Optical coherence tomography (OCT) images, like those produced by ultrasound scanners, are contaminated with speckles. Coherent noise in images of highly scattering tissues acquired by OCT reduces the visibility of microscopic features. Images are usually formed from the envelope of the measured interference signals. Computation of the absolute value of the signals for measurement of the envelope is a non-linear process that destroys phase information. This study explores the idea of recording and processing the phase of the OCT interference signal before calculation of magnitudes for display. Processing of the partially coherent OCT signals in the complex domain provides the opportunity to correct phase aberrations responsible for speckle noise in OCT images. Rapid phase changes indica...[ Read more ]

Optical coherence tomography (OCT) images, like those produced by ultrasound scanners, are contaminated with speckles. Coherent noise in images of highly scattering tissues acquired by OCT reduces the visibility of microscopic features. Images are usually formed from the envelope of the measured interference signals. Computation of the absolute value of the signals for measurement of the envelope is a non-linear process that destroys phase information. This study explores the idea of recording and processing the phase of the OCT interference signal before calculation of magnitudes for display. Processing of the partially coherent OCT signals in the complex domain provides the opportunity to correct phase aberrations responsible for speckle noise in OCT images. Rapid phase changes indicate the locations at which speckle occurs in OCT. A-line signals. These changes can be detected by obtaining the unwrapped phase angle of quadrature-demodulated signals or zeros of the z-transform of windowed A-lines. A speckle-reduction technique that works in the complex domain, called the zero-adjustment procedure (ZAP), is investigated as an example of complex-domain processing. To improve the performance of ZAP, signal phase information is used to make the correction process more robust and linear. The methods are evaluated on analytical models and applied to OCT images of living skin. The results show that the methods reduce speckle contrast in regions where scatterer density is high and expand the dynamic range of the images. Besides single-channel based processing methods, incoherent summation of interference signals from multiple array elements has been shown to improve the signal-to-noise ratio of OCT images at the expense of resolution loss. This study demonstrates that image quality can be improved without loss of resolution by applying an adaptive beamforming technique based on calculation of the covariance of array signals. A prototype four-detector OCT scanner is built to evaluate the method on particle-in-gelatin phantoms and living tissue. The system incorporates a quadrature-demodulation scheme for accurate recording of the phase and amplitude of OCT signals. Compared to the images formed from the single-channel and coherently added signals, the processed images were found to be substantially sharper and less noisy. Simulations of an array-detector OCT system are done to evaluate effects of varying number of detectors and different beamforming methods. It is shown that increasing number of detectors improves beamforming results and the performance of adaptive beamforming exceeds coherent and incoherent additions.