epiPAPA is a neural network based seizure detection algorithm that runs in real-time, or offline with a variety of digital EEG file formats including EDF.
epiPAPA operates in five different modes:
- Generic scalp seizure detection
- Generic intracranial seizure detection
- Patient-specific trainable scalp seizure detection
- Patient-specific trainable intracranial seizure detection
- Patient-specific seizure onset/preonset detection
The epiPAPA module extracts various spectral, time-domain and wavelet analysis features from the raw EEG and uses an artificial neural network (ANN) to produce a certainty measure for seizure recognition.
In our testing, we evaluated epiPAPA against a a standard seizure detection package. In all modes, epiPAPA reduced the false alarm rate as measured in false alarms per hour. In most modes, the number of true seizures that were detected either stayed the same, or increased.