Design and Implementation of Wearable Devices for sEMG Detection and Analysis of Muscle Synergies

This thesis is part of a larger project, called MITOR (a collaboration between Politecnico di Torino (POLITO), Massachusetts Institute of Technology (MIT) and Spaulding Rehabilitation Hospital’s Motion Analysis Laboratory (an affiliate of Harvard Medical School)). This project aims to develop a wearable, wireless, low-power surface ElectroMyoGraphy (sEMG) acquisition system for long-term monitoring of muscle synergies and diagnostic motor rehabilitation. The system’s hardware specification was defined specifically for sEMG acquisition channels. Where the resulting signals are stored locally on a microSD card. The circuit diagrams were drawn referencing the schematics of the previous version of the MITOR system. In this revision, the most significant changes from the previous MITOR is an upgraded microcontroller and additional components in the acquisition channel (so the board can detect Threshold Crossing (TC) signals). The TC signal is a quasi-digital signal obtained by comparing sEMG signal and a threshold. Each time the sEMG signal exceeds this threshold an event is generated. The number of events is proportional to the electrical muscle activity. From this signal, it is possible to calculate the Average Threshold Crossing (ATC) by noting the number of TC events into a fixed window of time. Applying this approach to sEMG is innovative as it involves a considerable reduction in power consumption due to drastically reducing the volume of data in need of processing, saving and transmission. Using Altium Designer, a two-layer Printed Circuit Board (PCB) was created. In this phase, all the electronic components were placed in order to minimize board size by optimizing its dimensions (50mm x 37.5mm). Most of the components were soldered manually or with the use of a pick-and-place machine. Once the components were set, the electrical connections were checked. A case for the PCB was created to ease the process of placing the devices on human subjects for sEMG signal recording. The microcontroller firmware was written in C++, using the IAR Embedded Workbench IDE. The firmware allows three different working modes: classic sEMG signal, sEMG envelope (using digital filters in the firmware), and ATC signal. The main feature of the firmware is wireless synchronization between the different MITOR boards (possible due to a PCB antenna). This is the crucial element that needed to be tested to guarantee long-term analysis. The results showed a maximum delay of 10 μs with a synchronization period of 5s. Another important aspect to analyze is power consumption. Each developed board, with a battery supply of 450mA/h, can store data for more than 42 hours in classic sEMG mode; or for more than 80 hours in ATC mode. A commercially available multi-channel EMG system from Motion Lab Systems was used to validate the sEMG signal recorded by the MITOR system. For the tests, muscles of the upper-limbs performing grasping and of the lower-limbs during gait were selected. The developed boards, in combination with the Motion Lab Systems’ probes, were used to extract muscle synergies during gait in a healthy subject. In addition to the classic method, the ATC approach was also tested to extract the synergies. The synergies with ATC signal, compared with the synergies of the classic method, showed good results. The performance was evaluated in terms of zero lag cross-correlation between the time activations and of cosine similarity between the muscle weights.