Author Type

Graduate Student

Date of Award

Spring 4-28-2026

Document Type

Thesis

Publication Status

Version of Record

Submission Date

May 2026

Department

Computer and Electrical Engineering and Computer Science

College Granting Degree

College of Engineering and Computer Science

Department Granting Degree

Electrical Engineering and Computer Science

Degree Name

Master of Science (MS)

Thesis/Dissertation Advisor [Chair]

Abhijit Pandya

Abstract

This thesis presents the design, implementation, and experimental validation of an artificial intelligence (AI)-driven system for detecting and quantifying nystagmus an involuntary, rhythmic oscillation of the eyes intended as a portable, low-cost complement to conventional Videonystagmography (VNG). The complete pipeline integrates six algorithmic stages: face landmark detection, contrast enhancement, background-aware pixel thresholding, grid-based vertical column filtering, connected-component cluster analysis, and centroid computation, operating in real time on standard smartphone video to extract a sub-pixel normalized iris position time-series without any specialized eye-tracking hardware or infrared illumination. The system supports diagnostic decision-making, highlighting its promise for incorporation into telemedicine settings. The beat-detection and slow phase velocity (SPV) algorithms were validated across eight video recordings obtained from three subjects under all four canonical optokinetic stimulation directions. Peak SPV values ranged from 18.57 to 22.23 deg/sec for horizontal recordings and from 15.40 to 20.57 deg/sec for vertical recordings, all within the clinically accepted 30°/sec ceiling and consistent with the expected physiological range of optokinetic nystagmus in healthy subjects.

Download

Share

COinS