PET Neuroimaging Enhancements for Improved Mild Cognitive Impairment Detection

Abstract

Mild Cognitive Impairment (MCI) is a neurode-generative condition characterized by cognitive decline that may progress to Alzheimer’s disease. Modern neuroimaging techniques, particularly Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET), serve as essential diagnostic tools by providing detailed structural and metabolic information of the brain tissue. Fluorodeoxyglucose PET (FDG-PET) has emerged as one of the most sensitive early indicators of MCI due to its ability to quantify glucose metabolism in specific regions of the brain. Patients with MCI exhibit characteristic patterns of reduced metabolic activity in distinct brain regions, which can be detected through automated analysis.Recent advances in Artificial Intelligence (AI) and Deep Learning (DL) have revolutionized medical image analysis through their superior pattern recognition capabilities. State-of-the-art (SOTA) image classification methods, including Convolutional Neural Networks (CNN) and Vision Transformers (ViT), demonstrate remarkable precision in distinguishing MCI patients from Cognitively Normal (CN) individuals. This study presents a novel methodology that uses FDG-PET as the primary diagnostic modality, incorporating histogram equalization technique, noise reduction and advanced augmentation pipeline. When combined with SOTA image classifiers, our approach achieves a 7% improvement in diagnostic accuracy over the baseline, highlighting how neuroimaging enhancement techniques can improve the already robust diagnostic capabilities of DL image classifiers.