Multimodal MRI Based Classification and Prediction of Alzheimer’s Disease Using Random Forest Ensemble

Abstract

Alzheimer’s disease (AD) is a neurodegenerative disorder that affects millions of people worldwide and it accounts for a significant decrease in the quality of life of patients and their families. Currently, available treatment options for AD is merely palliative and no drugs are available for the inexorable progression of the disorder that is diagnosed during the later stage of the disease. So the early diagnosis of AD is an optimal strategy in formulating the treatment plan. Neuroimaging modalities like Magnetic Resonance Imaging (MRI), resting-state functional Magnetic resonance imaging (rs-fMRI), Diffusion Tensor Imaging (DTI) and Positron emission tomography (PET) are used to diagnose the structural and functional alteration caused by AD. For the past few years, machine learning methods are widely used to analyze the neuroimaging data acquired from MRI imaging modalities for the diagnosis and prediction of neurological disorder. In this work, the random forest classification algorithm is used to classify and predict Alzheimer’s disease. The data set that is used in this study is TADPOLE data set, which has been acquired from Alzheimer’s neuroimaging Initiative (ADNI). In this work, the multiclass classification that distinguishes the different level of Alzheimer’s disease has achieved an accuracy comparable to current research in the prediction of AD.