Feature Of The Week 2/13/11: Influence and Compensation of Autocorrelation Terms in Depth-Resolved Spectroscopic FD-OCT: A Review of Work from Bern University of Applied Sciences
Feature Of The Week 2/13/11: An extension to regular Fourier-Domain OCT, spectroscopic optical coherence tomography (SOCT), enables the extraction of spectral characteristics from the OCT signal of a sample or substance under investigation. For medical imaging, especially the power of SOCT to detect depth and spectrally resolved attenuation coefficients of endogenous chemicals and tissue is of interest as it could pave the way towards a variety of applications. Possible areas of usage include non-invasive assessments and monitoring of blood oxygen saturation and blood glucose levels or the detection of precancerous cells, but also other applications, for example in dermatology, are subject to intense research worldwide.
However, because of the complex and scattering nature of biological tissue, it has been a big challenge in SOCT research to obtain a stable and reliable measurement of attenuation coefficients and to associate them to the corresponding tissue structures. For applications in samples with multiple reflective layers such as biological tissue, an additional obstruction, the beating effects caused by autocorrelation terms, might prevent a meaningful detection of spectral characteristics.
If, as presented in here, a narrow filtering of the spatial-domain signal is employed to isolate the signal parts from certain sample depths and if the peak from the sample interface is positioned too close to an autocorrelation peak, both peaks will be present within the filter bandwidth. The resulting spectral analysis will show a superposition of the cosine functions of both the Fourier-transformed reflection and autocorrelation peak. Due to vibrations and other effects, even minimal changes in the optical path length difference will result in a phase drift and will significantly alter the spectral response due to the beats modulated onto the reflection spectrum.
The need for strong sample peaks and the resulting choice of strongly reflecting layers for the SOCT analysis leads to comparable magnitudes for the sample and reference reflectivities. The artifacts severely derogate the measurements of the reflection spectra and therefore the extraction of the spectral absorption values. The effect described becomes more severe if strong autocorrelation terms are present which is unavoidable if strong sample signals are needed. However, an influence of the autocorrelation terms on the reflection spectrum could also be seen with autocorrelation signals not much stronger than the noise level.
In the Article below, the authors focus on measurements of the spectral absorption of liquid Indocyanine-Green (ICG) samples with a self designed OCT system, realized with a simple, low cost spectrometer and a broadband multiple S-LED light source. Additionally, the authors present the influence of autocorrelation terms and sub-wavelength vibration on the recorded spectral attenuation values and propose a simple and effective method to diminish those effects in SOCT analysis.
A robust prevention of autocorrelation terms in the signal to be analyzed is needed for a non-constrained and stable SOCT post-processing. It is shown that a thorough background subtraction will gradually reduce the autocorrelation peaks present in the signal. Nevertheless, with changing background signals due to fluctuations of the signal, changes in the sample and because of minimal phase shifts between sample and reference arm, the background subtraction has to be done immediately before the actual measurements, possibly by using shutters in both sample and reference path.
For more information see recent Article. Courtesy Patrick Steiner.