Optimization method for Knox–Thompson paths in phase-based optoretinography

Abstract

Optoretinography (ORG) involves imaging the intrinsic optical signal (IOS) of the retina and when used with optical coherence tomography (OCT) offers a noninvasive, objective evaluation of retinal function. The recently proposed Knox-Thompson (KT) method, a phase-based ORG, induces fluctuations and noise in the ORG signal based on the selected path, referred to as the KT path. However, the strategy for selecting this path has not been firmly established. Therefore, we implemented Dijkstra's algorithm, a renowned shortest path algorithm, to select the KT path to minimize noise in phase-based ORG data. Our study used a healthy pigmented rabbit as the test subject. IOS images were captured using point-scan swept-source OCT with a white light-emitting diode with a 400-800 nm wavelength range used as a light stimulus for approximately 6 s during the ORG. The light intensity was adjusted to ensure a photopigment bleaching level of 63%. In the KT method, the reference time is constantly updated to construct the KT path. Changes in thickness were calculated by correlating the phase difference with varying reference times. The KT path was optimized with Dijkstra's algorithm. Thus, the ORG signal from the optimized KT path exhibited less phase dispersion than that from the KT path with a fixed time interval and produced smoother changes over time in the thickness between the ellipsoid zone and the retinal pigment epithelium. We believe that this improved method will contribute to the practical implementation of OCT-ORG in clinical settings.