Executive Summary : | Alzheimer's disease (AD) is a widespread form of dementia that affects the elderly population above 65 years old, with an estimated 13.8 million affected by 2060. Predicting AD at the transitional stage, mild cognitive impairment (MCI), is crucial. Approximately 15% of 65-year-olds experience MCI without any notable symptoms. Some MCI patients are converted to AD within 5 years, known as converter MCI (cMCI), while others are reversible to cognitive normal. Studies have explored the alterations in structural atrophies and brain networks using structural and functional magnetic resonance imaging (MRI) scans of AD patients. Structural MRI (sMRI) visualizes AD and MCI-related symptoms, while functional MRI (fMRI) investigates the intrinsic connectivity of AD-related neurological disorders. A multimodality approach, including MRI and fMRI, can improve prediction of MCI conversion to AD and reliable diagnosis. The project aims to develop a deep learning-based open-source toolbox to assist physicians in diagnosing cognitive normal (CN), MCI, and AD patients and predicting the risk of degeneration progression. The model will integrate clinical assessment tests, MRI and fMRI scans, and carry out a two-stage classification. The first layer will perform multiclass classification to categorize subjects into CN, MCI, and AD using a deep learning-based model. The second layer will observe MCI to AD conversion using three years follow-up from baseline diagnosis. The developed model will be verified by expert neurologists from AlIMS, Delhi, and will be available as open-source to help medical experts with AD diagnosis and predict disease progression in MCI patients. |