Ultra-low-power instrumentation amplifiers with cuff electrodes for detection of epileptic seizures

While many advances have been done for identifying and processing data generated by epileptic seizures, there are still many challenges of how to deal with it without medicines. In order to detect these biosignals, some prerequisites must be respected, otherwise the whole sensing is degraded by noise or interferences. Not only the electrodes but also the amplification chain determines the quality of the observed signal that will be utilized in all processes. This thesis discusses the types of electrodes used for ENG signals that could be extended to identifying epileptic seizures, alternatives for the core instrumentation amplifier with low-power consumption, configurations that make better use of all features presented and how to overcome non-idealities generated by the variable behaviour of the electrode-tissue interfaces. The source of the signals will be the vagus nerve, a complex but large nerve in the human body. A key element in the whole process is how to deal with artefacts, since the order of magnitude from the captured useful ENG signals (µV) can be two or three orders of magnitude smaller than the one from muscles (EMG signals), not allowing proper measurements and analysis. For this, some configurations with the instrumentation amplifiers were analyzed checking advantages and drawbacks for each of them and the decision was made after the investigation of another non-ideality intrinsic from cuff electrodes: the impedance imbalances, that greatly degrades the property of linearized external potential interferences from the cuff electrodes. This issue was dealt by an addition of digital control over the gain from the amplifying chain that is possible in only one of these architectures.