An Efficient Framework for EEG Signals Processing Using Backpropagation Multilayer Convolutional Neural Network for Brain Tumor Diagnosis

The average patient's life expectancy is shortened by a Brain Tumour (BT), one of the most dangerous and aggressive diseases. The prognosis for patients with BTs is worse when they get inadequate medical therapy or a wrong diagnosis. An electroencephalogram (EEG) records electrical activity in the brain in a non-invasive manner. Several modern imaging methods including MRI, SPECT, and PET, EEG offer superior tumor-specific diagnoses. This study offers a diagnostic method for identifying if an EEG signal is cancerous or normal. Initially, EEG signals were first used to capture brain data, which was then preprocessed using the stationary wavelet transform (SWT) method. The features of the brain data are then selected using a heuristic assessment. To extract features, a Grey-Level Co-Occurrence Matrix (GLCM) is utilized. Furthermore, we suggest a novel BackPropagationMultilayer Convolutional Neural Network (BP-MCNN) approach for BT diagnostics for tumor categorization. A set of database samples was used to provide the neural network with the statistical information that had been extracted from the EEG signal during the training phase. After being trained, a neural network can accurately categorize a test set of samples into normal or cancerous. The suggested approach provides superior classification accuracy than the current methods, according to experimental data.