In this paper, we systematically demonstrate two real-time CS SD OCT systems based on a conventional desktop having three GPUs. The first one takes fast Fourier transform (FFT) as the sensing technique and under-sampled linear wavenumber spectral sampling as input data, while the second one uses non-uniform fast Fourier transform (NUFFT) and under-sampled nonlinear wavenumber spectral sampling, respectively. The maximum reconstruction speed of 72k and 33.5k A-line/s were achieved for these two systems, respectively, with A-scan size 2048. It is >100 times faster than the C++ implementation and >400 times faster than the MATLAB implementation. Finally, we present real-time dispersion compensated image reconstruction for both systems.
In this paper, we proposed a novel compressive sensing (CS) method in spectral domain optical coherence tomography (SD OCT), which reconstructs B-scan image using a subset of the spectral data that is under-sampled in both axial and lateral dimensions. Thus a fraction of the A-scans for a B-scan are acquired; the spectral data of each acquired A-scan is under-sampled. Compared with the previous studies, our method further reduces the overall size of the spectral measurements. Experimental results show that our approach can obtain high quality B-scan image using 25% spectral data, which takes 50% number of A-scans and acquires 50% spectral data for each selected A-scan.
In this paper, we describe a novel CS method that incorporates dispersion compensation into the CS reconstruction of spectral domain OCT (SD OCT) signal. We show that A-scans with dispersion compensation can be obtained by multiplying the dispersion correcting term to the undersampled linear-in-wavenumber spectral data before the CS reconstruction. We also implemented fast CS reconstruction by taking the advantage of fast Fourier transform (FFT). The matrix-vector multiplication commonly used in the CS reconstruction is implemented by a two-step procedure. Compared to the CS reconstruction with matrix multiplication, our method can obtain dispersion compensated A-scan at least 5 times faster. Experimental results show that the proposed method can achieve high quality image with dispersion compensation.
In this paper, we performed an in-depth assessment of current state-of-the-art compressive sensing (CS) reconstruction algorithms, including YALL1, CSALSA, NESTA, SPGL1, TwIST and SpaRSA for use in spectral domain optical coherence tomography (SD-OCT). A brief description of mentioned algorithms and criterion in assessing performance between constraint and unconstraint algorithms are presented. The performance of all algorithms is initially assessed using a set of artificial noiseless A-scan signals with different spatial-domain dynamic range. Reconstruction error, computation time, noise tolerance and reliability of each algorithm are used as key metrics. A fair speed comparison is then implemented. Finally, computation time, SNR and local contrast of the algorithms are evaluated on real OCT Bscan data. Our results show that SPGL1 and YALL1 have moderately better performance.