Paper Title
Mining of Patterns Through Brain Images and EEG Signals for Early Detection of Alzheimer’s Disease

Abstract
Alzheimer’s disease is a neuro-degenerative disease that initially presents itself as dementia, and slowly leads to the death of brain cells, cohorts typically experience loss of memory and a change in social behavior. To identify the disease at an early stage is particularly challenging as the symptoms are subtle and hard to diagnose at early stages. The diagnosis of the disease is usually done by identifying protein deposits, amyloid proteins, known as plaques, or tau and through studying brain images. Another efficient method used in diagnosing the disease is through studying the hippocampus region of the brain. EEG signals have also been fruitful in diagnosing the disease. This paper aims at conducting a comprehensive study on the most effective ways to identify Alzheimer’s disease at a rudimentary stage. Using a combination of various extraction methods and an eclectic set of data: Positron Emission Tomography(PET), Functional Magnetic Resonance Imaging(fMRI), Magnetic Resonance Imaging(MRI)and EEG data, the paper presents an unprecedented approach to diagnose Alzheimer’s at an early stage. The paper approaches the problem by first working on a proof of concept approach, experimentally validating popular diagnostic methods such as brain atrophy, hippocampal shrinkage, activation of brain and trends in EEG signals.The early diagnosis of the disease was performed using a myriad of techniques using a variable collection of data. Image data-set collected from Alzheimer’s Disease Neuro-imaging Initiative(ADNI) was pre-processed and utilized to extract features and particularly hippocampus region extraction was performed using various methodologies like the U-Net. After successful hippocampus region extraction, feature extraction was performed using techniques such as Gray Level Coocurrence Matrix (GLCM).Finally, the features were classified using Random forest. By leveraging the f-MRI data the active voxels of the image were identified and used for classification by 3D CNN. In recent years, there has been an increased interest in the use of deep learning models for analyzing electroencephalography (EEG) data. In this paper an attempt was made to convert the EEG data into images using the grammian angular matrix. These images were then fed into deep learning models such as convolutional neural networks (CNNs) for classification. The advantage of using images is that CNNs are well suited for image processing tasks and can learn important features directly from the raw data. Furthermore, converting EEG data to images can provide visual insights into the underlying patterns and structures in the data, making it easier to interpret and analyze. Keywords - Alzheimer’s, hippocampus extraction, ROI based feature extraction, multi-modal analysis