Dynamic Seizure Recognition: Invelling Epileptic Patterns with CNN-LSTM Networks

Epilepsy is a condition affecting the nervous system. which is diagnosed with the aid of Electroencephalography (EEG)through a neurologist or experts. Frequent and unpredictable seizures characterise it may cause loss of consciousness, altered awareness, or unusual sensations. Reducing the use of conventional diagnostic methods is crucial, as is making the diagnosis of this condition early on, before behavioural. Signs appear. This project aims to introduce an intelligent framework for diagnosing neurological disorders (epilepsy) based on EEG recordings utilising the techniques of deep learning. In EEG signals the epileptic seizures are recognised with the help of sharp spikes of the signals. The focus lies on developing a system to transform the subjective qualitative diagnostic criteria into a more objective quantitative prognosis criterion and to analyse hidden dynamics of the Neurological data for extracting more information about the pathological versus normal status of the signals. 1D-Convolutional Neural Network and a long short-term memory hybrid model of deep learning have been employed to identify epileptic seizures. In this study, the authors used a CNN-LSTM network with 20 epochs on two separate datasets to get an optimal detection rate of EEG data exhibiting seizures compared to those without. By adding noise to the EEG signals the suggested model's adaptability has been examined. Neurologists will find the suggested methods useful for detecting seizures in real-time.