Phase-corrected complex averaging improves signal-to-noise ratio for <i>in vivo</i> retinal imaging using optical coherence tomography

Optical coherence tomography (OCT) has advanced toward higher A-scan rates, enabling faster acquisitions and reduced motion artifacts. However, due to the maximum permissible exposure in biomedical imaging, increased imaging speed compromises signal-to-noise ratio (SNR). While magnitude averaging is widely used to boost SNR, complex averaging provides higher theoretical SNR gains. Nevertheless, its in vivo application is hindered by phase fluctuations from bulk tissue movement and pulsatile deformations. In this study, we conduct theoretical analyses to highlight the importance of phase stability in achieving effective complex averaging. Accordingly, we propose a signal processing pipeline—including subpixel image registration and subsequent bulk or local phase correction—to minimize undesired phase fluctuations. We validate the proposed phase-corrected complex averaging method using a moving phantom and in vivo retinal imaging in rodents, a non-human primate, and a human subject, utilizing both spectral-domain and swept-source OCTs. The results demonstrate that phase-corrected complex averaging consistently achieves higher SNR gains compared to magnitude averaging in the retina. Notably, in low-SNR scenarios, complex averaging reveals retinal features otherwise obscured by noise, thereby improving diagnostic potential. Our study highlights the significance of integrating complex averaging techniques into clinical OCT systems to enhance SNR without requiring hardware modifications.