MRI-Based Epilepsy Seizure Detection Using Combined Approach of CNN With SVM Networks

The quality of life for those affected by the neurological illness epilepsy is greatly diminished by the frequent occurrence of seizures. Nearly half a million individuals throughout the globe live with this chronic disease. Better results and faster treatment could result from the early identification of seizures via the examination of medical imaging. When applied to MRI and EEG data, deep learning algorithms have shown encouraging results for epilepsy classification. Consistently improving these models' accuracy and reliability is key to achieving great performance in epilepsy detection. To solve the issue of automatically detecting instances of epilepsy in MRI images, this study introduces a new hybrid machine learning method. Based on these specific qualities, the support vector algorithm (SVM) may detect activity related to seizures in a scan. Researchers looked at two sets of magnetic resonance imaging (MRI) images from people with epilepsy that had seizures in order to test the notion. This research successfully detects and predicts the beginning of seizures associated with epilepsy in MRI examinations using the suggested hybrid technique, which combines a neural network based on convolution (CNN) with a vector machine with support (SVM). It is clear from these results that computerized prediction systems have achieved significant progress in medical diagnosis. Using 5-fold cross-validation, the proposed SVM-CNN system achieves a detection accuracy greater than 98.69% with dataset 2 and 98.47% with dataset 1, surpassing previous standards. Seizure detection becomes more accurate and interpretable when deep learning and shallow learning are combined. This diagnostic method allows for the early administration of anti-epileptic medication, which may lead to better patient outcomes.