Type 2 diabetes is a metabolic disorder that may cause major problems to several physiological systems. Exercise has proven to be very effective in the prevention, management and improvement of this pathology in patients. Muscle metabolism is often studied with near-infrared spectroscopy (NIRS), a noninvasive technique that can measure changes in the concentration of oxygenated (O2Hb) and reduced hemoglobin (HHb) of tissues. These NIRS signals are highly non-stationary, non-Gaussian and nonlinear in nature. The empirical mode decomposition (EMD) is used as a nonlinear adaptive model to extract information present in the NIRS signals. NIRS signals acquired from the tibialis anterior muscle of controls and type 2 diabetic patients are processed by EMD to yield three intrinsic mode functions (IMF). The sample entropy (SE), fractal dimension (FD), and Hurst exponent (HE) are computed from these IMFs. Subjects are monitored at the beginning of the study and after one year of a physical training programme. Following the exercise programme, we observed an increase in the SE and FD and a decrease in the HE in all diabetic subjects. Our results show the influence of physical exercise program in improving muscle performance and muscle drive by the central nervous system in the patients. A multivariate analysis of variance performed at the end of the training programme also indicated that the NIRS metabolic patterns of controls and diabetic subjects are more similar than at the beginning of the study. Hence, the proposed EMD technique applied to NIRS signals may be very useful to gain a non-invasive understanding of the neuromuscular and vascular impairment in diabetic subjects.
Keywords: Empirical mode decomposition; Higuchi fractal dimension; Hurst exponent; MANOVA; Muscle metabolism; Near-infrared spectroscopy; Sample entropy; Type 2 diabetes.
Copyright © 2015 Elsevier Ltd. All rights reserved.