Toward Early Alzheimer’s Disease Diagnosis: Development of an Integrated EEG-fNIRS System and EEG-Based Evaluation of P300 Latency as a Physiological Marker

Abstract

The primary focus of this dissertation is the neurophysiological analysis of event-related potentials (ERPs), with particular emphasis on P300 latency, to identify reliable biomarkers for the early detection of Alzheimer’s Disease (AD). By investigating electrophysiological signals associated with early cognitive decline, this research aims to distinguish the preclinical stages of AD. Based on the analytical requirements and insights derived from this investigation, we subsequently developed and validated a novel multimodal brain monitoring system that integrates electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) to enhance detection sensitivity and spatial resolution. Chapter 1 presents a comprehensive overview of Alzheimer’s Disease (AD), beginning with the increasing global burden of the disease and the pressing need for early detection during its asymptomatic and prodromal AD stages. It outlines the clinical progression of AD, from healthy aging through the preclinical and mild cognitive impairment phases to overt dementia, highlighting key pathological hallmarks such as amyloid-beta accumulation, tau neurofibrillary tangles, neuroinflammation, synaptic dysfunction, and structural and functional brain alterations. The chapter critically evaluates existing diagnostic approaches, including the AT(N) framework and other biomarker-based criteria, while emphasizing the limitations of conventional neuroimaging and cerebrospinal fluid analyses in early-stage detection due to their invasiveness, cost, and limited accessibility. As a response, it introduces electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) as non-invasive, cost-effective alternatives capable of capturing complementary neurophysiological information. Special focus is placed on the event-related P300 component, elicited via the oddball paradigm, due to its sensitivity to attention and working memory processes that are often compromised early in the AD continuum. This rationale supports the need for integrated, time-efficient, and scalable brain monitoring tools that can serve as early biomarkers and enhance both clinical screening and public health intervention strategies. Chapter 2 investigates P300 ERP indices—specifically peak latency and amplitude—as potential biomarkers for distinguishing between healthy controls (HC), individuals with asymptomatic AD (AAD), and those with prodromal AD (PAD). EEG data were collected during a visual oddball task from 79 participants. Findings reveal that P300 peak latency, but not amplitude, significantly differentiates PAD from both AAD and HC. Moreover, P300 latency correlates significantly with memory domain scores, reinforcing its role as a neurophysiological marker of early cognitive impairment. This chapter focuses exclusively on EEG-based analysis, as fNIRS-related investigations are being pursued by other research groups. The integration of EEG and fNIRS data remains a direction for future work. Chapter 3 extends the investigation to a larger cohort of 117 participants and explores the topographical distribution of P300 latency across the left, middle, and right brain regions. Results show that P300 latency from the left hemisphere distinguishes HC from AAD, while latency from all regions separates HC from PAD. Receiver Operating Characteristic (ROC) analyses confirm the diagnostic potential of P300 latency with favorable sensitivity and specificity, particularly in the left and middle regions. However, differentiation between AAD and PAD remains limited, suggesting a need for complementary multimodal or longitudinal approaches. Chapter 4 details the design and implementation of a portable, low-cost EEG/fNIRS brain function monitoring system. The system incorporates the ADS1298IPAG analog front-end and a Teensy 3.2 microcontroller, allowing for simultaneous acquisition of two-channel EEG and six-channel fNIRS signals. The system’s hardware, software, and graphical user interface (GUI) are described, and performance validation confirms its suitability for real-time, dual-modal brain monitoring. Chapter 5 concludes by summarizing the successful development of a low-cost, integrated EEG/fNIRS system and the identification of P300 latency as a viable physiological marker for early AD screening. This dissertation contributes a scalable, accessible neurodiagnostic platform that bridges engineering innovation with clinical neuroscience. Future directions include system miniaturization, the integration of EEG and fNIRS data for Alzheimer's Disease classification, the incorporation of machine learning models to enhance diagnostic accuracy, and validation across larger and more diverse populations. Together, this work advances biomedical engineering by delivering a novel brain monitoring solution that is both physiologically informative and practically applicable for early Alzheimer’s Disease detection.