Abstract
Automated measurements of the human cone mosaic requires the identification of individual cone photoreceptors. The current gold standard, manual labeling, is a tedious process and can not be done in a clinically useful timeframe. As such, we present an automated algorithm for identifying cone photoreceptors in adaptive optics optical coherence tomography (AO-OCT) images. Our approach fine-tunes a pre-trained convolutional neural network originally trained on AO scanning laser ophthalmoscope (AO-SLO) images, to work on previously unseen data from a different imaging modality. On average, the automated method correctly identified 94% of manually labeled cones when compared to manual raters, from twenty different AO-OCT images acquired from five normal subjects. Voronoi analysis confirmed the general hexagonal-packing structure of the cone mosaic as well as the general cone density variability across portions of the retina. The consistency of our measurements demonstrates the high reliability and practical utility of having an automated solution to this problem.
Highlights
Adaptive optics (AO) techniques have been used to facilitate the visualization of the retinal photoreceptor mosaic in ocular imaging systems by improving the lateral resolution [1,2,3]
Similar to AO-SLO and AO-FIO, AO-optical coherence tomography (OCT) systems can provide high-resolution en face images of the photoreceptor mosaic in which the cones appear as bright circles surrounded by dark regions [7,8,9], but with the added benefit of high axial resolution which allows for cross-sectional tomography images in which the different outer retinal layers can be clearly delineated
We present an effective transfer learning algorithm for retraining a convolution neural networks (CNNs) originally trained on manually segmented confocal AO-SLO images in order to detect cones in AO-OCT images with a different field-of-view (FOV)
Summary
Adaptive optics (AO) techniques have been used to facilitate the visualization of the retinal photoreceptor mosaic in ocular imaging systems by improving the lateral resolution [1,2,3]. As manual segmentations are both subjective and laborious, several automated methods for detecting cones in AO images using a variety of traditional image processing techniques such as local intensity maxima detection [22,23,24,25], graph-theory and dynamic programming (GTDP) [26], and estimation of cone spatial frequency [27,28,29] have been developed These mathematical methods work for the specific data for which they were designed, reliance on ad hoc rules and specific algorithmic parameters does not allow for alternative imaging conditions, such as different resolutions, areas within the retina, and imaging modalities
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
