High-Resolution OCT To Improve Retinal Imaging

An automated algorithm developed by the Universitat Politècnica de Valènci (UPV) proved capable of distinguishing between healthy and glaucomatous eyes with an accuracy of 0.81, sensitivity of 0.79 and specificity of 0.82, making use only of OCT volumes focused on the optic nerve head of the retina. Image analysis methods termed Class Activation Maps were used to identify the regions monitored by the deep-learning network for classification.

This level of sensitivity was found to be within the range of commercial supercontinuum-based OCT systems, but more importantly the axial resolution was found to be below the key 1 micron threshold in tissue, making the OCT system capable of resolving single cells in the area of interest. This resolution was said to be possible through careful resampling and correction of the unbalanced dispersion introduced within the Michelson interferometer of the OCT system, and was described as setting “the new state-of-the-art for PS-OCT systems” by the project.

Read more