Functional Biomedical Images of Alzheimer's Disease. A Green's Function-based Empirical Mode Decomposition Study

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Positron emission tomography (PET) is crucial for detecting dementia in human brains.
• Two-dimensional empirical mode decomposition (2D-EMD) analyzes PET images by extracting texture features at various spatial scales.
• These texture features are vital for subsequent classification tasks in disease diagnosis.

Purpose of the Study:

• To develop a novel, computationally efficient variant of 2D-EMD for analyzing PET images.
• To enhance the speed and stability of the EMD process for neuroimaging applications.
• To explore the potential of this new method in computer-aided diagnosis (CAD) systems for dementias.

Main Methods:

• Proposed a new 2D-EMD variant utilizing a Green's function-based estimation with a tension parameter.
• Developed a method for fast and reliable estimation of envelope hypersurfaces from 2D image intensity distributions.
• Implemented and validated the method on PET images from Alzheimer's disease patients.

Main Results:

• The new bi-dimensional EMD method significantly accelerates computations, achieving approximately a 100-fold speed-up.
• Demonstrated the method's stability and reliability in processing PET image data.
• Showcased the potential of extracted features for classifying diseases like Alzheimer's.

Conclusions:

• The developed 2D-EMD variant offers a substantial computational advantage for analyzing functional brain images.
• This technique, combined with classifiers, can form effective CAD systems for early disease detection.
• The study highlights the utility of advanced image analysis for diagnosing neurological conditions using PET.