Farid Hassainia

Quantitative EEG and Statistical Mapping of Wakefulness and REM Sleep in the Evaluation of Mild to Moderate Alzheimer’s Disease

Statistical probability mapping was used to quantify and localize EEG differences between 27 patients with Alzheimers disease (AD) and 25 age- and gener-matched controls. Differences in mean activity in four EEG frequency bands (delta, theta, alpha, beta) for wakefulness and for REM sleep were examined, t-statistic maps clearly highlighted common pattern anomalies in AD patients in the two states. More specifically, Alzheimer patients were more affected than control subjects in parieto-temporal and frontal regions. These differences were more prominent in REM sleep and consisted primarily in an increase in absolute delta and theta activities, and a decrease in absolute alpha and beta activities. Discriminant analysis, using a ratio of slow over fast frequencies, yielded a classification rate of 90.4% (sensitivity 81.5%, specificity 100%) for REM sleep. For wakefulness, the same measure allowed correct classification of 80.8% of the subjects (sensitivity 66.7%, specificity 96%).

Significance probability mapping: the final touch in t-statistic mapping.

SummarySignificance Probability Mapping (SPM), based on Studentst-statistic, is widely used for comparing mean brain topography maps of two groups. The map resulting from this process represents the distribution oft-values over the entire scalp. However,t-values by themselves cannot reveal whether or not group differences are significant. Significance levels associated with a fewt-values are therefore commonly indicated on map legends to give the reader an idea of the significance levels oft-values. Nevertheless, a precise significance level topography cannot be achieved with these few significance values. We introduce a new kind of map which directly displays significance level topography in order to relieve the reader from converting multiplet-values to their corresponding significance probabilities, and to obtain a good quantification and a better localization of regions with significant differences between groups. As an illustration of this type of map, we present a comparison of EEG activity in Alzheimers patients and age-matched control subjects for both wakefulness and REM sleep.

Three dimensional representation of brain electrical activity

SummaryBrain topography mapping is a useful technique for the representation of electrical activity recorded on the scalp. It clarifies spatial and temporal relationships between different cortical areas. In this work we propose a system which includes several enhancements over those previously proposed, such as an optimised interpolation method and a three dimensional reconstruction of maps. This system is available in a personal computer environment. Results clearly show a superiority of the 3D representation over 2D maps obtained with different projections. The performance of this system in terms of speed and precision is comparable to that of dedicated image processing and image synthesis workstations proposed for brain mapping.

Topographical study of the heterogeneity of impairments in early Alzheimer's disease patients

The purpose of the present study was to demonstrate the power of significance probability mapping in localizing EEG slowing in Alzheimers Disease (AD) patients relative to a control group. The t-statistic mapping was used to highlight the common pattern changes in AD patients, and the z-statistic mapping served to show the diversity of impairments in these patients.

Orthogonal polynomial trend analysis helps interpreting abnormalities in event-related brain potentials in high functioning autism

Orthogonal polynomial trend analysis allows the extraction of event-related brain activity on basis of a few stimulus presentations enabling the study of time on task changes in event-related brain activity. The impact of this information on the interpretation of conventionally averaged event-related potentials is illustrated by comparing a high functioning autism group to matched controls.

Nonparametrical statistical mapping

Statistical mapping is widely used in electrophysiological research. T-statistic mapping is commonly used in between-group map comparison. The t-map, which is based upon the equal variance of groups, was the first type of statistical mapping to be used in this field. Later, a t-map, and a p-map using the unequal variance hypothesis were introduced. Although, the latter is more adequate for comparisons where the variance of both groups cannot be assumed to be equal. Its use is not appropriate in the case where the respective distributions of the populations are not normal. To face this situation, we introduce a new kind of statistical mapping based upon nonparametric tests. This type of mapping corresponds to a distribution-free test in the sense that the population distribution is not assumed to be normal. This technique therefore increases the degree of freedom of the researcher when comparing groups with a non-normal, or more generally an unknown distribution.

The use of F-statistic mapping as a complementary tool to t-statistic mapping in group comparisons

The purpose of the present paper was to introduce a new kind of brain mapping displaying significance probability level associated to Fishers test. The latter measures the degree of difference between variances of two groups. This map can be considered as complementary to that representing the between-group mean differences. The interpretation of data is enhanced by the joint use of both maps.

Statistical mapping in the evaluation of right anatomical hemispherectomized patients in REM sleep

The goal of the present study was to quantify and localize brain electrical activity differences in REM sleep between right anatomical hemispherectomized patients and a matched control group. Z-statistic maps showed a significant increase in fast activities bands over the intact hemisphere.