Search

The aim of fetal monitoring is to identify fetuses at risk of an adverse outcome based on the ability to understand how the fetus reacts to stress before it becomes compromised. Recent works have shown that the ST waveform of the fetal electrocardiogram (fECG) provides continuous information on the capacity of the fetal heart muscle to respond to the stress of labour. The fECG can be ob- tained by a non-invasive technique that consists in collecting elec- trical signals by some sensors on the body of the mother. Unfortu- nately the fetal heartbeat signal yielded by this recording technique is quite weaker than the mother heartbeat signal, also due to the atten- uation during the propagation caused by the tissues; moreover, many other artifacts are superimposed to the two heartbeats. In this paper we propose a method based on wavelet decomposition and indepen- dent component analysis in order to approximate the real shape of the fECG. We also try to evaluate the ST waveform for fetal monitoring.

Visual Evoked Potentials (VEPs) are referred to electrical potentials due to brief visual stimuli which can be recorded from scalp overlying visual cortex. A way to measure VEPs is through encephalogram (EEG). VEPs are very important because they can quantify functional integrity of the optic pathway. Their study allows to detect abnormalities that affect the visual pathways or visual cortex in the brain, and so methods that permit to identify VEPs components in EEG signals must be defined. However, the background activity measured from EEG hides VEPs components because they have a low voltage. So it is necessary to define a robust method to extract features, which best describe these potentials of interest. In this work Empirical Mode Decomposition (EMD) method is used to separate the EEG components and to detect VEPs. EMD decomposes a signal into components named Intrinsic Mode Functions (IFM). The results, obtained from the study of EEG records of a normal person, suggest that IMFs may be used to determine VEPs in EEG and to obtain important information related to brain activity by a time and frequency analysis of IMF components. It is well comparable with the known Wavelet Transform method, but it is characterized from a greater simplicity of implementation because the basis used in the analysis is generated by the same analyzed signal.

A neural network model is proposed to treat inverse problems in electromagnetics, which includes wavelet functions to improve local approximation capabilities. This processor couples the advantages of an interpretation of the problem based on "features" to the accuracy derived from using wavelets where local corrections are needed. The combined model allows one to cope with singularities of the mapping and to slightly modify the mapping in real time. The detection and characterization of a circular defect in a conducting plate by using eddy current testing is shown to take advantage from the proposed approach in a test case, when unforeseen disturbances are present.

The application of a rotating field like that of induction machines is proposed for the eddy current testing of conducting cylindrical structures. This method allows for fast tracking of the position and geometrical parameters of cracks. In order to improve the capability to discriminate the depth of buried cracks, a suitable range of frequencies is considered. The measurement data are processed with a hybrid neuro-fuzzy system. Two artificial neural networks are used, one is specialized for surface cracks and the other one for buried cracks. The fuzzy system is used to perform the switching strategy between the two networks.