Lekang Yuan

I am a third-year Ph.D. student in the ECE department at Princeton University, advised by Prof. Felix Heide. Previously, I received my M.S. and B.S. degrees from the Department of Automation at Tsinghua University, advised by Prof. Qionghai Dai.

My research focuses on building some of the world's smallest cameras, pushing the boundaries of where imaging can go, with applications spanning neuroimaging, microscopy, multimodal imaging, augmented reality, and robotics. I'm also exploring how miniaturized systems can enable scalable, AI-driven discovery and design in the physical world.

Email  /  CV  /  Scholar  /  Github

Research

* indicates equal contribution.

A compact nanophotonic spectral imager that brings ultrafast spectral imaging out of the lab, deployable on smartphones, robotic manipulators and ultralight mini drones.

AirScope is a 1-gram wireless mesoscope for cortex-wide neuroimaging during natural behavior, recording up to 8,500 neurons across a 6 mm field of view in freely roaming mice.

Glint-free eye tracking for smart glasses that jointly optimizes a nanophotonic metasurface, the sensor pixel pitch, and the reconstruction network, so the optics physically disentangle eye rotation from glasses slippage — a 0.75° gaze error at 480 MMACs.

A Systematically Optimized Miniaturized Microscope (SOMM) for studying neural populations in freely moving animals. By optimizing a diffractive optical element (DOE) under strict form factor constraints, SOMM achieves 4 µm resolution across a 3.6 × 3.6 mm² field of view and 300 µm depth of field, all within a lightweight 2.5-gram system. This design enables the recording of over 3,000 neurons with minimal impact on the animals' motion.

This paper introduces a progressive optimization pipeline to optimize aspherical lenses and diffractive optical elements with significant memory reduction. The optimized microscope that outperforms a commercial 5×, NA 0.1 objective microscope while being only 0.15 cm³ and 0.5 g—five orders of magnitude smaller. The microscope is integrated into a cell phone for portable diagnostics, offering a novel approach to designing miniaturized, high-performance imaging systems.

DNA is an appealing data storage medium in the era of data explosion due to its very high information density and longevity. This paper introduce DeSP, a systematic DNA data storage error Simulation Pipeline which covers all types of error generated in the data storage process. DeSP enables systemically optimization of the redundancy design in silico to combat the channel's particular noise structure.

Website template from Jon Barron.