IMPLEMENTATION STUDY OF CONVOLUTIONAL NEURAL NETWORK FOR ALZHEIMER’S DISEASE MRI IMAGE CLASSIFICATION USING TENSORFLOW AND KERAS ON GOOGLE COLABORATORY

• Main Author: Rizalul Yahya, Fadilla
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/71631
• Institution: Institut Teknologi Bandung

Global statistics data from Global Burden of Disease Study stated that Alzheimer's disease became one of the diseases that increased among fifty diseases with the most cases of death in the period of 1990 to 2013. Observation and identification of Alzheimer's disease needs high accuracy because of the disease has a certain level and type, as to provide appropriate medical action recommendations. One of method to help with the classification of Alzheimer's disease that expected to reduce errors in diagnosis is deep learning. This study aims to evaluate the performance of four CNN model in classifying Alzheimer's disease MRI images based on TensorFlow and Keras on Google Colaboratory. The dataset used was obtained from ADNI website consists of T1-weighted images from 20 patients in DICOM format converted into JPG. The dataset is divided into three different folders and carried out with the process of data augmentation to increase the diversity of data. During modelling, each model are given the same hyperparameters. The results showed that the model has a different values of accuracy. Model D which is VGG-19 achieved an accuracy of 99% with an error of 2.06%, which indicates that the model has good ability to classify images. The results of this study may provide a useful reference for future research in the field of medical image analysis using deep learning method.