Neurodegenerative Diseases classified based on Salient Brain Patterns

Abstract

The MRI brain images can save various lives by identifying defects. A method for identifying neurodegenerative Alzheimer's disease in brain MRI images involves a comprehensive workflow. First, convert the input brain image into a grayscale format to simplify the data. Following this,  generate a saliency map from the pre-processed image, which highlights the most relevant areas in the brain image. Normalization of this saliency map ensures that the data falls within a standardized range, enhancing consistency. The critical step involves applying kernel fusion to the normalized saliency map to extract features from the image. These features encompass elements like strength, textural characteristics, statistical information, and binary tissue segmentations. Feature extraction is crucial as it reduces the dimensionality of the data while preserving essential information. Finally, you employ a Support Vector Machine (SVM) classifier, which is fed with these extracted features to classify brain images as indicative of Alzheimer's disease or not. While your proposed algorithm aims to streamline the computational process and reduce irrelevant features, it is essential to validate its effectiveness using diverse MRI brain images and collaborate with medical professionals for thorough evaluation and clinical applicability. The claim that salient regions identified by your method are systematically relevant for Alzheimer's disease discrimination should be substantiated through empirical evidence and validation studies to ensure its clinical utility.