Clinical Algorithms

EEG cap with TEGs will record an electrical activity of a brain using multiple electrodes placed on the scalp. Instead of raw data storage, sensor will do feature extraction for automatic seizure detection. Several medical algorithms are considered to trade off between diagnostic power and power consumption.

Spikes stand out from a normal brain activity on the EEG. Spike density bio-markers are of the great importance for the diagnosis of epilepsy in children. Spike detection algorithms are being implemented within clinical application of real-time EEG signal analysis. They are based on template matching, an approach for signal pattern recognition often used for bio-medical signals. Real life data obtained from Kinderspital is being used to test the algorithm implementation.


The trigger for abnormal behaviour can be found in the environment. Patient environment monitor (PEM) is used to keep a track of the environment and includes accelerometer and gyroscope to identify the amount of movement of the patient. Using very low power microphone and low resolution camera, there is the ability to determine characteristics of the environment of the patient such as level of noise and level of light.

Wearable ICT only allows low-density EEG (LDEEG) for monitoring of patients. However, it suffers from some limitations such as functional connectivity estimation and source estimation. The functional connectivity (FC) is an appealing feature of the brain electromagnetic activity that has been shown to deteriorate in a number of psychiatric and neurodegenerative diseases including Alzheimer’s disease. Long-term FC provides important data for diagnosis. The project seeks to find an approach to assess FC from LDEEG and make measurements with wearable ICT useful.

Analyzing relevant information extracted from multiple bio-signals such as ECG, respiration rate and skin conductance, there is also a possibility of assessing emotions.

Optimal System Design

Body energy harvesters produce unreliably small amounts of power and moreover at lower voltages in compare to other applications. On the other hand, local processing operation in body sensor nodes can be power hungry. Designing an optimal power converter includes low power blocks and techniques employed in circuit level as well as minimizing the loss of the system by trade-offs between switches sizing and working frequency.

Acceptable precision of EEG signal monitoring devices is maintained by placing an active amplifier and signal conditioning next to the electrodes. Those components are integrated into an active EEG electrode.

Thermoelectric generators (TEGs) directly convert thermal energy into electrical energy using the Seebeck effect and satisfy crucial requirements in wearables e.g. small form factor, robustness and maintenance-free operation without movable parts. However, due to the low temperature gradients, thermal harvesters worn on the body need to be highly optimized to meet the power requirements of a wearable system. The idea of zero-power active electrode combines micro-TEGs with active EEG electrodes. Therefore, an autonomous active EEG electrode is designed to contain an EEG electrode, Thermo-Electric Generator (TEG) harvester, TEG power converter and amplifier.


Harvesting kinetic energy from low frequency body movements provides the additional energy required by the project. From a point of human compliance and device’s complexity, nanocomposite energy harvesting devices are found the most suitable.