Neural Network-Based On-Chip Spectroscopy Using a Scalable Plasmonic Encoder.

Conventional spectrometers are limited by trade-offs set by size, cost, signal-to-noise ratio (SNR), and spectral resolution. Here, we demonstrate a deep learning-based spectral reconstruction framework using a compact and low-cost on-chip sensing scheme that is not constrained by many of the design trade-offs inherent to grating-based spectroscopy. The system employs a plasmonic spectral encoder chip containing 252 different tiles of nanohole arrays fabricated using a scalable and low-cost imprint lithography method, where each tile has a specific geometry and thus a specific optical transmission spectrum. The illumination spectrum of interest directly impinges upon the plasmonic encoder, and a CMOS image sensor captures the transmitted light without any lenses, gratings, or other optical components in between, making the entire hardware highly compact, lightweight, and field-portable. A trained neural network then reconstructs the unknown spectrum using the transmitted intensity information from the spectral encoder in a feed-forward and noniterative manner. Benefiting from the parallelization of neural networks, the average inference time per spectrum is ∼28 µs, which is much faster compared to other computational spectroscopy approaches. When blindly tested on 14 648 unseen spectra with varying complexity, our deep-learning based system identified 96.86% of the spectral peaks with an average peak localization error, bandwidth error, and height error of 0.19 nm, 0.18 nm, and 7.60%, respectively. This system is also highly tolerant to fabrication defects that may arise during the imprint lithography process, which further makes it ideal for applications that demand cost-effective, field-portable, and sensitive high-resolution spectroscopy tools.

Contact lens-based lysozyme detection in tear using a mobile sensor.

We report a method for sensing analytes in tear-fluid using commercial contact lenses (CLs) as sample collectors for subsequent analysis with a cost-effective and field-portable reader. In this study we quantify lysozyme, the most prevalent protein in tear fluid, non-specifically bound to CLs worn by human participants. Our mobile reader uses time-lapse imaging to capture an increasing fluorescent signal in a standard well-plate, the rate-of-change of which is used to indirectly infer lysozyme concentration through the use of a standard curve. We empirically determined the best-suited CL material for our sampling procedure and assay, and subsequently monitored the lysozyme levels of nine healthy human participants over a two-week period. Of these participants who were regular CL wearers (6 out of 9), we observed an increase in lysozyme levels from 6.89 ± 2.02 µg mL-1 to 10.72 ± 3.22 µg mL-1 (mean ± SD) when inducing an instance of digital eye-strain by asking them to play a game on their mobile-phones during the CL wear-duration. We also observed a lower mean lysozyme concentration (2.43 ± 1.66 µg mL-1) in a patient cohort with dry eye disease (DED) as compared to the average monitoring level of healthy (no DED) human participants (6.89 ± 2.02 µg mL-1). Taken together, this study demonstrates tear-fluid analysis with simple and non-invasive sampling steps along with a rapid, easy-to-use, and cost-effective measurement system, ultimately indicating physiological differences in human participants. We believe this method could be used in future tear-fluid studies, even supporting multiplexed detection of a panel of tear biomarkers toward improved diagnostics and prognostics as well as personalized mobile-health applications.