Project Summary For cellular imaging in deep tissue, adaptive optics OCT (AO-OCT) has been intensively developed by reshaping the wavefront of the illumination beam to focus the beam to diffraction-limited point spread function (PSF) in a targeted region. Due to its complexity, cost, and size, wavefront sensor-based AO-OCT is challenging to be translated into clinics. Less complicated sensorless AO-OCT(SAO-OCT) optimizes the PSF using image metrics, but cannot ensure global optimization and is susceptible to motion artifacts because image metrics must be strong and steady during the optimizing iteration. A trained Artificial neuron network (ANNs) can optimize the wavefront immediately, much more efficiently than the conventional optimization through multiple iterations. However, training ANNs with the image metric limits the generality of the ANN. We believe that the best metric for SAO-OCT should be either the PSF or its frequency domain equivalent, modulated transfer function (MTF), as they are the goals for optimization and are independent of the imaged subjects and system optics. However, the technology of accessing PSF/MTF in a scattering medium with OCT has not been proposed.OCT images originate from backscattered photons due to refractive index variation in tissue. New contrast, tissue property-related optical attenuation coefficient (OAC), has been extensively investigated to improve the diagnostic capability of OCT. However, deriving OAC is mainly based on the single-scattering model, which ignores MSPs, as conventional OCT cannot distinguish LSPs and MSPs. In addition, the single-scattering model relies on at least three interdependent parameters. Prior knowledge is needed to ensure deriving OAC successfully, but obtaining it in a clinical setting is not practical. These limitations have prohibited OAC measuring from being translated into clinics. Here, we propose reconstructing backscattered photon distribution(BPD) in a scattering medium with beam-offset OCT (BO-OCT) to resolve the above challenges. In conventional OCT, the illumination and detection beams share the same optical paths. In BO-OCT, the detection beam acquires images at offset positions from the illumination beam. The BPD can then be reconstructed with the offset images. Our theoretical prediction and preliminary data show that the distribution of LSPs is equivalent to the depth-resolved MTF, suggesting SAO-OCT can be implemented using the MTF as the metric. With the BPD, we also show it is feasible to separate LSPs and MSPs, allowing for accurately retrieving OAC by using just the LSPs to fit the single-scattering model. Real-time accessing focal depth and Rayleigh range through the BPD allow incorporating the variation of these parameters into modeling, suggesting a new method immune from motion artifacts.

Project Narrative Optical coherence tomography (OCT) is an important imaging modality in clinics. We will develop and validate a novel concept of OCT that significantly improves OCT clinical diagnostic capability, especially for high resolution deep tissue imaging and measure tissue optical property.