作者: Hoda Daou , Fabrice Labeau
DOI: 10.1109/JBHI.2014.2346493
关键词: Scalp 、 Novel technique 、 Electroencephalography 、 Eeg analysis 、 Dipole 、 Artificial intelligence 、 Speech recognition 、 Computer science 、 Pattern recognition 、 Coding (social sciences) 、 Inverse 、 Inverse problem
摘要: A novel technique for electroencephalogram (EEG) compression is proposed in this paper. This models the intrinsic dependence inherent between different EEG channels. It based on methods borrowed from dipole fitting that usually used order to find a solution classic problems analysis: inverse and forward problems. To compress signals, model approximated source dipoles first provide an approximation of recorded signals. Then, (based smoothness factor) appropriate coding techniques are suggested residuals process. Results show works well recordings patients, even able near-lossless certain types recordings.