A Robust Gaze Estimation Approach via Exploring Relevant Electrooculogram Features and Optimal Electrodes Placements.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Human-Computer Interaction

Background:

• Gaze estimation aids in disability assistance and diagnosing conditions like autism spectrum disorder (ASD), Parkinson's disease, and attention deficit hyperactivity disorder (ADHD).
• Electrooculography (EOG) provides an economical and effective method for gaze estimation, suitable for practical applications.
• Optimizing electrode placement and feature extraction is crucial for enhancing EOG-based gaze estimation accuracy.

Purpose of the Study:

• To systematically investigate optimal EOG electrode locations and informative features for gaze estimation.
• To select the best channels and features using a forward stepwise strategy to eliminate irrelevant information.
• To evaluate the impact of electrode placement and feature contributions on gaze estimation performance using classic models.

Main Methods:

• Systematic investigation of EOG electrode locations around the orbital cavity.
• Extraction of temporal-spectral domain features from seven differential channels.
• Application of a forward stepwise search algorithm for optimal channel and feature selection.
• Comparative analysis using six classic models and 18 subjects.

Main Results:

• Achieved promising performance in Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) for gaze estimation ranging from -50° to +50°.
• Optimized approach reached MAE of 2.80° and RMSE of 3.74° using only 10 features from 2 channels.
• Demonstrated significant performance improvements compared to prevailing EOG-based techniques, with MAE and RMSE reductions ranging from 0.70° to 5.48° and 0.66° to 5.42°, respectively.

Conclusions:

• Proposed a robust EOG-based gaze estimation approach through systematic optimization of channel and feature combinations.
• Experimental results highlight the superiority of the proposed method and its potential for clinical applications.
• The accurate gaze estimation via EOG signals holds promise for assisting individuals with disabilities and aiding in the diagnosis of neurological and developmental disorders.