Optimizing a 3D convolutional neural network to detect Alzheimer's disease based on MRI

Area of Science:

• Neuroscience
• Medical Imaging
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is a progressive neurological disorder causing cognitive decline, with irreversible damage occurring before symptom onset.
• Early diagnosis of AD is critical for timely intervention, slowing progression, and enhancing patient quality of life.
• Machine learning and neuroimaging, especially deep learning with MRI, show promise for early AD detection.

Purpose of the Study:

• To develop and evaluate a 3D convolutional neural network (3D-CNN) for enhanced classification accuracy in early Alzheimer's disease detection.
• To leverage the OASIS-3 database and advanced preprocessing techniques for robust diagnostic model development.

Main Methods:

• A 3D-CNN model was developed to process full 3D MRI scans, preserving spatial information unlike 2D approaches.
• Advanced preprocessing techniques, including intensity normalization and noise reduction, were applied to MRI data.
• The model was trained and validated using data from the OASIS-3 database.

Main Results:

• The proposed 3D-CNN model achieved a classification accuracy of 91% for Alzheimer's disease detection.
• The 3D approach outperformed traditional 2D methods by preventing information loss during dimensionality reduction.
• Enhanced image quality through preprocessing contributed to improved classification performance.

Conclusions:

• Deep learning, specifically the developed 3D-CNN, shows significant potential for reliable and efficient early Alzheimer's disease diagnosis.
• The findings support the use of advanced neuroimaging and AI for improved clinical decision-making in AD.
• This approach paves the way for better patient outcomes through earlier intervention.