Robustness and real-time adaptation of EMD-based spindle segmentation

Assess the sensitivity of the Empirical Mode Decomposition plus LOESS spindle segmentation algorithm to changes in noise level, sampling rate, and the presence of other oscillatory components, and develop or validate adaptations that enable real-time detection and operation in closed-loop control systems.

Background

The authors present a spindle segmentation procedure using Empirical Mode Decomposition and thresholded envelope extrema, and test robustness against additive Brownian noise in controlled simulations.

They explicitly pose open questions regarding broader robustness to acquisition and signal conditions and whether the method can be adapted for real-time use or closed-loop applications.

References

Despite the progress offered by our model, several computational and segmentation questions remain open: Detection robustness: How sensitive is the EMD-based segmentation algorithm to changes in noise level, sampling rate, and the presence of other oscillatory components? Can the algorithm be adapted for real-time detection or used in closed-loop control systems?

— Modeling, Segmenting and Statistics of Transient Spindles via Two-Dimensional Ornstein-Uhlenbeck Dynamics (2512.10844 - Sun et al., 11 Dec 2025) in Section 7, Discussion and Open Problems