Classification method based on surf and sift features for alzheimer diagnosis using diffusion tensor magnetic...

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Biomedical Engineering

Background:

• Alzheimer's disease (AD) diagnosis is challenging due to progressive microstructural brain changes not visible with standard MRI.
• Diffusion Tensor Imaging (DTI) offers insights into these subtle alterations, crucial for early detection.

Purpose of the Study:

• To develop and evaluate a computer-aided diagnosis (CAD) framework for characterizing Alzheimer's disease progression using DTI.
• To differentiate between Alzheimer's disease (AD), mild cognitive impairment (MCI), and normal controls (NC) based on DTI-derived visual patterns.

Main Methods:

• Utilized DTI data, specifically Fractional Anisotropy (FA), Mean Diffusivity (MD), and Radial Diffusivity (RD) maps.
• Employed SIFT and SURF feature descriptors with a bag-of-words model to create AD-specific signatures in the hippocampus.
• Investigated multiclass and binary classification, including late fusion of visual map features for improved decision-making.

Main Results:

• The proposed DTI-based CAD system achieved high accuracies, with feature fusion reaching 95.2% in multiclass classification.
• Binary classification using FA features attained 97.5% accuracy.
• Decision-level fusion demonstrated robust performance, with accuracies up to 97.5% for AD/NC discrimination and a precision boost of 96%.

Conclusions:

• The DTI-based CAD framework shows significant potential as a reliable tool for detecting Alzheimer's disease progression.
• The system's ability to capture microstructural changes suggests promise for earlier diagnosis and improved patient outcomes.
• Feature fusion techniques enhance diagnostic accuracy, highlighting the value of integrating multiple DTI-derived metrics.