Abstract
Fluctuation-based dispersion entropy (FDispEn) is a new approach to estimate the dynamical variability of the fluctuations of signals. It is based on Shannon entropy and fluctuation-based dispersion patterns. To quantify the physiological dynamics over multiple time scales, multiscale FDispEn (MFDE) is developed in this paper. MFDE is robust to the presence of baseline wanders or trends in the data. We evaluate MFDE, compared with popular multiscale sample entropy (MSE), multiscale fuzzy entropy (MFE), and the recently introduced multiscale dispersion entropy (MDE), on selected synthetic data and five neurological diseases' datasets. The MFDE avoids the problem of the undefined MSE values and, compared with the MFE and MSE, leads to more stable entropy values over the scale factors for white and pink noises. Overall, the MFDE is the fastest and most consistent method for the discrimination of different states of neurological data.
Introduction: The Challenge of Measuring Biological Complexity
In biomedical signal processing, "complexity" is a hallmark of physiological health. A healthy biological system, such as a heart rate or a walking gait, exhibits rich, multi-layered variability that allows for adaptation to internal and external stressors. Conversely, the onset of disease often leads to a "decomplexification" of these signals.
To quantify this complexity, researchers frequently employ entropy-based metrics. However, traditional tools like Multiscale Sample Entropy (MSE) and Multiscale Fuzzy Entropy (MFE) face significant hurdles:
- Computational Inefficiency: These methods often have $O(N^2)$ complexity, making them prohibitively slow for long-term monitoring or large datasets.
- Data Sensitivity: They frequently yield "undefined" results when applied to short or noisy data segments.
- Baseline Wander: Real-world signals often contain trends or "drifts" (e.g., eye movements or EEG baseline shifts). Traditional measures often mistake these simple trends for structural complexity, leading to inaccurate assessments.
Methodology: Multiscale Fluctuation-Based Dispersion Entropy (MFDE)
The proposed Multiscale Fluctuation-Based Dispersion Entropy (MFDE) addresses these limitations by focusing on signal fluctuations rather than absolute values.
The FDispEn Algorithm
The core of the approach is Fluctuation-Based Dispersion Entropy (FDispEn). Unlike standard dispersion entropy, which categorizes absolute signal amplitudes, FDispEn analyzes the differences between adjacent points. This inherent focus on change makes the algorithm robust to baseline wander and slow-moving trends.
The process involves mapping signal values into discrete classes using a Normal Cumulative Distribution Function (NCDF), identifying "dispersion patterns" based on fluctuations, and calculating Shannon entropy from the resulting probability distribution.
Multiscale Analysis
To capture dynamics across different time horizons, MFDE employs a "coarse-graining" procedure. The original signal is averaged over increasing window sizes (scales). This allows the metric to distinguish between signals that are merely irregular (like white noise) and those that possess true structural complexity across scales (like pink noise).
Evaluation on Neurological Datasets
The researchers validated MFDE across a diverse range of synthetic and clinical datasets, including epilepsy (EEG), Alzheimer's disease (EEG), and various movement disorders.
Gait Analysis in Huntington's Disease and ALS
One critical application was the analysis of walking stride intervals. In diseases like Huntington's Disease (HD), the rhythmic nature of walking is disrupted by involuntary movements and neurological decay.
MFDE successfully quantified the altered complexity in HD and ALS patients compared to healthy controls. While traditional MSE often failed to provide values for shorter gait recordings, MFDE remained stable and discriminative.
Eye Movement and Discrimination Power
In eye-movement data—where signals are notoriously prone to shifting means as the subject's gaze moves—MFDE was the only method capable of significantly discriminating between Ataxia and Parkinson’s Disease across all analyzed time scales.
Key Results and Advantages
- Speed: MFDE operates with $O(N)$ linear complexity. In tests with 100,000 samples, MFDE took approximately 0.4 seconds, compared to over 300 seconds for MFE.
- Robustness: The method is uniquely resistant to noise and baseline drift, which are common in clinical environments.
- Consistency: MFDE avoids the "undefined" results that plague MSE, providing a reliable metric even for short-duration recordings.
Conclusion and Future Directions
MFDE represents a significant advancement in the robust and rapid assessment of biological signal complexity. Its ability to handle non-stationary data and its computational efficiency make it a prime candidate for integration into real-time diagnostic tools and wearable health monitors. Future work will explore more advanced coarse-graining techniques and multivariate extensions to analyze data from multiple sensors simultaneously.
Summary updated