Semester

Summer

Date of Graduation

2022

Document Type

Problem/Project Report

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Lane Department of Computer Science and Electrical Engineering

Committee Chair

Nasser M.Nasrabadi

Committee Co-Chair

Omid Dehzangi

Committee Member

Omid Dehzangi

Committee Member

Jeremy Dawson

Abstract

Optical Coherence Tomography (OCT) has been identified as a noninvasive and cost-effective imaging modality for identifying potential biomarkers for Alzheimer's diagnosis and progress detection. Current hypotheses indicate that retinal layer thickness, which can be assessed via OCT scans, is an efficient biomarker for identifying Alzheimer's disease. Due to factors such as speckle noise, a small target region, and unfavorable imaging conditions manual segmentation of retina layers is a challenging task. Therefore, as a reasonable first step, this study focuses on automatically segmenting retinal layers to separate them for subsequent investigations. Another important challenge commonly faced is the lack of clarity of the layer boundaries in retina OCT scans, which compels the research of super-resolving the images for improved clarity.

Deep learning pipelines have stimulated substantial progress for the segmentation tasks. Generative adversarial networks (GANs) are a prominent field of deep learning which achieved astonishing performance in semantic segmentation. Conditional adversarial networks as a general-purpose solution to image-to-image translation problems not only learn the mapping from the input image to the output image but also learn a loss function to train this mapping. We propose a GAN-based segmentation model and evaluate incorporating popular networks, namely, U-Net and ResNet, in the GAN architecture with additional blocks of transposed convolution and sub-pixel convolution for the task of upscaling OCT images from low to high resolution by a factor of four. We also incorporate the Dice loss as an additional reconstruction loss term to improve the performance of this joint optimization task. Our best model configuration empirically achieved the Dice coefficient of 0.867 and mIOU of 0.765.

Share

COinS