This paper proposes an automatic method for extraction of useful information content from time-frequency distributions of nonstationary signals heavily corrupted by additive noise. The proposed method, which does not require prior knowledge of the signal, initially performs a one-dimensional clustering of the time-frequency distribution aimed at segmenting it into a fixed number of classes. This procedure points out basic structural differences of noise and signal components (i.e., useful information) in the time-frequency plane. In fact, noise presents a flat spectrum with large time-frequency supports, while signal components are narrow energy ridges. The time-frequency supports are estimated for each of the obtained classes and used as input for an algorithm aimed at discriminating “noise” affine classes, containing mainly non-zero coefficients located outside the components time-frequency supports, from “useful information” affine classes, containing mainly coefficients inside time-frequency supports of the signal components. Simulations show the superiority of the proposed algorithm, mainly in terms of error rate reduction of classified time-frequency coefficients, compared to hard amplitude thresholding methods, as well as to a recently proposed method for information content extraction based on adaptive thresholding, which is also outperformed in terms of computational complexity and execution time (by up to 43 and 28 times, respectively).

This paper approaches the problem of signal denoising in time-variable noise conditions. Non-stationary noise results in variable degradation of the signal’s useful information content over time. In order to maximize the correct recovery of the useful part of the signal, this paper proposes a denoising method that uses a criterion based on amplitude segmentation and local Rényi entropy estimation which are limited over short time blocks of the signal spectrogram. Local estimation of the signal features reduces the denoising problem to the stationary noise case. Results, presented for synthetic and real data, show consistently better performance gained by the proposed adaptive method compared to denoising driven by global criteria.

Medical electrodes used for measuring low amplitude signals, such as EEG electrodes, have to be robust and guarantee a high level of reliability. Corkscrew electrodes, considered in this paper, can become faulty due to cold solder that can appear immediately after the manufacturing process or due to mechanical stress after a few months of use. This problem is hard to detect and is usually manifested as noisy output signal. Commonly used method for monitoring the reliability of materials or circuit interconnects is the resistance measurement. Although very easy to implement, this method does not provide a reliable failure detection. Motivated by these facts, in this paper we propose a computer model based on resistance measurements, for predicting and detecting failure in EEG electrodes supported by laboratory measurements. Level and type of noise is obtained from the comparison of resistance fluctuations of the electrodes tip recorded under stress, and simulated signals. Time-frequency analysis has been applied to real and simulated reference and faulty electrode signals and results compared in order to establish a failure detection measure. Since the energy spectrum of the signal is shown to be an unreliable indicator of the failure appearance, the Rényi entropy is used to determine the difference between reference and faulty electrodes. This measure is applied to measured and simulated spectrograms, denoised using the K-means algorithm. It is shown that the difference between global entropies of the reference and faulty electrode spectrograms is significant when K-means based denoising is applied, thus providing a method for reliable failure detection.

Modelling and detection of failure in medical electrodes

We present a numerical study on resistance fluctuations in a series of nanowire‐based grids. Each grid is made of GaAs nanowires arranged in parallel with metallic contacts crossing all nanowires perpendicularly. Electrical properties of GaAs nanowires known from previous experimental research are used as input parameters in the simulation procedure. Due to the nonhomogeneous doping, the resistivity changes along nanowire. Allowing two possible nanowire orientations (“upwards” or “downwards”), the resulting grid is partially disordered in vertical direction which causes resistance fluctuations. The system is modeled using a two‐dimensional random resistor network. Transfer‐matrix computation algorithm is used to calculate the total network resistance. It is found that probability density function (PDF) of resistance fluctuations for a series of nanowire grids changes from Gaussian behavior towards the Bramwell‐Holdsworth‐Pinton distribution when both nanowire orientations are equally represented in the grid.