Use this URL to cite or link to this record in EThOS:
Title: Wearable electroencephalography for long-term monitoring and diagnostic purposes
Author: Iranmanesh, Saam
ISNI:       0000 0004 7427 7504
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Truly Wearable EEG (WEEG) can be considered as the future of ambulatory EEG units, which are the current standard for long-term EEG monitoring. Replacing these short lifetime, bulky units with long-lasting, miniature and wearable devices that can be easily worn by patients will result in more EEG data being collected for extended monitoring periods. This thesis presents three new fabricated systems, in the form of Application Specific Integrated Circuits (ASICs), to aid the diagnosis of epilepsy and sleep disorders by detecting specific clinically important EEG events on the sensor node, while discarding background activity. The power consumption of the WEEG monitoring device incorporating these systems can be reduced since the transmitter, which is the dominating element in terms of power consumption, will only become active based on the output of these systems. Candidate interictal activity is identified by the developed analog-based interictal spike selection system-on-chip (SoC), using an approximation of the Continuous Wavelet Transform (CWT), as a bandpass filter, and thresholding. The spike selection SoC is fabricated in a 0.35 μm CMOS process and consumes 950 nW. Experimental results reveal that the SoC is able to identify 87% of interictal spikes correctly while only transmitting 45% of the data. Sections of EEG data containing likely ictal activity are detected by an analog seizure selection SoC using the low complexity line length feature. This SoC is fabricated in a 0.18 μm CMOS technology and consumes 1.14 μW. Based on experimental results, the fabricated SoC is able to correctly detect 83% of seizure episodes while transmitting 52% of the overall EEG data. A single-channel analog-based sleep spindle detection SoC is developed to aid the diagnosis of sleep disorders by detecting sleep spindles, which are characteristic events of sleep. The system identifies spindle events by monitoring abrupt changes in the input EEG. An approximation of the median frequency calculation, incorporated as part of the system, allows for non-spindle activity incorrectly identified by the system as sleep spindles to be discarded. The sleep spindle detection SoC is fabricated in a 0.18 μm CMOS technology, consuming only 515 nW. The SoC achieves a sensitivity and specificity of 71.5% and 98% respectively.
Supervisor: Rodriguez-Villegas, Esther Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral