Real time sEMG based muscle fatigue monitoring using low power integrated circuits

• Main Author: Koutsos, Ermis
• Other Authors: Georgiou, Pantelis
• Published: Imperial College London 2016
• Subjects: 621.3
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.739600

Electromyogram (EMG), the recording of the electrical impulses of the muscles, is a rich source of information, which can facilitate such an insight into our muscles and especially their activation and fatigue level. Muscle fatigue has been shown to be one of the most important biofeedback parameters of EMG in rehabilitation, ergonomics and training, by using measured results from the body to change the way we behave, improve our performance and achieve better compliance to rehabilitation. This thesis addresses the challenge of reliably and efficiently estimating a muscle’s fatigue state though monitoring surface EMG signals, with the use of low power integrated circuits. CMOS technology facilitates localised real-time processing to achieve complete miniaturisation, resulting in an information driven system rather that conventionally data driven system. Thus, reducing requirements on data transmission, saving power and increasing the degree of freedom for the user. Several EMG properties progressively change during muscle fatigue and can be quantified in the time and frequency domains using different processing techniques. CMOS technology allows to significantly reduce the power and size of the developed EMG processing technology. Firstly, a CMOS system is presented, capable of measuring the instantaneous Median Frequency (iMDF) of the EMG signal, which is considered the golden standard for muscle fatigue assessment. Continuing, a novel bit-stream cross-correlator design that greatly simplifies the sEMG signal without any loss of information is presented for the estimation of the EMG conduction velocity, which is associated with the physiological changes of the muscle during fatigue. Furthermore, a new metric similar to iMDF is introduced, combining the advantages of the bit-stream approach that can accurately track the spectral compression of the sEMG during fatigue with one bit representation. Lastly, a complete muscle fatigue monitoring System-on-Chip (SoC) is presented, offering complete insight into the underlying mechanisms and physiological changes during muscle fatigue through sEMG analysis while operating under both static and dynamic contractions. The proposed approach is scalable, as several muscle fatigue monitoring SoCs can operate in parallel and periodically relay key information about the muscle, thus reducing data transmission costs and bandwidth requirements. Finally, a succession of wearable EMG devices are presented, introducing the use of custom Application Specific Integrated Circuits in wearable electronics for unsupervised muscle fatigue monitoring. The wearable nodes are wireless while user ergonomics, power and weight were the primary design considerations.