Diffusion tensor imaging (DTI) Analysis Based on Tract-based spatial statistics (TBSS) and Classification Using...

Area of Science:

• Neuroimaging
• Machine Learning
• Neurology

Background:

• Alzheimer's disease (AD) involves cerebral cortex atrophy and neurofibrillary tangles.
• Early identification of high-risk individuals is crucial for timely intervention in AD.
• Diffusion tensor imaging (DTI) combined with machine learning can reveal unique anatomical patterns differentiating AD from healthy controls (HC).

Purpose of the Study:

• To investigate the utility of tract-based spatial statistics (TBSS) and machine learning in classifying Alzheimer's disease patients.
• To identify specific white matter fiber tracts affected in AD.
• To evaluate the effectiveness of integrating multiple diffusion metrics for improved classification accuracy.

Main Methods:

• Utilized DTI data from 37 AD patients and 36 healthy controls (HCs) from the Alzheimer's Disease Neuroimaging Initiative.
• Applied tract-based spatial statistics (TBSS) for white matter analysis.
• Employed support vector machine recursive feature elimination (SVM-RFE) for multi-metric classification.

Main Results:

• TBSS identified significant white matter damage in the corona radiata, corpus callosum, and superior longitudinal fasciculus in AD patients.
• The SVM-RFE method improved classification performance.
• Integrating fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD) effectively distinguished AD patients from HCs, with FA being the most significant metric.

Conclusions:

• TBSS and machine learning methods show promise in guiding clinical diagnosis of Alzheimer's disease.
• DTI-derived metrics, particularly FA, are valuable biomarkers for AD detection.
• The integration of multiple diffusion metrics enhances the ability to differentiate AD from healthy states.