Catálogo de publicaciones - libros

Nowadays there exist different approaches to cancel out noise effect and baseline drift in biomedical signals. However, none of them can be considered as completely satisfactory. In this work an artificial neural network (ANN) based approach to cancel out baseline drift in electrocardiogram signals is presented. The system is based on a grown ANN allowing to optimize both the hidden layer number of nodes and the coefficient matrixes. These matrixes are optimized following the simultaneous perturbation algorithm, offering much lower computational cost that the traditional back propagation algorithm. The proposed methodology has been compared with traditional baseline reduction methods (FIR, Wavelet-based and Adaptive LMS filtering) making use of cross correlation, signal to interference ratio and signal to noise ratio indexes. Obtained results show that the ANN-based approach performs better, with respect to baseline drift reduction and signal distortion at filter output, than traditional methods.

This paper studies a sequential decomposition method suitable for real-time separation of linear mixtures of finite-length signals. The signals are modelled as channel responses in a multiple-input multiple-output (MIMO) model, with positive pulse trains as channel inputs. Our decomposition method compensates the channel responses and aims to reconstruct the input pulse trains in real time. Tests on synthetic surface electromyograms (SEMG) show how well the proposed method performs, in comparison to its batch version, and how robust it is.

In this paper we introduce a new method for muscle force estimation from multi-channel surface electromyograms. The method combines a motor unit twitch model with motor unit innervation pulse trains, which are estimated from multi-channel surface electromyograms. The motor unit twitches are then aligned to the innervation pulse trains and summed up to obtain the total muscle force. The method was tested on real surface EMG signals acquired during force ramp contractions of abductor pollicis brevis muscle in 8 male subjects. With 22 ± 5 (mean ± std. dev.) motor units identified per subject, the force estimation error of our method was 16 ± 4 % RMS. These results were compared to the method which uses the EMG amplitude processing to estimate muscle force. The results of our new concept proved to be completely comparable to those of EMG amplitude processing.

Today’s maternal/fetal monitoring lacks the capability to diagnose labor and predict delivery. The objective of this work was to demonstrate that uterine electromyography (EMG) is proven to be a viable alternative to current monitoring techniques. Uterine EMG was monitored noninvasively and trans-abdominally from pregnant patients using surface electrodes. Several aspects of the uterine EMG were investigated: contraction plotting, diagnosing labor, and predicting delivery. Contractions were seen to correspond well with tocodynamometer- (TOCO-) plotted contractions. As well, increases in electrical activity were indicative of labor and imminent delivery. Uterine EMG could be a valuable tool for obstetricians if implemented on a routine basis in the clinic.

We present efforts to objectively assess cervical ripeness in humans. The hypothesis was that the cervical EMG basal activity might reflect readiness of the cervix for delivery. 47 women at initiation of delivery were involved in the study. EMG parameters – amplitude ( U _ RMSA ) and frequency contents ( MF _ A ) – were related to Cumulative Bishop Score values (CBS) assessed by an obstetrician in each woman at the labor onset. The results show that the parameters are predictive of the CBS, both correlating negatively with the CBS value. Hence, EMG parameters have potentials to become objective indicators in assessment of cervical ripeness in humans. This would empower an obstetrician in his/her decisions how to further conduct the labor.

This paper describes two simple algorithms for detection of uterine contractions during labour using the uterine electromyogram recorded from the abdominal surface. The location of a contraction is extracted from the signal’s energy using either an amplitude- or derivative-based algorithm. For our recordings, these algorithms managed to correctly locate the majority of contractions, with an average success rate of 87.4% for the derivative-based algorithm and 85.6% for the amplitude-based algorithm.

Hospitals throughout Europe hold vast amounts of data in the form of patient records. Performing on-the-fly analyses of these data and their actual transformation into information and knowledge may help improve medical procedures, treatments or prevent illnesses. Grid technology has recently emerged to address the needs for efficient and effective exploitation of heterogeneous and geographically distributed resources, such as large and distributed data, open source or proprietary programs for data analysis, massive storage devices and high-performance computers. A de facto standard framework for building grid environments is the Open Grid Service Architecture (OGSA) and the corresponding Web Service Resource Framework (WSRF). The Globus Toolkit version 4, is a fully WSRF-compliant grid middleware, which addresses the needs for secure, flexible, interoperable and seamless use of grid resources. The DataMiningGrid© (www.datamininggrid.org) system was recently built on top of existing Globus technology inter alia to address the requirements of a community of medical users and enable them to perform on-the-fly analysis of geographically distributed medical databases. DataMiningGrid© is a set of grid services and user-friendly workflow editing and managing tools, which facilitate manipulation of distributed data, registering, discovery and use of grid-enabled statistical and data mining programs, their execution in the grid environment and a provenance tracking mechanism. The software is now freely available at SourceForge.net under Apache License V2. The present work illustrates the use of the DataMiningGrid© system to perform analysis of nine regional medical databases in Slovenia.

By exposing cells to high voltage electric pulses, cells’ membrane permeability increases significantly. Phenomenon is known as electroporation and is widely used in biotechnology, biology and medicine, as a way of introducing into a cell molecules which otherwise are deprived of membrane transport mechanisms. Besides cell membrane permeability, the cells’ geometrical and electrical properties change significantly due to electroporation. These changes have a huge impact on dielectrophoretic force, which could allow us to separate the electroporated and non-electroporated cells. Usually, a test whether a cell is electroporated or not is performed by exposing cells to a dye. After such test cells are most often not useful for further use. For this reason cell separation based on dielectrophoretic force could be very useful, because cells are not destroyed or changed due to dielectrophoresis. In this study we report the results of an attempt to separate the electroporated and non-electroporated cells by means of dielectrophoresis. In several experiments we managed to separate the electroporated and non-electroporated cells suspended in a medium with conductivity 0.174 S/m by exposing them to a non-uniform electric field at a frequency of 2 MHz. Because experimental results did not match theoretical predictions entirely we presume that cell membrane permittivity decreases after electroporation for at least ten times.

A custom made DAQ-card based bioimpedance measurement system is presented. The signals are processed by the digital lock-in technique. The system was tested on an electrical model of skin with underlying tissues over a frequency range of 20 Hz to 1 MHz. The measurements performed directly with the DAQ-card are compared to the measurements with the instrumentation amplifier interface. The highest achieved accuracy without special calibration and compensation is about 0.5 % for the impedance magnitude and 0.02° for the impedance phase angle in the low frequency region, whereas in the high frequency region the respective values are approximately 1 % and 1°.

Dielectric properties of biological substances are usually deduced ex vivo by the way of the impedance measurement of a cell loaded by the investigated medium. At low frequency it is well known that the bioimpedance depends on the polarization effects that occur at the electrodes interface. Measurements being affected for frequencies lower than 50 kHz for standard electrodes, black platinum was used to decrease polarization effects. In this paper, dielectric properties of blood measured at different temperatures are presented for frequencies varying between 100 Hz and 1 MHz.