Multispectral optical coherence tomography: longitudinal spatial resolution improvement and limitations

Abstract

Optical coherence tomography (OCT) is a non-invasive, label-free imaging modality with broad applications in biomedical diagnostics. Its longitudinal (axial) spatial resolution is primarily determined by the spectral bandwidth of the light source used to acquire the interference signal. In this work, we propose a multispectral approach for improving longitudinal resolution by acquiring interference signals in several non-overlapping spectral bands and computationally reconstructing the unmeasured spectral regions.

The proposed method exploits the spectral smoothness and low-rank structure of the OCT interference signal. Reconstruction is performed using an iterative singular value decomposition (SVD) procedure. At each iteration, the dimension of the signal subspace associated with the underlying interference signal is estimated using information-theoretic criteria. Singular values and corresponding singular vectors attributed to noise or model mismatch are then discarded, and the signal is reconstructed from the retained subspace. The procedure is repeated until a predefined convergence criterion is satisfied.

The performance and limitations of the proposed multispectral reconstruction approach are evaluated using both simulated and experimental data. Particular attention is given to the influence of diffuse scattering, which may violate the assumptions underlying signal reconstruction. We quantify the range of scattering levels for which resolution enhancement remains reliable and identify conditions under which reconstruction accuracy deteriorates. The results demonstrate the potential of multispectral OCT to improve longitudinal spatial resolution while clarifying its practical limitations in scattering media.
The work is supported by Russian Science Foundation, Project 23-72-10095-П

Speaker

Nikolai Ushakov
Peter the Great St. Petersburg Polytechnic University
Russia

Discussion

Ask question