Abstract:
Alzheimer’s is the most common type and cause of dementia (loss in cognitive skills)
and it does lead to the death of the victim. Stages of Alzheimer’s Disease can be
predicted by the use of Magnetic Resonance Imaging (MRI) images. The objective of
this project is to develop image recognition system to predict stage of Alzheimer’s
disease through MRI images. This report explains the techniques, methods and
approaches used to predict the stage of Alzheimer’s disease such as image pre processing, features extraction, classification. The project uses the class ofdeep neural
network known as Convolution Neural Network (CNN), Tensor flow library Keras to
develop the software. The benefit ofusing CNN lies within its capability of adaptively
learning spatial hierarchies offeatures using the backpropagation technique provided
by the building blocks, such as convolution layers, pooling layers, and fully connected
layers. After trials and errors, a suitable set of training parameters are defined and
network structure that consist of 1 input layer, 2 hidden layers and 1 output layer with
69 input neurons, 324 neurons for both hidden layers and 38 neurons for output layer
is created. The system first proceeds with the pre-process ofthe captured image with
threshold, inverting and smoothing. Filtering, segmentation, resizing and features
extraction are also performed in the process. Next, the feed forward process through
the network is invoked to yield an output matrix. Based on the output matrix, the
recognized character can be determined. This system is designed to customize the
network for an individual user. With an accuracy of 95% we have also included
recommendations and conclusions for future development and in the report.
