Freezing of gait (FOG) is a paroxysmal dyskinesia, which is common in patients with advanced Parkinson’s disease (PD). It is an important cause of falls in PD patients and is associated with serious disability. In this study, we implemented a novel FOG detection system using deep learning technology. The system takes multi-channel acceleration signals as input, uses one-dimensional deep convolutional neural network to automatically learn feature representations, and uses recurrent neural network to model the temporal dependencies between feature activations. In order to improve the detection performance, we introduced squeeze-and-excitation blocks and attention mechanism into the system, and used data augmentation to eliminate the impact of imbalanced datasets on model training. Experimental results show that, compared with the previous best results, the sensitivity and specificity obtained in 10-fold cross-validation evaluation were increased by 0.017 and 0.045, respectively, and the equal error rate obtained in leave-one-subject-out cross-validation evaluation was decreased by 1.9%. The time for detection of a 256 data segment is only 0.52 ms. These results indicate that the proposed system has high operating efficiency and excellent detection performance, and is expected to be applied to FOG detection to improve the automation of Parkinson’s disease diagnosis and treatment.

Power line interference (PLI) is a major source of noise in sEMG signals. As the bandwidth of PLI overlaps with the sEMG signals, it can easily affect the interpretation of the signal. The processing methods used in the literature are mostly notch filtering and spectral interpolation. However, it is difficult for the former to reconcile the contradiction between completely filtering and avoiding signal distortion, while the latter performs poorly in the case of a time-varying PLI. To solve these, a novel synchrosqueezed-wavelet-transform (SWT)-based PLI filter is proposed. The local SWT was developed to reduce the computation cost while maintaining the frequency resolution. A ridge location method based on an adaptive threshold is presented. In addition, two ridge extraction methods (REMs) are proposed to fit different application requirements. Parameters were optimized before further study. Notch filtering, spectral interpolation, and the proposed filter were evaluated on the simulated signals and real signals. The output signal-to-noise ratio (SNR) ranges of the proposed filter with two different REMs are 18.53–24.57 and 18.57–26.92. Both the quantitative index and the time–frequency spectrum diagram show that the performance of the proposed filter is significantly better than that of the other filters.