Computer-Aided Diagnosis System for Alzheimer's Disease Using Different Discrete Transform Techniques.

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Computational Neuroscience

Background:

• Alzheimer's disease (AD) diagnosis relies on identifying significant biomarkers.
• Feature extraction techniques are crucial for developing effective computer-aided diagnosis (CAD) systems.
• Discrete transform techniques like DCT, DST, and DWT are established methods for feature extraction.

Purpose of the Study:

• To propose and evaluate a CAD system for Alzheimer's disease (AD) recognition.
• To compare the effectiveness of various discrete transform techniques (DCT, DST, DWT) and MFCCs for feature extraction in AD detection.
• To identify the most significant features for improved AD diagnosis performance.

Main Methods:

• Utilized discrete cosine transform (DCT), discrete sine transform (DST), discrete wavelet transform (DWT), and mel-scale frequency cepstral coefficients (MFCCs) for feature extraction.
• Developed a computer-aided diagnosis (CAD) system incorporating these feature extraction methods.
• Employed a linear support vector machine (SVM) as the classification algorithm.

Main Results:

• The proposed CAD system demonstrated superior performance when utilizing MFCCs for AD recognition.
• MFCCs enabled the extraction of highly significant features, leading to improved system performance.
• The MFCC-based CAD system outperformed systems using DCT, DST, DWT, and hybrid transform combinations.

Conclusions:

• Mel-frequency cepstral coefficients (MFCCs) are highly effective for feature extraction in Alzheimer's disease (AD) detection.
• The proposed CAD system with MFCCs offers a promising approach for accurate and efficient AD diagnosis.
• Further research can explore MFCCs in conjunction with advanced machine learning models for enhanced diagnostic capabilities.