Signal to noise ratio in balanced OCT under wide band illumination

In optical coherence tomography (OCT), it is often assumed that the signal-to-noise ratio is limited by shot noise. However, a high data acquisition rate and a low target reflectivity require operation under high optical power. Balance detection is used with the aim to bring the noise close to shot noise regime, but this is not all the time achievable in practice. If the balance detection is ideal, the limiting factor is the beat noise. This is proportional to the stray reflectance in the object arm and inverse proportional to the effective noise bandwidth. It was often noticed that the limiting theoretical value for the S/N ratio for a given stray reflectance and optical source bandwidth could not be experimentally achieved. In the present study we develop a new model where we address this issue by taking into account the limited spectral response of fiber based balanced detection receivers. We show that due to mismatches in the balanced receiver, excess photon noise has a larger contribution than normally expected, with important implications in the maximum achievable SNR. The theoretical model developed leads to a redefinition of the effective noise bandwidth to take into account the non-flat response of the directional coupler used in the balanced stage. The model is capable to explain the limitation of SNR observed in practice when stray reflectances within the interferometer are brought to infinitesimal values. The model guides optimization of parameters in order to maximize system performance, for a given optical source power and directional coupler characteristics. In this paper we present experimental results to validate the theoretical model. Such S/N analysis is paramount in the modern OCT technology, which makes use of wide bandwidth sources in the quest for high-depth resolution.