Computer-Aided Diagnosis System for Alzheimer's Disease Using Positron Emission Tomography Images

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is an irreversible neurodegenerative disorder with no cure.
• Early diagnosis is crucial for effective patient management and treatment.
• Analyzing brain scans for accurate AD detection presents significant challenges.

Purpose of the Study:

• To design a Computer Aided Diagnosis (CAD) system utilizing Long-Term Short Memory (LSTM) for improved Alzheimer's disease classification.
• To identify key attributes that effectively differentiate AD patients from Healthy Control (HC) subjects.
• To enhance the accuracy of brain image analysis for neurological disorders.

Main Methods:

• Preprocessing 3D PET images and converting them into 2D image subsets.
• Extracting combined feature vectors, including statistical, Gray Level Co-occurrence Matrix, and wavelet energy features.
• Employing a Long-Term Short Memory (LSTM) network for classifying PET brain images into AD and HC groups.

Main Results:

• The developed CAD system achieved a classification accuracy of 98.9% using combined features.
• Validation was performed on 18FDG-PET images from 188 subjects (105 HC, 83 AD) from the ADNI database.
• The system demonstrated outstanding performance in differentiating AD from HC subjects.

Conclusions:

• The LSTM-based CAD system shows high efficacy for accurate Alzheimer's disease detection from PET images.
• Feature extraction and combination are vital for improving classification performance.
• This AI-driven approach offers a promising tool for early and accurate diagnosis of Alzheimer's disease.