Automatic seizure detection by convolutional neural networks with computational complexity analysis

•Early diagnosis, including the combination of high accuracy and low computational complexity, is important in real applications of computer-aided diagnosis systems.•We present an approach to seizure detection from electroencephalogram (EEG) signals with complexity analysis and comparison.•The methodology was tested on the short-term Bonn EEG dataset and the long-term CHB-MIT EEG dataset. The results lead to the conviction that the designed solution represents a suitable tool.•The proposed solution should be implemented in all clinical or home environments for decision support.