Seizure detection by convolutional neural network-based analysis of scalp electroencephalography plot images

• Published in: NeuroImage clinical Vol. 22; p. 101684
• Main Authors: Emami, Ali, Kunii, Naoto, Matsuo, Takeshi, Shinozaki, Takashi, Kawai, Kensuke, Takahashi, Hirokazu
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier Inc 01.01.2019; Elsevier
• Subjects: Adolescent; Adult; Child; Convolutional neural networks; Deep Learning; Electroencephalography - methods; Electroencephalography - standards; Epilepsies, Partial - diagnosis; Epileptic seizure; Female; Humans; Image Processing, Computer-Assisted - methods; Image Processing, Computer-Assisted - standards; Male; Middle Aged; Radiology; Regular; Scalp; Scalp electroencephalogram; Seizure detection; Seizures - diagnosis; Sensitivity and Specificity; Young Adult; CNN; STD; Sz; PLE; OLE; PFA; TLE; Epileptic seizure; Deep learning; rec; Seizure detection; FIAS; MF; DMF; PWE; RSA; ROC; Sup; Sb; RED; FBTCS; FAS; Convolutional neural networks; Scalp electroencephalogram; FLE; TS; seizure; subject; rhythmic slow activity; recording; occipital lobe epilepsy; multifocal; suppression; parietal lobe epilepsy; receiver operating characteristic; repetitive epileptiform discharge; temporal lobe epilepsy; patients with epilepsy; paroxysmal fast activity; focal aware seizure; focal to bilateral tonic-clonic seizure; frontal lobe epilepsy; focal impaired awareness seizure; diffuse multifocal; tonic seizure; standard deviation
• ISSN: 2213-1582; 2213-1582
• DOI: 10.1016/j.nicl.2019.101684

We hypothesized that expert epileptologists can detect seizures directly by visually analyzing EEG plot images, unlike automated methods that analyze spectro-temporal features or complex, non-stationary features of EEG signals. If so, seizure detection could benefit from convolutional neural networks because their visual recognition ability is comparable to that of humans. We explored image-based seizure detection by applying convolutional neural networks to long-term EEG that included epileptic seizures. After filtering, EEG data were divided into short segments based on a given time window and converted into plot EEG images, each of which was classified by convolutional neural networks as ‘seizure’ or ‘non-seizure’. These resultant labels were then used to design a clinically practical index for seizure detection. The best true positive rate was obtained using a 1-s time window. The median true positive rate of convolutional neural networks labelling by seconds was 74%, which was higher than that of commercially available seizure detection software (20% by BESA and 31% by Persyst). For practical use, the median of detected seizure rate by minutes was 100% by convolutional neural networks, which was higher than the 73.3% by BESA and 81.7% by Persyst. The false alarm of convolutional neural networks' seizure detection was issued at 0.2 per hour, which appears acceptable for clinical practice. Moreover, we demonstrated that seizure detection improved when training was performed using EEG patterns similar to those of testing data, suggesting that adding a variety of seizure patterns to the training dataset will improve our method. Thus, artificial visual recognition by convolutional neural networks allows for seizure detection, which otherwise currently relies on skillful visual inspection by expert epileptologists during clinical diagnosis. [Display omitted] •Artificial visual recognition of scalp EEG plot images successfully detects seizures.•CNN-based automatic detection performed better than commercial software.•Customized CNN learning using large datasets improves detection.