Real-time tomographic imaging without x-rays: a smart pixel array with massively parallel signal processing for real-time optical coherence tomography performing close to the physical limits

Abstract

Previous research has demonstrated the feasibility of pointwise cross-sectional imaging of semi-transparent, scattering biological and technical objects with micrometer resolution, using Optical Coherence Tomography (OCT). The goal of the present work was to understand the fundamental limitations of OCT imaging and to design and implement a practical real-time OCT system that performs close to the theoretical limit. The key to this realization has been the design of a dedicated CMOS image sensor with massively parallel signal processing on the pixel level. The pixel architecture is based on a modified I-Q direct demodulation principle, carefully optimized for photon shot-noise limited signal detection. At the same time, the signal chain has been designed for low-power operation, e.g. by employing switched capacitor techniques for low-pass filtering, requiring a power dissipation of only a few μW per channel at full OCT data acquisition and processing speed. Based on this dedicated CMOS imager, a working prototype of the fastest time-domain OCT system has been implemented, and close to shot-noise performance has been measured. This accomplishment has led to the creation of a start-up company offering the world’s first real-time OCT systems for tomographic/topographic applications in industry and in life sciences.