A fine-tuned convolutional neural network model for accurate Alzheimer's disease classification

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Medical Diagnostics

Background:

• Alzheimer's disease (AD) is a leading cause of dementia, necessitating early diagnosis for effective management.
• Conventional machine learning models for AD detection face limitations including data dependency, poor generalization, and lengthy retraining periods.
• Deep learning models, while powerful, demand substantial computational resources and extensive datasets.

Purpose of the Study:

• To develop an improved method for early Alzheimer's disease diagnosis.
• To leverage transfer learning with convolutional neural networks (CNNs) for enhanced AD detection using MRI scans.
• To overcome the limitations of traditional machine learning and deep learning approaches in AD diagnosis.

Main Methods:

• Utilized magnetic resonance imaging (MRI) data for Alzheimer's disease (AD) diagnosis.
• Employed transfer learning techniques with pre-trained convolutional neural network (CNN) architectures: AlexNet, GoogleNet, and MobileNetV2.
• Tested various solvers including Adam, Stochastic Gradient Descent (SGD), and Root Mean Square Propagation (RMSprop).

Main Results:

• Achieved high classification accuracy: 99.4% on the Kaggle MRI dataset and 98.2% on the Open Access Series of Imaging Studies (OASIS) database.
• Demonstrated the effectiveness of transfer learning in improving diagnostic accuracy even with limited training data.
• Showcased reduced training costs and improved model generalization compared to conventional methods.

Conclusions:

• Transfer learning offers an effective solution to enhance the accuracy and efficiency of Alzheimer's disease diagnosis.
• The proposed CNN-based transfer learning model enables earlier and more precise identification of AD.
• Improved diagnostic capabilities can lead to better patient treatment management and insights into disease progression.