2016, he never imagined his work would take him into the neurosciences. But when the Neurophotonics Center at BU launched the following year, bringing together a multidisciplinary team from all four corners of the university, he couldn’t help but get involved.
“When I started with BU, neuroscience was not one of my ambitions,” he says. “I was thinking about other things.” After a series of conversations with NPC faculty, though, he realized the potential impact of his work in neuroscience studies. “A lot of people these days talk about how super-important data is for the biomedical science. After talking with researchers in the Neurophotonics Center, I realized that, if we can help them process the data better, they can see more, understand more, in the studies they are doing.”
The main thrust of Tian’s research has always been computational imaging, developing computational algorithms and optical imaging technologies that work together synergistically to address a host of applications. For instance, one of the threads of his work has been the development of a computational label-free microscope combining programmable illumination, multiple-scattering physical modeling and advanced reconstruction algorithms to enable tomographic reconstruction of unstained biological samples – work that was recognized by an NSF CAREER award in 2019.
Today, in collaboration with investigators from across the NPC, Tian and his group are creating cutting-edge solutions for studies producing very different kinds of neuroimaging data – and, in doing so, advancing the state of the art for computational imaging.
Just one example: In an October 2020 issue of Science Advances, Tian’s group, in collaboration with Ian Davison’s team in the Neurophotonics Center and NPC director David Boas, reported a wearable microscope for use in mouse models. Wearable instrumentation has recently made considerable waves in the neurosciences: not least because it enables studies in freely moving animals, yielding insights into the brain that would not have been possible in conventional imaging studies, where the animals are typically physically constrained. But because of engineering and other challenges inherent to miniaturized wearable microscopes, many of the instruments developed for such applications have been limited in application.
Tian and colleagues tackled these challenges decisively in the 2020 Science Advances paper, in which they describe wearable instrumentation called the Computational Miniature Mesoscope, or CM2. (The acronym doubles as a reference to the area of coverage it provides – 1 centimeter squared – which is approximately the area of a mouse’s brain.) In developing the CM2, they employed computational imaging – augmenting the optics with algorithms – to facilitate a robust design enabling single-shot 3D imaging across an 8 mm by 7 mm field of view with a 2.5-mm depth of field with high resolution.
Ultimately, they hope to wield the technology to image large populations of neurons across the entire brain cortex of a mouse, potentially opening the door to an array of new neuroscience applications.
—–