GlobeMetrics: A Healthcare Framework for Video Based Saccade Characterization

Neog, DR and Salman Choudhary, P and Pathak, M (2024) GlobeMetrics: A Healthcare Framework for Video Based Saccade Characterization. In: 2024 IEEE Conference on Artificial Intelligence, June 25-June 27, 2024, Singapore, pp. 638-643.

PDF con_art_eng_2024.pdf - Published Version Restricted to Registered users only Download (1MB)

Abstract

Eye movement analysis is extensively utilized in understanding mechanisms governing perception, cognition, and action and proves valuable in exploring neurological and neurodegenerative diseases. This paper introduces GlobeMetrics, a healthcare application designed for ocular analysis from video data. Our emphasis in this study is on estimating the saccadic profile of the subjects utilizing the GlobeMetrics framework. The framework includes a setup for data recording, an appearance-based gaze estimation system, a module for analyzing gaze data specifically for saccade analysis, and an interactive GUI application that interfaces with each aspect of the framework. The proposed gaze estimation network consists of a convolutional neural network (CNN) based segmentation and regression network that maps input frames to gaze points. The proposed gaze estimation architecture achieves a prediction error of 0.467± 0.133 cm on our database. Additionally, the segmentation network attains mean IOUs of 95.19 and 97.39 for sclera and iris, respectively. Our proposed framework, GlobeMetrics, offers an interactive platform for conducting ocular analysis in clinical settings. This application seamlessly integrates data recording, stimulus generation, database management, and data analysis within a unified framework. The overall framework is accurate, robust, and generalizes well to new subjects. © 2024 IEEE.

Item Type:	Conference Paper
Publication:	Proceedings - 2024 IEEE Conference on Artificial Intelligence, CAI 2024
Publisher:	Institute of Electrical and Electronics Engineers Inc.
Additional Information:	The copyright for this article belongs to Institute of Electrical and Electronics Engineers Inc.
Keywords:	Convolution; Convolutional neural networks; Health care; Information management; Neurodegenerative diseases, Appearance based; Convolutional neural network; Estimation systems; Eye movement analysis; Gaze estimation; GUI applications; Health care application; Network-based; Saccade analyse; Video data, Eye movements
Department/Centre:	Division of Electrical Sciences > Electrical Communication Engineering
Date Deposited:	06 Sep 2024 05:22
Last Modified:	06 Sep 2024 05:22
URI:	http://eprints.iisc.ac.in/id/eprint/85980

Actions (login required)

View Item