A novel framework based on biclustering for automatic epileptic seizure detection

摘要: Automatic epileptic seizure detection based on electroencephalogram is crucial to epilepsy diagnosis and treatment. However, the large numbers of time series make it quite challenging establish a high performance automatic method. Considering different physiological states brain could be characterized by distinct combinations or interactions similar discontinuous local temporal patterns, novel framework biclustering for proposed in this paper. First, CC algorithm used identify patterns. Then, bicluster membership matrix using new similarity measurement constructed reduce dimensionality. At last, ELM classifier adopted discriminate between seizure-free EEG signals. With extensive comparative studies evaluations publicly available Bonn dataset, indicates that not only automatically detect predict an with performances respect accuracy, robustness efficiency, but also implicitly provide valuable knowledge studying mechanisms epilepsy.