Abstract:

Abstract Schizophrenia is a devastating mental illness that affects millions of people throughout the world and profoundly alters the way they think, feel, and act. The illness may be accurately diagnosed thanks to the EEG signal. Researchers can capture brain activity non-invasively using an electroencephalogram (EEG). It's a useful technique for diagnosing a wide range of neurological illnesses and diseases of the brain. An electrooculogram, a large electrical potential around the eyes produced by blinking or eye movement, is recorded. When extracortical activity spreads to the scalp, it contaminates EEG data. Ocular artifacts are a term used to describe these Ocular artifacts (OAs). When it comes to EEG analysis, one of the most significant types of interference is the ocular artifact. An important part of the analysis before processing EEG data is OAs removal/reduction. To eliminate the common classical OAs, a multi-channel EEG or a second electrooculogram recording is required. This paper developed a Fejer-Korovkin pre-processing model for OA elimination in the EEG signal. The developed model has termed the FK_DARKNET model with DARKET integrated with the RCNN (Recurrent Convolutional Neural Network) applied over the YOLO architecture. The proposed FK_DARKNET comprises the YOLO model for the classification of Schizophrenia and normal EEG signals in the images. The proposed FK_DARKNET model's efficacy is evaluated in light of the state-of-the-art method. The results of the comparison show that the suggested FK_DARKNET performs better than its competitors in terms of accuracy, TPR, and TNR.

Keywords: