NPD

NPD detects and analyzes microsaccades by applying a statistical parametrization and maximum likelihood estimation to gaze-position time series for precise saccadic event detection and parameter estimation.

Key Features:

• Statistical Model Integration: Incorporates a parametrization previously used to model saccades to capture the saccadic velocity profile.
• Maximum Likelihood Estimation (MLE): Implements MLE to simultaneously detect saccadic events and estimate saccade duration, amplitude, and maximum velocity.
• Adaptive and Statistically Optimal Detection: Uses an adaptive approach and statistical optimality to improve microsaccade detection performance.
• Comparative Performance: Demonstrates superior or comparable performance to convolutional neural network and Bayesian methods while modeling a biologically accurate velocity profile.

Scientific Applications:

• Perception and Cognition Research: Provides precise saccadic parameter estimates to mitigate biases in studies of microsaccades' roles in perception and cognition.
• Human-Computer Interaction (HCI): Enables online saccade detection for real-time eye-movement analysis in HCI applications.

Methodology:

Processes two input CSV files, X.csv and Y.csv, containing X and Y gaze-coordinate matrices of identical square dimensions; performs detection independently on each row of these matrices using the saccade parametrization and maximum likelihood estimation to detect events and estimate saccadic parameters.