The Effectiveness Assessment of Massage Therapy Using Entropy-Based EEG Features Among Lumbar Disc Herniation Patients Comparing With Healthy Controls

Massage therapy (MT) is a useful complementary and alternative therapy widely used in treating low back pain (LBP), including lumbar disc herniation (LDH). However, few studies revealed the quantitative entropy-based features of electroencephalography (EEG) for the MT effectiveness for the LDH patients. This study investigated the immediate effects of Chinese massage on four EEG rhythms, using the eight entropy-based features (approximation entropy (ApEn), Sample Entropy (SampEn), wavelet entropy (WaveEn), Hilbert-Huang Transform Marginal spectrum entropy (HHTMSEn), normalized energy, permutation entropy (PE), Fuzzy entropy (FuzzyEn), and Inherent fuzzy entropy (IFE)) in the 26 LDH patients and 24 healthy controls. Results showed that after MT in LDH group, ApEn, SampEn, WaveEn, PE, FuzzyEn, and IFE features of the delta rhythm decreased, the normalized energy of delta rhythm significantly increased in the left hemisphere and theta/alpha rhythms significantly decreased. Furthermore, HHTMSEn feature of theta and beta rhythms showed the significant difference for two groups (LDH and control group) in two states (before and after massage). Ten classifiers were applied in classifying two groups or two states in one group, and most of them reached high averaged accuracies and the area under the curve (AUC), even with different length per epoch (1024, 512, and 256 points). It indicated that the entropy-based features and permutation disalignment index (PDI) of EEG rhythms could be promising indices of the massage effectiveness for LDH patients and control group. Rhythms' SampEn scalp topography in two groups showed the significant different complexity change between two states (before and after massage).