Alzheimer' disease prediction and classification using CT images through machine learning.

Area of Science:

• Neuroimaging and computational neuroscience
• Artificial intelligence in medicine
• Neurology and cognitive science

Background:

• Early detection of Alzheimer's disease (AD) is crucial for effective treatment before irreversible cognitive decline.
• Neuroimaging data analysis is a key area for identifying AD symptoms.
• Automated algorithms can significantly improve the speed and accuracy of early AD diagnosis.

Purpose of the Study:

• To develop and evaluate an automated system for the early detection of Alzheimer's disease using deep learning techniques.
• To assess the efficacy of advanced neural network architectures in classifying AD from neuroimaging data.
• To identify individuals with mild cognitive impairment (MCI) who are at risk of progressing to AD.

Main Methods:

• Utilized a deep learning approach, specifically the Improved Faster Recurrent Convolutional Neural Network (IFRCNN) model, adapted from the Visual Geometry Group (VGG)-16 architecture.
• Employed action recognition as a feature extraction method within a mathematical model for image categorization.
• Trained and validated the system using the Alzheimer's Neuroimaging Initiative (ADNI) dataset.

Main Results:

• The proposed IFRCNN system demonstrated high accuracy in classifying Alzheimer's disease.
• Achieved an accuracy rate of 98.32% on the ADNI dataset, indicating robust performance.
• The system effectively identified patterns in neuroimaging data indicative of early-stage AD and mild cognitive impairment.

Conclusions:

• The developed deep learning system shows significant promise for the early and accurate identification of Alzheimer's disease.
• Automated analysis of neuroimaging data using advanced ML models can enhance diagnostic capabilities.
• This approach supports timely therapeutic interventions, potentially improving patient outcomes and managing the expected risk of AD progression.