Feature extraction for snore sound via neural network processing

Snore sound (SS) is the earliest and the most common symptom of Obstructive Sleep Apnea (OSA) which is a serious disease caused by the collapse of upper airways during sleep. SS should carry vital information on the state of the upper airways and is simple to acquire and rich in features but their analysis is complicated. In this study we use neural network (NN) based method to model SS via a simple second order one-step predictor. We show that the some hidden information/feature of a SS can be conveniently captured in the connection-weight-space (CWS) of the NN, after a process of supervised training. The availability of the proposed method is investigated by performing independent component analysis (ICA) on CWS.

Recognition mechanism of a neural network for character recognition

We have developed a small scale four-layered neural network (NN) model for simple character recognition, which can recognize the patterns transformed by affine conversion. In this study 24 patterns are presented as input patterns. An input pattern is divided into 64 local patterns and connected with the 1st hidden layer. After the training, we investigated the recognition mechanism of NN using Alopex algorithm. Effectiveness of this method is demonstrated.

Systematic identification for inert region of a brain from EEG

Evaluation of biomedical signals is important in the diagnosis of neurology diseases, such as dementia, in neurology through the use of electroencephalograms (EEG). While automated techniques exist for EEG analysis, it is likely that additional information can be extracted from EEG signal through the use of new methods. We describe a method for identifying inert region from EEG. This method uses EEG as input to an artificial neural network with five outputs: activity in whole regoin and inert region separated by four regions.

Tracking the states of a nonlinear and nonstationary system in the weight-space of artificial neural networks

We propose a novel interpretation and usage of Neural Network (NN) in modeling physiological signals, which are allowed to be nonlinear and/or nonstationary. The method consists of training a NN for the k-step prediction of a physiological signal, and then examining the connection-weight-space (CWS) of the NN to extract information about the signal generator mechanism. We define a novel feature, Normalized Vector Separation (gamma(ij)), to measure the separation of two arbitrary states "i" and "j" in the CWS and use it to track the state changes of the generating system. The performance of the method is examined via synthetic signals and clinical EEG. Synthetic data indicates that gamma(ij) can track the system down to a SNR of 3.5 dB. Clinical data obtained from three patients undergoing carotid endarterectomy of the brain showed that EEG could be modeled (within a root-means-squared-error of 0.01) by the proposed method, and the blood perfusion state of the brain could be monitored via gamma(ij), with small NNs having no more than 21 connection weight altogether.

Dipole source localization of MEG by BP neural networks

The purpose of this study was to examine the usefulness of BP neural network for source localization of MEG. Since the performance of this method does not depend on the complexity of brain parameters and source models, a homogeneous brain model and a single current dipole source are assumed for convenience. Localization accuracy was examined in relation to the configuration and scale of the network. As a result, average error for position and moment estimations was within 2% while the maximum error was about 5%. It was therefore concluded that the neural network method, was useful for MEG source localization, though some improvements are still necessary.