Christos N. Schizas

Neural network models in EMG diagnosis

In previous years, several computer-aided quantitative motor unit action potential (MUAP) techniques were reported. It is now possible to add to these techniques the capability of automated medical diagnosis so that all data can be processed in an integrated environment. In this study, the parametric pattern recognition (PPR) algorithm that facilitates automatic MUAP feature extraction and Artificial Neural Network (ANN) models are combined for providing an integrated system for the diagnosis of neuromuscular disorders. Two paradigms of learning for training ANN models were investigated, supervised, and unsupervised. For supervised learning, the back-propagation algorithm and for unsupervised learning, the Kohonens self-organizing feature maps algorithm were used. The diagnostic yield for models trained with both procedures was similar and on the order of 80%. However, back propagation models required considerably more computational effort compared to the Kohonens self-organizing feature map models. Poorer diagnostic performance was obtained when the K-means nearest neighbor clustering algorithm was applied on the same set of data. >

Computer-aided detection of breast cancer nuclei

A computer-aided detection system for tissue cell nuclei in histological sections is introduced and validated as part of the Biopsy Analysis Support System (BASS). Cell nuclei are selectively stained with monoclonal antibodies, such as the anti-estrogen receptor antibodies, which are widely applied as part of assessing patient prognosis in breast cancer. The detection system uses a receptive field filter to enhance negatively and positively stained cell nuclei and a squashing function to label each pixel value as belonging to the background or a nucleus. In this study, the detection system assessed all biopsies in an automated fashion. Detection and classification of individual nuclei as well as biopsy grading performance was shown to be promising as compared to that of two experts. Sensitivity and positive predictive value were measured to be 83% and 67.4%, respectively. One major advantage of BASS stems from the fact that the system simulates the assessment procedures routinely employed by human experts; thus it can be used as an additional independent expert. Moreover, the system allows the efficient accumulation of data from large numbers of nuclei in a short time span. Therefore, the potential for accurate quantitative assessments is increased and a platform for more standardized evaluations is provided.

Artificial neural networks for modelling the starting-up of a solar steam-generator

An experimental solar steam generator, consisting of a parabolic trough collector, a high-pressure steam circuit, and a suitable flash vessel has been constructed and tested in order to establish the thermodynamic performance during heat-up. The heat-up energy requirement has a marked effect on the systems performance because solar energy collected during the heating-up period is lost at night due to the diurnal cycle. This depends mostly on the dimensions and the inventory of the flash vessel, and the prevailing environmental conditions. Experimental data were obtained and used to train an artificial neural network in order to implement a mapping between easily measurable features (environmental conditions, water content and vessel dimensions) and the system temperatures. Such mapping may be useful to system designers when seeking to find the optimal vessel-dimensions. The trained network predicted very well the response of the system, as indicated by the statistical R-squared value of 0.999 obtained and a maximum deviation between predicted and actual values being less than 3.9%. This degree of accuracy is acceptable in the design of such systems. The results are important, because the system was tested during its heat-up cycle, under transient conditions, which is quite difficult to model analytically.

Artificial neural nets in computer-aided macro motor unit potential classification

The use of macro electromyography to obtain a macro motor unit potential (MMUP) is described. At least 20 potentials are measured from a single muscle to obtain a reasonable estimate of the parameters of an average motor unit potential. The MMUP data are analyzed by means of the peak-to-peak amplitude and the integral of the central 50 ms of the signal. The possibility of using artificial neural networks (ANNs) to analyze the macro data in a way that makes no assumptions about the relationships between the parameters and without recourse to conventional modeling methods is discussed. The results of an analysis carried out on 820 MMUPs recorded from 41 subjects who were classified on the basis of a clinical opinion and the appearance of a muscle biopsy are presented and discussed.<<ETX>>

Practical evaluation of GPRS use in a telemedicine system in Cyprus

The unceasing emergence of new technologies in wireless and mobile telecommunication networks, combined with the simultaneous rapid advances in information technology, are leading to many new solutions in the field of telemedicine, thus offering more opportunities for improving further existing and supporting new advanced services for healthcare. The objective of this paper is to carry out a practical evaluation of the performance of the GSM and GPRS systems in the transmission/reception of X-ray images and video in emergency orthopedics cases. As expected, the performance of GPRS is superior to that of GSM. The data transfer rate achieved with GPRS were in the range of 32 Kbps with the download time for typical X-ray images of a file size of 200 Kbytes to the mobile device to be in the region of 60 seconds. Similar performance was also recorded in the case of a moving station (simulating the ambulance) for the biggest part of the journey. In conclusion, although the medical imaging downloading timing was in the range of a few minutes, the physicians were very pleased by the benefits offered by the system through the freedom of access, anywhere and anytime even in motion.

Genetics-based machine learning for the assessment of certain neuromuscular disorders

Clinical electromyography (EMG) provides useful information for the diagnosis of neuromuscular disorders. The utility of artificial neural networks (ANNs) in classifying EMG data trained with backpropagation or Rohonens self-organizing feature maps algorithm has recently been demonstrated. The objective of this study is to investigate how genetics-based machine learning (GBML) can be applied for diagnosing certain neuromuscular disorders based on EMG data. The effect of GBML control parameters on diagnostic performance is also examined. A hybrid diagnostic system is introduced that combines both neural network and GBML models. Such a hybrid system provides the end-user with a robust and reliable system, as its diagnostic performance relies on more than one learning principle. GBML models demonstrated similar performance to neural-network models, but with less computation. The diagnostic performance of neural network and GBML models is enhanced by the hybrid system.

Artificial Neural Network Learning: A Comparative Review

Various neural learning procedures have been proposed by different researchers in order to adapt suitable controllable parameters of neural network architectures. These can be from simple Hebbian procedures to complicated algorithms applied to individual neurons or assemblies in a neural structure. The paper presents an organized review of various learning techniques, classified according to basic characteristics such as chronology, applicability, functionality, stochasticity etc. Some of the learning procedures that have been used for the training of generic and specific neural structures, and will be reviewed are: Hebbian-like (Grossberg, Sejnowski, Sutton, Bienenstock, Oja & Karhunen, Sanger, Yuile et al., Hasselmo, Kosko, Cheung & Omidvar), Reinforcement learning, Min-max learning, Stochastic learning, Genetics-based learning, Artificial life-based learning. The various learning procedures will be critically compared, and future trends will be highlighted.

Mobile health systems: A brief overview

Rapid advances in information technology and telecommunications, and more specifically wireless and mobile communications, and their convergence (telematics) are leading to the emergence of a new type of information infrastructure that has the potential of supporting an array of advanced services for healthcare. The objective of this paper is to provide a snapshot of the applications of mobile technology in healthcare. A brief review of the spectrum of these applications and the potential benefits of these efforts will be presented, followed by success case studies in electronic patient record, emergency telemedicine, teleradiology, and home monitoring. It is anticipated that the progress carried out in these efforts, and the potential benefits of emerging mobile technologies will trigger the development of more applications, thus enabling the offering of a better service to the citizen.

An overview of recent end-to-end wireless medical video telemedicine systems using 3G

Advances in video compression, network technologies, and computer technologies have contributed to the rapid growth of mobile health (m-health) systems and services. Wide deployment of such systems and services is expected in the near future, and its foreseen that they will soon be incorporated in daily clinical practice. This study focuses in describing the basic components of an end-to-end wireless medical video telemedicine system, providing a brief overview of the recent advances in the field, while it also highlights future trends in the design of telemedicine systems that are diagnostically driven.

Classification capacity of a modular neural network implementing neurally inspired architecture and training rules

A three-layer neural network (NN) with novel adaptive architecture has been developed. The hidden layer of the network consists of slabs of single neuron models, where neurons within a slab-but not between slabs- have the same type of activation function. The network activation functions in all three layers have adaptable parameters. The network was trained using a biologically inspired, guided-annealing learning rule on a variety of medical data. Good training/testing classification performance was obtained on all data sets tested. The performance achieved was comparable to that of SVM classifiers. It was shown that the adaptive network architecture, inspired from the modular organization often encountered in the mammalian cerebral cortex, can benefit classification performance.