چكيده به لاتين
With the prevalence of 1% of the world population, epilepsy is one of the most common neurological disorders that can only lead to stroke. Patients with epilepsy suffer from frequent seizures that affect their quality of life due to the inability of the person and the unpredictability of the disease. For decades, treatment options for epilepsy have been largely drug-based and have undergone a low amount of surgery. Therefore, one of the important discussions that have recently been taken into consideration by the researchers is the diagnosis and classification of epilepsy. Although BCI-enabled devices are not usable compared to other controls, for brain-computer-based devices, the most important means of communicating with the outside world is for those who are unable to move. . These devices are used to command controls. These tools can extend the capabilities of a person. In this thesis, an BCI system is proposed for the diagnosis of epilepsy, which is used in the classification process for artificial neural networks. Having a good feature set helps to improve the accuracy of the BCI system. One of the most important steps in the BCI system is the extraction of the feature; therefore, researchers are more focused on this area. One of the important parameters in the brain-computer system is speed; therefore, it is always a high-speed system design that has a high performance at the same time. In this thesis, evolutionary algorithms (EAs) are applied after applying the known features extraction methods and the results show that when these feature extraction algorithms are used alone, high accuracy is not achieved. But when combined with these well-known extraction techniques and evolutionary algorithms, it becomes more precise to find that this is very appropriate for a common spatial pattern (CSP) algorithm. It should be noted that in the classification stage, the classification of neural networks is also used. Therefore, in addition to the fact that the accuracy of the system has improved, due to the fact that the number of selected features of 178 properties ranges from 70 to 80, the significant amount of complexity of the problem has been reduced. In this thesis, a BCI system is proposed for the diagnosis and classification of epilepsy. In the proposed BCI system, adding a feature selection block that performs a feature-specific process, in addition to lowering computational complexity, makes the accuracy even considerably improved.
Keywords:
BCI, EEG, CSP, Epilepsy, Classification, Neural Networks, Evolutionary Algorithms.
