ALAM-DISSERTATION-2020.pdf (11.51 MB)
Download file

Quantitative Analysis and Automated Classification of Retinal Images

Download (11.51 MB)
posted on 01.08.2020, 00:00 authored by Minhaj Nur Alam
Retina is a complex sensory tissue located at the back of the eye, often considered crucial for diagnosis of systematic diseases and retinopathies. Therefore, quantitative retinal imaging and developing imaging biomarkers are of great scientific and clinical interest. In current literature, color fundus photography has been most commonly used for eye disease screening, diagnosis and treatment assessment, but the spatial resolution and image contrast are limited to reveal subtle distortions in early stages of eye diseases. Other imaging modalities such as scanning laser ophthalmoscopy (SLO) and adaptive optics (AO) imaging systems are unable to differentiate individual retinal neural layers and vascular plexuses. In recent years, optical coherence tomography (OCT) has been extensively employed for depth-resolved examination of morphological abnormalities due to its unprecedented capability to differentiate individual functional layers. Adding power to the OCT, OCT angiography (OCTA) is a new imaging modality that provides high resolution blood flow information in individual retina plexuses. However, since it is a new imaging modality, quantitative OCTA analysis and investigative studies are required to standardize objective interpretation of clinical outcomes. In this dissertation, extensive studies have been conducted to investigate OCTA features for quantitative analysis and objective classification of different retinopathies. Within the overarching scope of this dissertation, new OCTA imaging biomarkers were developed, strategies for artery-vein (AV) classification in OCTA were demonstrated, and the OCTA features were demonstrated and validated for diagnostic analysis and machine learning based automated classification of retinal diseases. Utilizing the developed OCTA imaging biomarkers, AV classification techniques and AI based classification tools demonstrated in this dissertation can be beneficial in providing diagnostic support to ophthalmologists and efficient clinical screening of different types of retinopathies.



Yao, Xincheng


Yao, Xincheng



Degree Grantor

University of Illinois at Chicago

Degree Level


Degree name

PhD, Doctor of Philosophy

Committee Member

Lim, Jennifer I Klatt, Dieter Royston, Thomas J Cao, Dingcai

Submitted date

August 2020

Thesis type