RESUMEN
The treatment of epilepsy, the second most common chronic neurological disorder, is often complicated by the failure of patients to respond to medication. Treatment failure with anti-seizure medications is often due to the presence of non-epileptic seizures. Distinguishing non-epileptic from epileptic seizures requires an expensive and time-consuming analysis of electroencephalograms (EEGs) recorded in an epilepsy monitoring unit. Machine learning algorithms have been used to detect seizures from EEG, typically using EEG waveform analysis. We employed an alternative approach, using a convolutional neural network (CNN) with transfer learning using MobileNetV2 to emulate the real-world visual analysis of EEG images by epileptologists. A total of 5359 EEG waveform plot images from 107 adult subjects across two epilepsy monitoring units in separate medical facilities were divided into epileptic and non-epileptic groups for training and cross-validation of the CNN. The model achieved an accuracy of 86.9% (Area Under the Curve, AUC 0.92) at the site where training data were extracted and an accuracy of 87.3% (AUC 0.94) at the other site whose data were only used for validation. This investigation demonstrates the high accuracy achievable with CNN analysis of EEG plot images and the robustness of this approach across EEG visualization software, laying the groundwork for further subclassification of seizures using similar approaches in a clinical setting.
Asunto(s)
Electroencefalografía , Epilepsia , Aprendizaje Automático , Redes Neurales de la Computación , Convulsiones , Humanos , Electroencefalografía/métodos , Convulsiones/diagnóstico , Convulsiones/fisiopatología , Epilepsia/diagnóstico , Epilepsia/fisiopatología , Adulto , Masculino , Algoritmos , Femenino , Persona de Mediana EdadRESUMEN
OBJECTIVE: Using video-EEG (v-EEG) diagnosis as a gold standard, we assessed the predictive diagnostic value of home videos of spells with or without additional limited demographic data in US veterans referred for evaluation of epilepsy. Veterans, in particular, stand to benefit from improved diagnostic tools given higher rates of PNES and limited accessibility to care. METHODS: This was a prospective, blinded diagnostic accuracy study in adults conducted at the Houston VA Medical Center from 12/2015-06/2019. Patients with a definitive diagnosis of epileptic seizures (ES), psychogenic nonepileptic seizures (PNES), or physiologic nonepileptic events (PhysNEE) from v-EEG monitoring were asked to submit home videos. Four board-certified epileptologists blinded to the original diagnosis formulated a diagnostic impression based upon the home video review alone and video plus limited demographic data. RESULTS: Fifty patients (30 males; mean age 47.7â¯years) submitted home videos. Of these, 14 had ES, 33 had PNES, and three had PhysNEE diagnosed by v-EEG. The diagnostic accuracy by video alone was 88.0%, with a sensitivity of 83.9% and specificity of 89.6%. Providing raters with basic patient demographic information in addition to the home videos did not significantly improve diagnostic accuracy when comparing to reviewing the videos alone. Inter-rater agreement between four raters based on video was moderate with both videos alone (kappaâ¯=â¯0.59) and video plus limited demographic data (kappaâ¯=â¯0.60). SIGNIFICANCE: This study demonstrated that home videos of paroxysmal events could be an important tool in reliably diagnosing ES vs. PNES in veterans referred for evaluation of epilepsy when interpreted by experts. A moderate inter-rater reliability was observed in this study.