Graphics processing unit based dispersion encoded full-range frequency domain OCT

Standard FD-OCT systems suffer from a limited useful depth range due to the inherent complex conjugate artifacts and continuous fall-off in sensitivity with distance from the zero delay. The techniques of dispersion encoded full range (DEFR) frequency-domain optical coherence tomography (FD-OCT) and its enhanced version fast DEFR use the dispersion mismatch between sample and reference arm to double the imaging depth range by iteratively suppressing complex conjugate artifacts. Previously the computational complexity of DEFR prevented its application to fields where real-time visualization or large volumetric datasets are needed. A graphics processing unit (GPU) with hundreds of processing cores provides highly parallel computation capability to FD-OCT in which processing for each A-line is identical and independent. In this paper, we adopted GPUs to accelerate DEFR, thereby significantly improving reconstruction speed by a factor of > 90 in respect to CPU based processing. A maximum display line rate of similar to 21 k-lines/s for 2048 points/A-line using 10 iterations of the fast DEFR algorithm has been successively achieved, thereby enabling the application of DEFR in fields where real time visualization is required. By comparison in the conjugate artifact suppressed cross-sectional image of a mouse eye, there is no significant qualitative difference between the corresponding CPU- and GPU-processed images.