About 10% to 15% of patients admitted at the urgent care experience some form of seizure while unconscious. Because they are unconscious, the doctors cannot observe the direct effects of the seizure, and have to rely on Electro-Encephalograms (EEGs) to detect that a seizure occurred. This is costly, time-consuming and invasive for the patients. Thus, this detection is currently almost never done, and the patients suffering from a seizure may experience some side-effects of them.
The FADING SUN project encompasses the hardware realization of a new lightweight and cheap electrodes cap, for easier capture of EEG signals, and the software side, the automatic detection of the occurrence or future occurrence of a seizure from the signals acquired by the cap. The VUB AI Lab is responsible for the second half, the signal processing and Machine Learning aspect of detecting future seizures from EEG signals.
Research topics:
- Supervised Learning
- Convolutional Neural Networks
- Training methodology (ensuring valid and good results on a challenging dataset)
