Shifts in Spectral Power Detected By Fourier And Wavelet Transforms During Muscular Contractions To Volitional Exhaustion

• Published in: Clinical kinesiology (2003) 2015-07, Vol.69 (2), p.5A-5A
• Main Authors: Craft, Amber, Croce, Ronald, Miller, John, Chamberlin, Kent, Filipovic, David
• Format: Article
• Language: English
• Subjects: Algorithms; Fatigue; Methods; Researchers; Signal processing; Studies; Wavelet transforms
• ISSN: 0896-9620

Surface electromyography (SEMG) time-frequency responses during fatiguing muscular contractions to volitional exhaustion computed by traditional short-time Fourier transform (STFT) and continuous wavelet transform (CWT) were compared using ANOVA and polynomial regression analyses. Twelve healthy adult male university students (mean ± SD age = 21.08 ± 1.25 years; stature = 179.27 ± 4.93 cm; mass = 88.57 ± 7.15 kg) performed maximum effort flexion-extension movements at 3.14 rad . s^sup -1^ until volitional exhaustion, rested for 2-min, and then completed a second bout of movements until volitional exhaustion. Average torque and median and mean frequencies of rectus femoris, vastus lateralis, and vastus medialis were examined at nine points across repetitions completed. Average torque decreased precipitously during early stages of repetitions completed, then decreased less sharply to exhaustion. Median and mean frequencies displayed a steep decrease during early stages of repetitions completed, then decreased less sharply to exercise cessation. There was a significant measurement technique effect for mean frequency (STFT higher values than CWT), and for both mean and median frequencies there were greater score standard deviations using STFT. For exertions used in this investigation, fractional polynomial and quadratic models explained fatigue with the highest precision. Despite concerns over non-stationarity of the electromyography waveform, the Fourier-based method was acceptable as a way to measure central frequency during fatiguing muscular contractions at moderate velocities. However, Fourier analysis can lead to measurement error, which can directly impact on standard deviations and data variance.