In the present study, we verified the working hypothesis that cortical generators of resting state eyes-closed electroencephalographic (rsEEG) rhythms could classify with good performance (area under receiver operating characteristic, AUROC > 0.8) healthy elderly (Nold), Alzheimer's disease (AD) and Parkinson's disease (PD) individuals. rsEEG recordings were performed in 110 AD, 110 PD, and 75 Nold patients. Individual delta, theta, low-frequency alpha, and high-frequency bands were measured. Fixed low-frequency beta and high-frequency beta were also evaluated. eLORETA freeware measured cortical generators of rsEEG rhythms from posterior sources. Classification analysis was performed with the ROC curve. Cortical generators of posterior rsEEG rhythms at delta, theta, low-frequency alpha, high-frequency alpha, low-frequency beta, and high-frequency beta allowed a good performance (AUROC > 0.8) in the classification between healthy and neurodegenerative (i.e., AD and PD) individuals. Moreover, cortical generators of posterior rsEEG rhythms at low-frequency alpha permitted only a moderate performance (AUROC > 0.75) in the classification between AD and PD individuals. Future prospective cross-validation studies will have to test these candidate rsEEG markers for clinical applications and drug discovery.

Pent et al. (see ibid., vol.SAC-6, no.1, p.34-41, 1988) proposed a semianalytic method for performance evaluation of communication channels with a bandpass nonlinearity. This method is based on a decomposition of the composite signal at the output of the nonlinear device which leads to the definition of an equivalent nonlinearity. In Pent et al. such an equivalent model is evaluated via numerical integration. In the present comment it is shown that a suitable series expansion of the nonlinearity characteristic yields to an evaluation of the equivalent nonlinearity in closed form.>