Biography
A. Q. M. Sala Uddin Pathan is a researcher and academic currently pursuing a PhD at the University of New South Wales (UNSW), focusing on ocular image analysis using machine learning. He also works as a casual academic at UNSW. Before relocating to Australia, Sala Uddin taught for five years in the Department of Computer Science and Telecommunication Engineering at Noakhali Science and Technology University in Bangladesh.
He is actively engaged in student affairs and serves as the...
Biography
A. Q. M. Sala Uddin Pathan is a researcher and academic currently pursuing a PhD at the University of New South Wales (UNSW), focusing on ocular image analysis using machine learning. He also works as a casual academic at UNSW. Before relocating to Australia, Sala Uddin taught for five years in the Department of Computer Science and Telecommunication Engineering at Noakhali Science and Technology University in Bangladesh.
He is actively engaged in student affairs and serves as the General Secretary of the UNSW Postgraduate Council (PGC). Sala Uddin has a passion for cricket and regularly plays for the Kryptonites Sporting Club, where he holds a leadership role. Additionally, he has certifications in cardiopulmonary resuscitation (CPR) and mental health first aid, demonstrating his commitment to professional development and community service.
Research title: Machine learning for ocular image analysis: automated segmentation and classification
Supervisor: A/Prof. Maitreyee Roy
Co-supervisors: Prof. Salil Kanhere & Prof. Matthew Simunovic
Research abstract: Machine learning (ML) has revolutionized ocular image analysis, offering automated solutions for disease diagnosis and segmentation. This study aims to develop an improved ML-based system for the precise and efficient diagnosis of ocular diseases using optical coherence tomography (OCT) and retinal fundus images. Traditional diagnostic methods are time-consuming, prone to errors, and dependent on expert interpretation. Automated segmentation and classification techniques can aid ophthalmologists in diagnosing conditions such as Epiretinal Membrane and Vitreomacular Adhesion with enhanced accuracy.
This research focuses on identifying optimal ML models for segmenting the retinal layers and correcting topology errors in segmented masks to enhance clinical decision-making. The methodology includes data preprocessing, boundary detection, and post-processing to ensure topologically accurate segmentation. By integrating deep learning techniques, this study aims to improve the robustness and reliability of automated ocular image analysis.
The expected outcome is an advanced ML-driven system that efficiently segments and classifies ocular images, assisting ophthalmologists in making precise clinical decisions. This work will contribute to reducing diagnostic time, improving disease detection, and addressing the increasing demand for automated ophthalmic analysis.
