< Back to previous page

Project

Deep Learning approaches for mining wearable brain monitoring data in epilepsy(CSC-2019-18)

Automated seizure detection from the electroencephalograms (EEG) signals plays an important role in epilepsy diagnosis. Although many approaches have shown their abilities in detecting seizure with scalp EEG, they have to face the issue of data insufficiency since the clinical scalp EEG signals are expensive and inconvenient to collect. Recently, wearable devices provide a much easier and cheaper way to collect EEG and therefore make it possible to monitor seizure in daily life. However, learning directly from wearable data is nontrivial since annotation of wearable data is often limited because the ground truth is only defined on scalp EEG. When labelled data is scarce, it becomes difficult to train a well performed detection framework. To remedy this issue, this project establish an automated seizure detection framework that leverage the unannotated wearable data as well as available annotated scalp data.

Date:21 Sep 2020 →  Today
Keywords:Seizure Detection, EEG
Disciplines:Multimedia processing not elsewhere classified, Data mining
Project type:PhD project