Beginning with the introduction of electrically powered prostheses more than 65 years ago surface electromyographic (sEMG) signals recorded from residual muscles in amputated limbs have served as the primary source of upper-limb myoelectric prosthetic control. The majority of these devices use one or more neural interfaces to translate the sEMG signal amplitude into voltage control signals that drive the mechanical components of a prosthesis. In so doing, users are able to directly control the speed and direction of prosthetic actuation by varying the level of muscle activation and the associated sEMG signal amplitude. Consequently, in spite of decades of development, myoelectric prostheses are prone to highly variable functional control, leading to a relatively high-incidence of prosthetic abandonment among 23-35% of upper-limb amputees. Efforts to improve prosthetic control in recent years have led to the development and commercialization of neural interfaces that employ pattern recognition of sEMG signals recorded from multiple locations on a residual limb to map different intended movements. But while these advanced algorithms have made strident gains, there still exists substantial need for further improvement to increase the reliability of pattern recognition control solutions amongst the variability of muscle co-activation intensities. In efforts to enrich the control signals that form the basis for myoelectric control, I have been developing advanced algorithms as part of a next generation neural interface research and development, referred to as Motor Unit Drive (MU Drive), that is able to non-invasively extract the firings of individual motor units (MUs) from sEMG signals in real-time and translate the firings into smooth biomechanically informed control signals. These measurements of motor unit firing rates and recruitment naturally provide high-levels of motor control information from the peripheral nervous system for intact limbs and therefore hold the greater promise for restoring function for amputees. The goal for my doctoral work was to develop advanced algorithms for the MU Drive neural interface system, that leverage MU features to provide intuitive control of multiple degrees-of-freedom. To achieve this goal, I targeted 3 research aims: 1) Derive real-time MU-based control signals from motor unit firings, 2) Evaluate feasibility of motor unit action potential (MUAP) based discrimination of muscle intent 3) Design and evaluate MUAP-based motion Classification of motions of the arm and hand.

This thesis has evaluated the use of Independent Component Analysis (ICA) on Surface Electromyography (sEMG), focusing on the biosignal applications. This research has identified and addressed the following four issues related to the use of ICA for biosignals: • The iterative nature of ICA • The order and magnitude ambiguity problems of ICA • Estimation of number of sources based on dependency and independency nature of the signals • Source separation for non-quadratic ICA (undercomplete and overcomplete) This research first establishes the applicability of ICA for sEMG and also identifies the shortcomings related to order and magnitude ambiguity. It has then developed, a mitigation strategy for these issues by using a single unmixing matrix and neural network weight matrix corresponding to the specific user. The research reports experimental verification of the technique and also the investigation of the impact of inter-subject and inter-experimental variations. The results demonstrate that while using sEMG without separation gives only 60% accuracy, and sEMG separated using traditional ICA gives an accuracy of 65%, this approach gives an accuracy of 99% for the same experimental data. Besides the marked improvement in accuracy, the other advantages of such a system are that it is suitable for real time operations and is easy to train by a lay user. The second part of this thesis reports research conducted to evaluate the use of ICA for the separation of bioelectric signals when the number of active sources may not be known. The work proposes the use of value of the determinant of the Global matrix generated using sparse sub band ICA for identifying the number of active sources. The results indicate that the technique is successful in identifying the number of active muscles for complex hand gestures. The results support the applications such as human computer interface. This thesis has also developed a method of determining the number of independent sources in a given mixture and has also demonstrated that using this information, it is possible to separate the signals in an undercomplete situation and reduce the redundancy in the data using standard ICA methods. The experimental verification has demonstrated that the quality of separation using this method is better than other techniques such as Principal Component Analysis (PCA) and selective PCA. This has number of applications such as audio separation and sensor networks.

During muscle contraction, electrical signals are generated by the muscle cells. The analysis of those signals is called electromyography (EMG). The EMG signal is mainly determined by physiological factors including so called central factors (central nervous system origin) and peripheral factors (muscle tissue origin). In addition, during the acquisition of EMG signals, technical factors are introduced (measurement equipment origin). The aim of this dissertation was to develop and evaluate methods to estimate physiological properties of the muscles using multichannel surface EMG (MCsEMG) signals. In order to obtain accurate physiological estimates, a method for automatic signal quality estimation was developed. The method’s performance was evaluated using visually classified signals, and the results demonstrated high classification accuracy. A method for estimation of the muscle fibre conduction velocity (MFCV) and the muscle fibre orientation (MFO) was developed. The method was evaluated with synthetic signals and demonstrated high estimation precision at low contraction levels. In order to discriminate between the estimates of MFCV and MFO belonging to single or populations of motor units (MUs), density regions of so called spatial distributions were examined. This method was applied in a study of the trapezius muscle and demonstrated spatial separation of MFCV (as well as MFO) even at high contraction levels. In addition, a method for quantification of MU synchronisation was developed. The performance on synthetic sEMG signals showed high sensitivity on MU synchronisation and robustness to changes in MFCV. The method was applied in a study of the biceps brachii muscle and the relation to force tremor during fatigue. The results showed that MU synchronisation accounted for about 40 % of the force tremor. In conclusion, new sEMG methods were developed to study muscle function and motor control in terms of muscle architecture, muscle fibre characteristics, and processes within the central nervous system.