Multiplicative multi-fractal modeling of electromyography signals for discerning neuropathic conditions

On Multiscaling of Parkinsonian Rest Tremor Signals and Their Classification

Self-similar stochastic processes and broad probability distributions are ubiquitous in nature and in many man-made systems. The brain is a particularly interesting example of (natural) complex system where those features play a pivotal role. In fact, the controversial yet experimentally validated "criticality hypothesis" explaining the functioning of the brain implies the presence of scaling laws for correlations. Recently, we have analyzed a collection of rest tremor velocity signals recorded from patients affected by Parkinson's disease, with the aim of determining and hence exploiting the presence of scaling laws. Our results show that multiple scaling laws are required in order to describe the dynamics of such signals, stressing the complexity of the underlying generating mechanism. We successively extracted numeric features by using the multifractal detrended fluctuation analysis procedure. We found that such features can be effective for discriminating classes of signals recorded in different experimental conditions. Notably, we show that the use of medication (L-DOPA) can be recognized with high accuracy.

Exploring Acute Changes in Hamstring EMG after Warm-up and Stretching Using a Multifractal Analysis

INTRODUCTION

This study aimed to apply multifractal detrended fluctuation analysis (MFDFA) to surface EMG to detect neuromuscular changes after realistic warm-up procedures that was followed by various stretching exercises.

METHODS

Sixteen volunteers conducted two experimental sessions. Testing included two maximal voluntary contractions before, after a standardized warm-up, and after a stretching exercise (static or neurodynamic nerve gliding technique). EMG was registered on biceps femoris and semitendinosus muscles. EMG was analyzed using different parameters obtained from the singularity Hurst exponent function and multifractal power spectrum (both obtained from the multifractal detrended fluctuation analysis).

RESULTS

The Hurst exponent, α maximum, and peak value of the multifractal spectrum significantly decreased after warm-up as compared with baseline for both biceps femoris ( P = 0.003, P = 0.006, and P = 0.003, respectively) and semitendinosus ( P = 0.006, P = 0.013 and P = 0.01, respectively) muscles. No further alteration was obtained after static or neurodynamic nerve gliding stretching as compared with post-warm-up ( P = 1.0). No significant difference was obtained for Hurst exponent range, width, and asymmetry of the multifractal spectrum ( P > 0.05).

CONCLUSIONS

From the present results, EMG depicted multifractal features sensitive to detect neuromuscular changes after a warm-up procedure. An increase in multiscale complexity is revealed after warm-up without any further alteration after stretching. The multifractal spectrum depicted dominant small fluctuations that shifted toward slightly larger fluctuations that could be attributed to motor unit recruitment.

Analysis of concentric and eccentric contractions in biceps brachii muscles using surface electromyography signals and multifractal analysis

Muscle contractions can be categorized into isometric, isotonic (concentric and eccentric) and isokinetic contractions. The eccentric contractions are very effective for promoting muscle hypertrophy and produce larger forces when compared to the concentric or isometric contractions. Surface electromyography signals are widely used for analyzing muscle activities. These signals are nonstationary, nonlinear and exhibit self-similar multifractal behavior. The research on surface electromyography signals using multifractal analysis is not well established for concentric and eccentric contractions. In this study, an attempt has been made to analyze the concentric and eccentric contractions associated with biceps brachii muscles using surface electromyography signals and multifractal detrended moving average algorithm. Surface electromyography signals were recorded from 20 healthy individuals while performing a single curl exercise. The preprocessed signals were divided into concentric and eccentric cycles and in turn divided into phases based on range of motion: lower (0°–90°) and upper (>90°). The segments of surface electromyography signal were subjected to multifractal detrended moving average algorithm, and multifractal features such as strength of multifractality, peak exponent value, maximum exponent and exponent index were extracted in addition to conventional linear features such as root mean square and median frequency. The results show that surface electromyography signals exhibit multifractal behavior in both concentric and eccentric cycles. The mean strength of multifractality increased by 15% in eccentric contraction compared to concentric contraction. The lowest and highest exponent index values are observed in the upper concentric and lower eccentric contractions, respectively. The multifractal features are observed to be helpful in differentiating surface electromyography signals along the range of motion as compared to root mean square and median frequency. It appears that these multifractal features extracted from the concentric and eccentric contractions can be useful in the assessment of surface electromyography signals in sports medicine and training and also in rehabilitation programs.

Discrimination and Characterization of Parkinsonian Rest Tremors by Analyzing Long-Term Correlations and Multifractal Signatures

GOAL

We analyze 48 signals of rest tremor velocity related to 12 distinct subjects affected by the Parkinson's disease. The subjects belong to two different groups, high- and low-amplitude rest tremors, with four and eight subjects, respectively. Each subject has been tested in four settings given by combining the use of deep brain stimulation and L-DOPA medication.

METHODS

We develop two main feature-based representations of the signals, which are obtained by considering 1) the long-term correlations and multifractal properties, and 2) the power spectra.

RESULTS

Our results show that, when medication is used, a qualitative change is observed in the related signals from anticorrelated to long term positively correlated. In addition, the medication effect yields statistically significant differences in both high- and low-amplitude tremor groups. We successively consider three different classification problems, involving the recognition of 1) the use of medication, 2) the use of deep brain stimulation, and 3) the membership to the high- and low-amplitude tremor groups. Classification results show that the best results are obtained with a parsimonious, two-dimensional (2-D) representation encoding the long-term correlations and multifractal properties of the signals.

CONCLUSIONS

Long-term correlations and multifractal signatures of time series provided an effective tool to analyze Parkinsonian rest tremor signals.

SIGNIFICANCE

The developed 2-D representation is a parsimonious and effective representation for rest tremor signals that could be adopted in clinical settings, even by considering resource-constrained scenarios.