Model-based time- and frequency-domain processing of single optical coherence tomography A-scans.

摘要

An essential goal of optical coherence tomography (OCT) is to develop methods and instrumentation that yield high-definition visualization of tissue structures co-registered with accurate and precise quantification of blood flow in non-stationary situations. Previous OCT enhancements have improved the acquisition speed, with, however, the need for averaging to compensate for lower signal-to-noise (SNR). To this day, information is extracted from OCT signals using Fourier analysis, which is limited by the trade-off between spatial and velocity resolution. We sought to improve measurements from a single OCT A-scan.; We developed a simulator that allows one to understand better the impact of source and tissue parameters on OCT signals. We introduced a model-based approach to processing OCT signals and obtaining statistics from a single A-scan. Statistics and error estimates allow for diminution of unreliable data, objective analysis, and information visualization of OCT measurements. Decimation and windowing were implemented to improve the SNR. A parabolic model based on linear regression was developed to provide rapid error estimates on the measurements. The parabolic model was then expanded to yield exact solutions to a conic model that can identify, isolate, and quantify asymmetry in the OCT power spectrum. Such conic models can potentially model and quantify shear rates, source asymmetry, or system distortions.; We then developed velocimetry techniques based on the Doppler effect and applied them to single A-scans acquired in ∼1 μs with a rapid-scan system (RSOD). Results indicate the accuracy and precision of both the amplitude and frequency (i.e., velocity) calculations based upon our time-domain processing. Spatial and velocity resolution are limited ultimately by the acquisition rate and resolution of the data acquisition board. We explain how time-domain processing of OCT signals can improve frequency and spatial resolution without violating the uncertainty principle that limits Fourier-based methods. Finally, we showed that a combination of OCT and ultrasound induces acousto-optic interactions that measure non-optical material properties with the resolution of optical methods. The volumetric nature of acoustically-enhanced OCT should allow one to image structures with diffuse boundaries, such as infiltrating tumors.