Skip to main content

Artificial Intelligence (AI) enabled Nystagmus Detection and classification for Diagnosis of Positional Vertigo Syndromes

Project Member(s): Prasad, M., Braytee, A.

Funding or Partner Organisation: The Balance Clinic and Laboratory (The Balance Clinic & Laboratory)
The Balance Clinic and Laboratory (The Balance Clinic & Laboratory)

Start year: 2023

Summary: This project proposes to automate both eye-tracking and video-based nystagmus classification using hybrid deep learning architectures. Our proposed hybrid deep learning model will be combined with iris segmentation and frame sampling for efficient diagnosis of positional vertigo syndromes. This thesis project will also focus on multimodal representation learning for additional supervisory signals from the multimodal data. The objectives of this projects are as follows: 1. To develop an accurate eye tracking programme to detect horizontal, torsional, and vertical eye-movements capable of identifying benign positional vertigo. 2. Implementing AI-enabled Computer-aided diagnosis (CAD) solutions to develop a diagnostic classifier to separate the 6 subtypes of BPV and test it against gold standard clinical diagnoses of real life experts for diagnostic accuracy. 3. Applying multimodal representation learning to process and link information of various modalities including nystagmus traces, history and video oculography (VOG) data for better diagnosis of patients with positional vertigo syndromes. 4. Develop, validate and implement an AI augmented smartphone tool to deliver an expert diagnosis to healthcare workers in remote locations.

FOR Codes: Artificial intelligence