The Electroencephalogram (EEG) can be used as a Biomarker for Dementia. This decades-old technology is finding new life in modern neuroscience.
Old Dog
Introduced to neurologists for the detection and classification of epilepsy by Frederick Gibbs’ in 1934, electroencephalography (EEG) has played an integral role in the history of neurological medicine.
Not even the introduction of Computed Tomography (CT), Magnetic Resonance Imaging (MRI), magnetoencephalography (MEG), and other neuroimaging technologies could replace EEG as a low-cost, noninvasive tool for evaluating cortical function.
Presently, the role of EEG remains limited by the relatively low spatial resolution. However, the superior temporal resolution made EEG an indispensable device in clinical neurology. EEG reading physicians with trained eyes can interpret subtle deviations of electrographic signals from normal brain activity. However, oftentimes these electrographic events cooccur with overt behavioral symptoms that seem to render EEG unnecessary. Nevertheless, these behavioral manifestations can be misleading. For instance, classifying psychogenic seizures, which do not have underlying abnormal brain activity, from epileptic seizures is primary importance for the subsequent treatment plan of the patient. This distinction can only be possible based on EEG evidence or the lack thereof.
Even so, interest in EEG has exploded over the last two decades. The recent leaps forward in computing power and advancements in machine learning have begun to change the practice of medicine. In neuroscience, attention has turned toward overcoming the limitation of the poor spatial resolution of EEG.
There are two streams of advancements in EEG data augmentation and analysis. One is to improve source localization and the other is to utilize the EEG to find biomarkers of diseases using Machine-learning algorithms. What makes the latter approach viable today is the aggregation of large amounts of EEG data on a scale previously impossible. The special characteristics that have kept EEG relevant for so long – non-invasive and inexpensive – have become even more compelling.
New Tricks
The first practical applications for these augmented EEG data examinations will most likely involve using EEG to locate biomarkers–tools to facilitate the early diagnosis of diseases, dementias especially. A biomarker is a piece of measurable and objective biological data used to diagnose or stage an illness. Presently, the most frequently used EEG detectible dementia-related biomarkers include those used to identify Alzheimer’s disease (AD).
Alzheimer’s disease biomarkers, identified using EEG, improve the objectivity to the detection and prognosis of AD. That objectivity can provide the much-needed confidence for physicians, and presentable evidence to the patients and families under evaluation. Researchers classify AD biomarkers in the following categories:
- Biochemical,
- Radiologic,
- Genetic,
- Neurophysiological.
Both biochemical and radiologic biomarkers have found their way onto AD diagnostic criteria.
Biochemical markers incorporated in AD work-ups include CSF analysis checking for the presence of amyloid, total tau, and hyperphosphorylation tau proteins. AD diagnostic evaluations frequently include radiologic biomarkers such as the atrophic findings on structural MRI or metabolic changes on fluorodeoxyglucose (FDG)-PET. Genetic markers have yet to merit inclusion as independent AD criteria. Genetic biomarkers, especially those focused on specific alleles of apolipoprotein E, have shown some promise as supporting markers.
Computer augmented EEG data techniques have been the focus of research attempting to establish neurophysiological biomarkers to add a function-based evaluation to AD diagnosis. Leveraging increased computing power has allowed denoising, improved localization, and non-linear analysis. These advanced EEG techniques now assist in Alzheimer’s disease and dementia diagnosis during preclinical stages. Candidates for EEG-derived biomarkers of Alzheimer’s disease include both linear-spectral and non-linear dynamic features. Linear features of AD typically include:
- Slowing of alpha power,
- Increase in delta power,
- Theta power is higher in patients with vascular dementia compared to AD,
- Zero crossing interval (ZCI) increased in slow activity associated with dementia.
Non-linear features may include:
- lower fractal dimension (FD) of the EEG
- lower Lempel-Ziv-Welch (LZW) complexity,
- and lower Tsallis entropy (TsEn)
- as compared to normal controls
Conclusion
The work of laying the foundation for EEG biomarkers in the diagnosis of Alzheimer’s disease and dementia continues. Neuroscientists continue decoding how functional changes in the dementia-impacted brain reveal themselves in EEG data. The list of EEG biomarkers will narrow and expand as our understanding improves. Ultimately, EEG biomarkers will sharpen our ability to diagnose and triage different types of dementia at an earlier phase of the disease than we do today. Utilizing EEG to detect early markers for Alzheimer’s disease and dementias may open the door to new therapeutic possibilities or create the opportunity for patients and their families to plan ahead. For diligent neurologists, the ongoing pursuit of EEG biomarkers will require sustained attention. The future has begun to arrive.