Shuffle Attention Mechanism Based Bi-Directional Long Short-Term Memory for Detecting Neurodegenerative Disorders

The neurodegenerative disorders are the vital diseases that affects all over the world and Parkinson”s Disease (PD) and Alzheimer's Disease (AD) are mainly affected in neurodegenerative disorders. A retina helps for detect early level of neurodegenerative disorder through eye. To detect early steps of neurodegenerative disorder is impossible. To overcome problem, this research proposed a Shuffle Attention Mechanism depends on Bi-directional Long Short-Term Memory (ShAM-Bi-LSTM) for detect early stages of neurodegenerative disorders. The Rim-one and Drishti dataset are the collection of images that fed input to the proposed method. The data normalization is used for the pre-processed the input images. The pre-processed images are provided into the feature extraction such as Gray Level Co-occurrence Matrix (GLCM) that helps for extract features. The extracted features are used for classification, ShAM-Bi-LSTM is used for detect early stages of neurodegenerative disorders. To estimate performance of model, the evaluation metric such as accuracy of 96.5% and 97.2% for Rim-one and Drishti dataset are compared with existing works of Glaucoma Network Classification (GNC) and Multi-Layer Perceptron (MLP) respectively.