Bilkent University
Department of Computer Engineering


Deep learning for segmentation of Henle’s Fiber Layer in optical coherence tomography imaging


Selahattin Cansız
Master Student
(Supervisor: Prof.Dr.Çiğdem Gündüz Demir)
Computer Engineering Department
Bilkent University

Abstract: The Henle's fiber layer (HFL) in the retina holds valuable information on the macular condition of an eye. However, in the common practice, this layer is not separately segmented, instead it is included in the outer nuclear layer since it is difficult to perceive HFL contours on standard optical coherence tomography (OCT) imaging. Due to its variable reflectivity under an imaging beam, directional OCTs are required to segment the HFL contours, which means additional imaging. In our work, we propose a shape-preserving network that achieves HFL segmentation in standard OCT scans with the target performance obtained when directional OCT scans are used. The cascaded network design benefits the shape prior of HFL in the network training. This design proposes to represent the shape prior by extracting Fourier descriptors on the HFL contours and defining an additional regression task of learning these descriptors. It then formulates HFL segmentation as concurrent learning of regression and classification tasks, in which Fourier descriptors are estimated from an input image to encode the shape prior and used together with the input image to construct the HFL segmentation map. The experiments on 1470 images of 30 OCT scans reveal that quantifying the HFL shape with Fourier descriptors and concurrently learning them with the main task of HFL segmentation lead to better results. This indicates the effectiveness of designing a shape-preserving network to improve HFL segmentation by reducing the need to perform directional OCT imaging.


DATE: 28 March, Monday @ 15:30