Feature Stability and Setup Minimization for EEG-EMG-Enabled Monitoring Systems

Delivering healthcare at home emerged as a key advancement to reduce healthcare costs and infection risks, as during the SARS-Cov2 pandemic. In particular, in motor training applications, wearable and portable devices can be employed for movements recognition and monitoring of the associated brain signals. In this context, it is essential to minimize the monitoring setup and the amount of data to collect, process, and share. In this paper, we address this challenge for a monitoring system that includes high-dimensional EEG and EMG data for hand movements classification. We fuse EEG and EMG into the magnitude squared coherence (MSC) signal, from which we extracted features using different algorithms (one from the authors) to solve binary classification problems. Finally, we propose a mapping-and-aggregating strategy to increase the interpretability of the machine learning results. The proposed approach provides very low mis-classification errors ( < 0. 1 ), with very few and stable MSC features ( < 10% of the initial set of available features). Furthermore, we identified a common pattern across algorithms and classification problems, i.e., the activation of the centro-parietal brain areas and arm ’s muscles in 8 ÷ 80 Hz, in line with previous literature. Thus, this study represents a step forward to the minimization of a reliable EEG-EMG setup to enable precise motor training at home.

Analysis of EEG-EMG Coherence in Low Frequency Bands

In this, study, an attempt is made to analyze the corticomuscular coupling of the brain and muscular system in the low-frequency components during ramp descent (RD) and stair descent (SD) locomotion. For this purpose, magnitude squared coherence (MSC) is computed from the simultaneous EEG and EMG signals recorded during the ramp and stair descent tasks. The MSC is extracted from the low- frequency bands such as delta (0.1–3 Hz) and theta bands (4–7 Hz). The study utilizes a publicly available database consisting of simultaneous recorded EEG, lower limb EMG and full body motion information from ten healthy subjects. The results show that there exists corticomuscular coupling between motor cortex (C1, C2 and Cz contacts) and tibialis anterior muscle activities during RD and SD. In addition, the MSC differs for both the tasks and frequency bands. In delta band frequencies, the MSC is found to be higher in C2 regions. In the case of theta, the MSC is higher in C1 during RD and in Cz during SD. Therefore, the MSC associated with the low frequency components could be used to detect walking intentions.

The Implementation of Fast Fourier Transform and Coherence Function to Detect the Millimetric Hole on Strip Iron Plate

In this study, Fast Fourier Transform (FFT) was used in order to detect bore hole in a structure. FFT is a common method in digital signal processing (DSP) to characterize the frequency emitted by some structure. This method is widely used because of its simplicity. Computational time needed for FFT is relatively lower than another method. The use of FFT to analyze defect in structure is not commonly used since FFT has some weakness for example spatial frequency cannot be extracted to point out the defect location. In this paper, defect was designated as a hole in a strip iron plate with 20 mm in diameter. The strip iron plate was 1 meter long, 38 mm wide and 3 mm thick. This strip iron plate was clamped at one of its ends while the other side is left free. In order to produce vibration signal, impact hammer Bruel Kjaer Type 8202 was used with plastic tip to limit the vibration frequency in to the range of 0 - 1000 Hz. The trigger point was 30 mm from its free end. Three accelerometers were placed series in one line with the trigger point with 300 mm distance of each accelerometer. The position of the hole was varied in three different position. The first position was between trigger point and first accelerometer, between first and second accelerometer and between the second and third accelerometer. The raw signal obtained from the accelerometer was processed by using FFT to understand the mode shape changes in the strip iron plate due to the bore hole. Furthermore, the FFT result was analyzed as function of receiver position to determine the position of hole. The result shows that the frequency characters were different in each case and further analysis by using magnitude-squared coherence function need to be used in order to quantitatively find the difference between FFT result.