Zu Soh

Theoretical and evolutionary parameter tuning of neural oscillators with a double-chain structure for generating rhythmic signals

A neural oscillator with a double-chain structure is one of the central pattern generator models used to simulate and understand rhythmic movements in living organisms. However, it is difficult to reproduce desired rhythmic signals by tuning an enormous number of parameters of neural oscillators. In this study, we propose an automatic tuning method consisting of two parts. The first involves tuning rules for both the time constants and the amplitude of the oscillatory outputs based on theoretical analyses of the relationship between parameters and outputs of the neural oscillators. The second involves an evolutionary tuning method with a two-step genetic algorithm (GA), consisting of a global GA and a local GA, for tuning parameters such as neural connection weights that have no exact tuning rule. Using numerical experiments, we confirmed that the proposed tuning method could successfully tune all parameters and generate sinusoidal waves. The tuning performance of the proposed method was less affected by factors such as the number of excitatory oscillators or the desired outputs. Furthermore, the proposed method was applied to the parameter-tuning problem of some types of artificial and biological wave reproduction and yielded optimal parameter values that generated complex rhythmic signals in Caenorhabditis elegans without trial and error.

An Artificial Neural Network Approach for Glomerular Activity Pattern Prediction Using the Graph Kernel Method and the Gaussian Mixture Functions

This paper proposes a neural network model for prediction of olfactory glomerular activity aimed at future application to the evaluation of odor qualities. The models input is the structure of an odorant molecule expressed as a labeled graph, and it employs the graph kernel method to quantify structural similarities between odorants and the function of olfactory receptor neurons. An artificial neural network then converts odorant molecules into glomerular activity expressed in Gaussian mixture functions. The authors also propose a learning algorithm that allows adjustment of the parameters included in the model using a learning data set composed of pairs of odorants and measured glomerular activity patterns. We observed that the defined similarity between odorant structure has correlation of 0.3-0.9 with that of glomerular activity. Glomerular activity prediction simulation showed a certain level of prediction ability where the predicted glomerular activity patterns also correlate the measured ones with middle to high correlation in average for data sets containing 363 odorants.

Unconstrained and noninvasive measurement of bioelectric signals from small fish

Recently, the technique of fish bioassay has attracted attention as a method for constant monitoring of aquatic contamination. The respiratory rhythms of fish are considered to be an efficient indicator for the monitoring of water quality, since they are sensitive to chemicals and can be measured indirectly from the bioelectric signals generated by their breathing. However, no method has yet been established to measure signals in small free-swimming fish. In this article, we propose a system to measure bioelectric signals in small fish and monitor the frequency component in real time. To cover the large measurement range required in a free-swimming environment, the signals are measured using multiple electrodes. Further, the system focuses on the frequency component of the signal to assess the condition of the fish using frequency analysis and a band-pass filter. Experiments were conducted with the purpose of enabling remote sensing and environment estimation. First, it was verified that the measured signals were synchronized with the breathing of the fish. Then, a remote sensing experiment was performed using medaka (Oryzias latipes) that were allowed to swim freely in a measurement aquarium. The results confirmed that bioelectric signals which were synchronized with breathing could be measured in unconstrained and noninvasive conditions.

A Novel Blood Viscosity Estimation Method Based on Pressure-Flow Characteristics of an Oxygenator During Cardiopulmonary Bypass.

During cardiopulmonary bypass (CPB), blood viscosity conspicuously increases and decreases due to changes in hematocrit and blood temperature. Nevertheless, blood viscosity is typically not evaluated, because there is no technology that can provide simple, continuous, noncontact monitoring. We modeled the pressure-flow characteristics of an oxygenator in a previous study, and in that study we quantified the influence of viscosity on oxygenator function. The pressure-flow monitoring information in the oxygenator is derived from our model and enables the estimation of viscosity. The viscosity estimation method was proposed and investigated in an in vitro experiment. Three samples of whole bovine blood with different hematocrit levels (21.8, 31.0, and 39.8%) were prepared and perfused into the oxygenator. As the temperature changed from 37°C to 27°C, the mean inlet pressure (Pin ) and outlet pressure (Pout ) of the oxygenator and the flow (Q) and viscosity of the blood were measured. The estimated viscosity was calculated from the pressure gradient (ΔP = Pin  - Pout ) and Q and was compared to the measured blood viscosity. A strong correlation was found between the two methods for all samples. Bland-Altman analysis revealed a mean bias of -0.0263 mPa.s, a standard deviation of 0.071 mPa.s, limits of agreement of -0.114-0.166 mPa.s, and a percent error of 5%. Therefore, this method is considered compatible with the torsional oscillation viscometer that has plus or minus 5% measurement accuracy. Our study offers the possibility of continuously estimating blood viscosity during CPB.

Hydrodynamic characteristics of a membrane oxygenator: modeling of pressure-flow characteristics and their influence on apparent viscosity

The viscosity obtained from pressure-flow characteristics of an oxygenator may help to detect factors that change oxygenator resistance. The objective of this study was to model pressure-flow characteristics of a membrane oxygenator with an integrated arterial filter and to quantify their influence on apparent viscosity of non-Newtonian fluids. One Newtonian fluid (glycerin solution) and two non-Newtonian fluids (whole bovine blood and a human red blood cell suspension) were perfused through an oxygenator and their pressure-flow characteristics examined systematically. Four resistance parameters for the pressure gradient characteristics approximation equation were obtained by the least squares method from the relational expression of pressure-flow characteristics and viscosity. For all three fluids, a non-linear flow to pressure change was observed with a coefficient of determination of almost 1 by exponential approximation. The glycerin solution had a higher pressure gradient (10-70%) than the other fluids; the apparent viscosity of the non-Newtonian fluids was around 35% lower than the static one measured by a torsional oscillation viscometer. Overall, our study demonstrated that the influence on the apparent viscosity of non-Newtonian fluids can be quantified by pressure gradient differences in a membrane oxygenator with an integrated arterial filter.

A Comparison Between the Human Sense of Smell and Neural Activity in the Olfactory Bulb of Rats

Generally, odor qualities are evaluated via sensory tests in which predefined criteria are assessed by panelists and stochastically analyzed to reduce human inconsistencies. Because this method requires multiple, well-trained human subjects, a more convenient approach is required to enable predictions of odor qualities. In this article, we propose an approach involving linking internal states of the olfactory system with perceptual characteristics. In the study, the glomerular responses of rats were taken to represent internal olfactory system states. Similarities between the glomerular responses of rats were quantified by correlations between glomerular activity patterns, overlap rate of strongly activated part across glomerular activity patterns, and the similarity between histograms of the strength of activity. These indices were then compared with perceptual similarities measured from human subjects in sensory tests. The results of experiments involving 22 odorants showed medium strength correlations between each index and perceptual similarity. In addition, when the 3 indices were combined using their Euclidean distance, we observed middle to high correlations (r = 0.65-0.79) to human perceptual similarity. We also report the results of our use of a machine learning technique to classify the odorants into a similar and dissimilar category. Although the correct rate of classification varied from 33.3% to 92.9%, these results support the feasibility of linking the glomerular responses of rats to human perception.

A neural network model of the olfactory system of mice: simulated the tendency of attention behavior

The demands for odor processing apparatuses have been increasing in fragrance or food industries. However, odors are extremely high dimensional information composed a combination from tens thousands of different odorant molecules, and thus requires vast amounts of computation. Therefore, it is considered learning from a living nose would be an efficient approach. From the odor discrimination experiments, it was found that mice have a feature extraction ability called Attention by which they could focus on the important odorants for odor discrimination. In this paper we propose a neural network model approximated to actual number of neurons and the structure of olfactory system. Simulation experiments of the proposed model were implemented based on the odor discrimination experiments on the living mice. From the simulation results of the model, we confirmed not only the proposed model had ability of Attention, but also the tendency of Attention was consistent with the living mice.

A blood viscosity estimation method based on pressure-flow characteristics of an oxygenator during cardiopulmonary bypass and its clinical application.

In this paper, we developed a model that uses pressure-flow monitoring information in the oxygenator to estimate viscosity of human blood. The comparison between estimated viscosity (ηe) and measured viscosity (η) was assessed in 16 patients who underwent cardiac surgery using mild hypothermia cardiopulmonary bypass (CPB). After initiation of CPB, ηe was recorded at three periods: post-establishment of total CPB, post-aortic cross-clamp, and post-declamp. During the same period, blood samples were collected from the circuit and η was measured with a torsional oscillation viscometer. The ηe was plotted as a function of η and the systematic errors and compatibility between two methods were assessed using Bland-Altman analysis. The parameters ηe and η were very strongly correlated at all points (R2=0.9616, p<;0.001). The Bland-Altman analysis revealed a mean bias of -0.001 mPas, a standard deviation of 0.03 mPas, limits of agreement of -0.06 mPas to 0.06 mPas, and a percent error of 3.3%. There was no fixed bias or proportion bias for the viscosity. As this method estimates blood viscosity with good precision during CPB continuously, it may be helpful for clinical perfusion management.

A novel tuning method for neural oscillators with a ladder-like structure based on oscillation analysis

Neural oscillators with a ladder-like structure is one of the central pattern generator (CPG) model that is used to simulate rhythmic movements in living organisms. However, it is not easy to realize rhythmical cycles by tuning many parameters of neural oscillators. In this study, we propose an automatic tuning method. We derive the tuning rules for both the time constants and the coefficients of amplitude by linearizing the nonlinear equations of the neural oscillators. Other parameters such as neural connection weights are tuned using a genetic algorithm (GA). Through numerical experiments, we confirmed that the proposed tuning method can successfully tune all parameters.

A neural network model of the olfactory system of mice: computer simulation of the attention behavior of mice for some components in an odor

Recently, it was observed that mice could identify an odor by paying attention to only a few of its components. Further, it has been reported that each individual is attracted to different components of an odor. This behavior is referred to as “attention”; however, its mechanism has yet to be completely elucidated. In this paper, we first propose a novel artificial neural network model based on the biological structure of an olfactory system. Then a series of computer simulations of odorant discrimination are performed to evaluate the attention ability of the proposed model. Finally, we changed the connective weights between the neurons to simulate individual differences. The simulation results indicate that the inhibitory connections from the piriform cortex to the olfactory bulb may contribute to the individual differences observed in the behavioral experiment.