Feature extraction and classification of Dementia with neural network

Abstract

Dementia, a neurodegenerative disorder related to aging, causes reduction in the cognitive abilities of man due to the degeneration of brain structures. Early detection of dementia with higher accuracy is essential for treatment. A technique for the accurate detection of Dementia from the Magnetic Resonance Images (MRI) of brain is proposed in this paper. Segmentation-based Fractal Texture Analysis (SFTA) technique is used for the extraction of features. Fractal dimensions and texture features are extracted from the binary images obtained after breaking down the image by the Two Threshold Binary Decomposition algorithm. Dementia classification is accomplished with Neural Network. This algorithm is successfully tested using 3D brain MRI images obtained from the OASIS dataset with a classification accuracy of 97.5%.