Francisco J. Fraga

Improving Alzheimer's Disease Diagnosis with Machine Learning Techniques:

There is not a specific test to diagnose Alzheimers disease (AD). Its diagnosis should be based upon clinical history, neuropsychological and laboratory tests, neuroimaging and electroencephalography (EEG). Therefore, new approaches are necessary to enable earlier and more accurate diagnosis and to follow treatment results. In this study we used a Machine Learning (ML) technique, named Support Vector Machine (SVM), to search patterns in EEG epochs to differentiate AD patients from controls. As a result, we developed a quantitative EEG (qEEG) processing method for automatic differentiation of patients with AD from normal individuals, as a complement to the diagnosis of probable dementia. We studied EEGs from 19 normal subjects (14 females/5 males, mean age 71.6 years) and 16 probable mild to moderate symptoms AD patients (14 females/2 males, mean age 73.4 years. The results obtained from analysis of EEG epochs were accuracy 79.9% and sensitivity 83.2%. The analysis considering the diagnosis of each individual patient reached 87.0% accuracy and 91.7% sensitivity.

P300 latency and amplitude in Alzheimer's disease: a systematic review.

UNLABELLED The P300 plays a key role as a method for monitoring and evaluating dementia, including Alzheimers disease. OBJECTIVE The goal of this study was to search for articles which analyzed P300 latency and amplitude values in Alzheimers disease. METHODS We searched in the following databases: Web of Science, Pub Med, Psyc Info, Medline, Biological Abstracts and Scielo using the following keywords: speed of information processing, processing speed, information processing, aged, older, elderly, older people, alzheimer dementia, alzheimer disease, Alzheimer and cross-references of selected articles. RESULTS We found eight studies matching the inclusion criteria. These studies showed that there is a consensus on a P300 latency increase of elderly patients with Alzheimers disease compared with subjects without the disease. However, it appears that, with respect to the P300 amplitude, there is still no consensus; however, it may be related to different methodological variables adopted in the reviewed studies. CONCLUSION There is a need to standardize the variables involved in P300 measurement for senior citizens with Alzheimers disease in order to be able to compare P300 latency and amplitude values for this population.The P300 plays a key role as a method for monitoring and evaluating dementia, including Alzheimers disease. OBJECTIVE: The goal of this study was to search for articles which analyzed P300 latency and amplitude values in Alzheimers disease. METHODS: We searched in the following databases: Web of Science, Pub Med, Psyc Info, Medline, Biological Abstracts and Scielo using the following keywords: speed of information processing, processing speed, information processing, aged, older, elderly, older people, alzheimer dementia, alzheimer disease, Alzheimer and cross-references of selected articles. RESULTS: We found eight studies matching the inclusion criteria. These studies showed that there is a consensus on a P300 latency increase of elderly patients with Alzheimers disease compared with subjects without the disease. However, it appears that, with respect to the P300 amplitude, there is still no consensus; however, it may be related to different methodological variables adopted in the reviewed studies. CONCLUSION: There is a need to standardize the variables involved in P300 measurement for senior citizens with Alzheimers disease in order to be able to compare P300 latency and amplitude values for this population.

EEG amplitude modulation analysis for semi-automated diagnosis of Alzheimer’s disease

Recent experimental evidence has suggested a neuromodulatory deficit in Alzheimer’s disease (AD). In this paper, we present a new electroencephalogram (EEG) based metric to quantitatively characterize neuromodulatory activity. More specifically, the short-term EEG amplitude modulation rate-of-change (i.e., modulation frequency) is computed for five EEG subband signals. To test the performance of the proposed metric, a classification task was performed on a database of 32 participants partitioned into three groups of approximately equal size: healthy controls, patients diagnosed with mild AD, and those with moderate-to-severe AD. To gauge the benefits of the proposed metric, performance results were compared with those obtained using EEG spectral peak parameters which were recently shown to outperform other conventional EEG measures. Using a simple feature selection algorithm based on area-under-the-curve maximization and a support vector machine classifier, the proposed parameters resulted in accuracy gains, relative to spectral peak parameters, of 21.3% when discriminating between the three groups and by 50% when mild and moderate-to-severe groups were merged into one. The preliminary findings reported herein provide promising insights that automated tools may be developed to assist physicians in very early diagnosis of AD as well as provide researchers with a tool to automatically characterize cross-frequency interactions and their changes with disease.

EEG spectro-temporal modulation energy: A new feature for automated diagnosis of Alzheimer's disease

There is recent indication that Alzheimers disease (AD) can be characterized by atypical modulation of electrophysiological brain activity caused by fibrillar amyloid deposition in specific regions of the brain, such as those related to cognition and memory. In this paper, we propose to objectively characterize EEG sub-band modulation in an attempt to develop an automated noninvasive AD diagnostics tool. First, multi-channel full-band EEG signals are decomposed into five well-known frequency sub-bands: delta, theta, alpha, beta, and gamma. The temporal amplitude envelope of each sub-band is then computed via a Hilbert transformation. The proposed ‘spectro-temporal modulation energy’ feature measures the rate with which each sub-band is modulated. Modulation energy features are computed for 19 referential EEG signals and seven bipolar signals. Salient features are then selected and used to train four different classifiers, namely, support vector machines, logistic regression, classification and regression trees, and neural networks. Experiments with a database of 34 participants, 22 of which have been clinically diagnosed with probable-AD, show a neural network classifier achieving over 91% accuracy, thus significantly outperforming a classifier trained with conventional spectral-based features.

Clinician’s Road Map to Wavelet EEG as an Alzheimer’s disease Biomarker:

Alzheimer’s disease (AD) is considered the main cause of dementia in Western countries. Consequently, there is a need for an accurate, universal, specific and cost-effective biomarker for early AD diagnosis, to follow disease progression and therapy response. This article describes a new diagnostic approach to quantitative electroencephalogram (QEEG) diagnosis of mild and moderate AD. The data set used in this study was composed of EEG signals recorded from 2 groups: (S1) 74 normal subjects, 33 females and 41 males (mean age 67 years, standard deviation = 8) and (S2) 88 probable AD patients (NINCDS-ADRDA criteria), 55 females and 33 males (mean age 74.7 years, standard deviation = 7.8) with mild to moderate symptoms (DSM-IV-TR). Attention is given to sample size and the use of state of the art open source tools (LetsWave and WEKA) to process the EEG data. This innovative technique consists in associating Morlet wavelet filter with a support vector machine technique. A total of 111 EEG features (attributes) were obtained for 162 probands. The results were accuracy of 92.72% and area under the curve of 0.92 (percentage split test). Most important, comparing a single patient versus the total data set resulted in accuracy of 84.56% (leave-one-patient-out test). Particular emphasis was on clinical diagnosis and feasibility of implementation of this low-cost procedure, because programming knowledge is not required. Consequently, this new method can be useful to support AD diagnosis in resource-limited settings.

Does EEG Montage Influence Alzheimer's Disease Electroclinic Diagnosis?

There is not a specific Alzheimers disease (AD) diagnostic test. AD diagnosis relies on clinical history, neuropsychological, and laboratory tests, neuroimaging and electroencephalography. Therefore, new approaches are necessary to enable earlier and more accurate diagnosis and to measure treatment results. Quantitative EEG (qEEG) can be used as a diagnostic tool in selected cases. The aim of this study was to answer if distinct electrode montages have different sensitivity when differentiating controls from AD patients. We analyzed EEG spectral peaks (delta, theta, alpha, beta, and gamma bands), and we compared references (Biauricular, Longitudinal bipolar, Crossed bipolar, Counterpart bipolar, and Cz reference). Support Vector Machines and Logistic Regression classifiers showed Counterpart bipolar montage as the most sensitive electrode combination. Our results suggest that Counterpart bipolar montage is the best choice to study EEG spectral peaks of controls versus AD.

EEG epochs with less alpha rhythm improve discrimination of mild Alzheimer's

BACKGROUND AND OBJECTIVE Eyes-closed-awake electroencephalogram (EEG) is a useful tool in the diagnosis of Alzheimers. However, there is eyes-closed-awake EEG with dominant or rare alpha rhythm. In this paper, we show that random selection of EEG epochs disregarding the alpha rhythm will lead to bias concerning EEG-based Alzheimers Disease diagnosis. METHODS We compared EEG epochs with more than 30% and with less than 30% alpha rhythm of mild Alzheimers Disease patients and healthy elderly. We classified epochs as dominant alpha scenario and rare alpha scenario according to alpha rhythm (8-13 Hz) percentage in O1, O2 and Oz channels. Accordingly, we divided the probands into four groups: 17 dominant alpha scenario controls, 15 mild Alzheimers patients with dominant alpha scenario epochs, 12 rare alpha scenario healthy elderly and 15 mild Alzheimers Disease patients with rare alpha scenario epochs. We looked for group differences using one-way ANOVA tests followed by post-hoc multiple comparisons (p < 0.05) over normalized energy values (%) on the other four well-known frequency bands (delta, theta, beta and gamma) using two different electrode configurations (parieto-occipital and central). RESULTS After carrying out post-hoc multiple comparisons, for both electrode configurations we found significant differences between mild Alzheimers patients and healthy elderly on beta- and theta-energy (%) only for the rare alpha scenario. No differences were found for the dominant alpha scenario in any of the five frequency bands. CONCLUSIONS This is the first study of Alzheimers awake-EEG reporting the influence of alpha rhythm on epoch selection, where our results revealed that, contrarily to what was most likely expected, less synchronized EEG epochs (rare alpha scenario) better discriminated mild Alzheimers than those presenting abundant alpha (dominant alpha scenario). In addition, we find out that epoch selection is a very sensitive issue in qEEG research. Consequently, for Alzheimers studies dealing with resting state EEG, we propose that epoch selection strategies should always be cautiously designed and thoroughly explained.

Event-related synchronisation responses to N-back memory tasks discriminate between healthy ageing, mild cognitive impairment, and mild Alzheimer's disease

In this study we investigate whether or not event-related (de)synchronisation (ERD/ERS) can be used to differentiate between 27 healthy elderly, 21 subjects diagnosed with amnestic mild cognitive impairment (aMCI) and 16 mild Alzheimers disease (AD) patients. Using 32-channel EEG recordings, we measured ERD responses to a three-level visual N-back task (N = 0, 1, 2) on the well-known delta, theta, alpha, beta and gamma bands. Our findings revealed that healthy elderly (HE) elicited consistently greater beta and alpha ERD responses than MCI and AD patients at many scalp electrodes, most of them located at fronto-central and temporal-parietal areas. Additionally, significant ERD differences were found on the gamma band in the MCI vs. AD comparison. Based on these findings, we conclude that ERD responses to a working memory (N-back) task could be useful for early MCI diagnosis or for improved AD diagnosis, and also for assessing the likelihood of MCI progression to AD.

Towards automated electroencephalography-based Alzheimer's disease diagnosis using portable low-density devices

Abstract Today, Alzheimer’s disease (AD) diagnosis is carried out using subjective mental status examinations assisted in research by scarce and expensive neuroimaging scans and invasive laboratory tests; all of which render the diagnosis time-consuming, geographically confined and costly. Driven by these limitations, quantitative analysis of electroencephalography (EEG) has been proposed as a non-invasive and more convenient technique to study AD. Published works on EEG-based AD diagnosis typically share two main characteristics: EEG is manually selected by experienced clinicians to discard artefacts that affect AD diagnosis, and reliance on EEG devices with 20 or more electrodes. Recent work, however, has suggested promising results by using automated artefact removal (AAR) algorithms combined with medium-density EEG setups. Over the last couple of years, however, low-density, portable EEG devices have emerged, thus opening the doors for low-cost AD diagnosis in low-income countries and remote regions, such as the Canadian Arctic. Unfortunately, the performance of automated diagnostic solutions based on low-density portable devices is still unknown. The work presented here aims to fill this gap. We propose an automated EEG-based AD diagnosis system based on AAR and a low-density (7-channel) EEG setup. EEG data was acquired during resting-awake protocol from control and AD participants. After AAR, common EEG features, spectral power and coherence, are computed along with the recently proposed amplitude-modulation features. The obtained features are used for training and testing of the proposed diagnosis system. We report and discuss the results obtained with such system and compare the obtained performance with results published in the literature using higher-density EEG layouts.

Comparing Methods for Determining Motor-Hand Lateralization Based on fTCD Signals

The lateralization index (LI) as determined from functional transcranial Doppler sonography (fTCD) can be used to determine the hemispheric organization of neural activation during a behavioral task. Previous studies have proposed different methods to determine this index, but to our knowledge no studies have compared the performance of these methods. In this study, we compare two established methods with a simpler method proposed here. The aim was to see whether similar results could be achieved with a simpler method and to give an indication of the analysis steps required to determine the LI. A simple unimanual motor task was performed while fTCD was acquired, and the LI determined by each of these methods was compared. In addition, LI determined by each method was related to behavioural output in the form of degree of handedness. The results suggest that although the methods differed in complexity, they yielded similar results when determining the lateralization of motor functions, and its correlation with behavior. Further investigation is needed to expand the conclusions of this preliminary study, however the new method proposed in the paper has great potential as it is much simpler than the more established methods yet yields similar results.