Enhanced spatio-temporal clustering in the detection of neonatal seizures using context-based rules
Joyeeta Mitra1, John R Glover, James D Frost
1Department of Electrical & Computer Engineering, Houston University, TX, USA.
Abstract:
This work describes the clustering stage of a three-stage automated neonatal seizure detection system. This stage clusters spatio-temporally the short candidate seizure segments detected in prior stages, and then applies a variety of context-based rules to eliminate false detections and determine the final detected seizures. The work discusses important considerations in the implementation of rules and presents preliminary results.
