Analysis of EEG signals and study of seizures for movement of prosthetic arm

Amputations, amounting to a staggering one million cases globally each year, pose significant challenges to affected individuals quality of life. Prosthetic limbs have emerged as an essential solution, enabling amputees to perform daily activities with proficiency comparable to their able-bodied peers. Integrating Electroencephalography (EEG) signals for prosthetic arm control has empowered users to move their artificial limbs intuitively. The impact of focal seizures on prosthetic arm control is investigated through EEG signal analysis and various classification techniques. EEG signals from an existing dataset are preprocessed and classified using Support Vector Machine, Fuzzy Logic and CNN and their accuracy is compared. Here, CNN is found to have the highest accuracy. In the proposed work, the CNN model is trained using transfer learning technique by employing an existing architecture AlexNet. An Arduino microcontroller is used to control a dc servo motor which is connected to the prosthetic arm via strings. By leveraging the classification results, the prosthetic arm is activated when the prediction is normal and deactivated when the prediction is seizure.