Enhanced Prediction of Heart Disease by Genetic Algorithm and RBF Network
Author: A. Durga Devi
Abstract
A huge amount of data is available in hospitals and clinics. These data become useful and meaningful, only when data mining techniques are applied on them to retrieve knowledge. Nowadays, most of the human deaths occur by heart diseases. It is very important to find a prediction system with more accuracy. This prediction system predicts heart disease risk with an enhanced accuracy. The enhanced accuracy is obtained by the attribute reduction. The data mining technique RBF Network is used to create the classifier. Genetic Algorithm is used for attribute selection or reduction. The result shows that the RBF Network performs better than the other data mining techniques Naïve Bayes and J48.
NIR Based Dental Imaging Techniques For Caries Detection
Authors: Renee Christina. R, Vanithamani. R
Abstract
Demineralization is the common problem that exists these days. Although traditional methods are commonly used to detect them, Near-infrared rays is found to be more beneficial than the other methods. Near-infrared (NIR) is a non-ionizing, minimally invasive approach to detect and quantify caries. In dental imaging, the tooth’s optical property exhibits deeper penetration into tooth because of the enamel which is highly transparent to NIR. The first step is using the NIR Light emitting Diode and IR camera the tooth image is acquired. The captured tooth image exhibits a clear variance between the healthy and caries affected region. Since the contrast and resolution of the image captured using IR camera is low, it is enhanced using morphological operation and top-hat transform processing methods. To identify and classify the tooth as caries affected tooth, histogram plot is used. To accurately determine the affected region, the Otsus’ segmentation method is used to extract the caries region.
New Technique for Reduction of THD in Single and Three Phase Multilevel Inverter
Authors: K. Sridhar, M. Bhasker
Abstract
Many recent multilevel inverter papers end up with voltage Total Harmonic Distortion (THD) values obtained from numerical voltage spectrum calculations (measurements). Motivated by IEEE Standard 519, a part of multilevel research community uses a limited harmonics count to evaluate multilevel voltage quality. First, this causes significant voltage THD underestimation, especially, for relatively high frequency PWM. Second, for a three-phase star-connected balanced load with an isolated neutral and phase symmetric modulation strategy, calculated load line and phase voltage THD become different. However, simple considerations show that line and phase voltage THD are essentially the same in this case. It may be difficult to judge about multilevel voltage quality given a numerically calculated (measured) voltage THD value that may be subject to computation errors. Presented are simple smooth hyperbolic voltage THD upper and lower bounds approximations for single- and three-phase inverters with nearest synchronous switching. They are valid for arbitrary modulation indices and uniformly distributed levels counts and may practically serve as good reference values.
A Novel Idea Based On Swirl Motion Of Water For Function Optimization
Authors: Mr. S. Sathish kumar, Mr. C. B. Sivaparthipan, Mr. J.Rathinaraja
Abstract
This paper proposes a new optimization algorithm based on vortex particle theory. Here particles represent multiple solutions that swirl (rotate) around a centre point called vortex. Vortex particle theory considers the fitness value of current position of particles that is obtained by applying an updating equation as well as strength update on previous position of particles. Since the proposed modifications make the vortex location in to a more steady position thus it widens the exploration capability of the algorithm. The performance of the proposed algorithm is tested using six bench mark functions. From the simulation result it is found the exploration ability of the proposed algorithm is not constrained through a narrow zone and produces better results without getting struck in to the local optima.
Low Power Based Ecg Detector For Cardiac Pacemaker
Authors: S. Sharmila, C. Saranya
Abstract
To achieve the low power design of wavelet based electrocardiogram (ECG) for implantable cardiac pacemaker is presented. The proposed wavelet based ECG detector consists of wavelet decomposer, QRS complex detector and noise detector with zero crossing points. To achieve the significant power consumption and QRS detection accuracy multi scaled product algorithm and soft threshold algorithm have been used in QRS complex detector. The standard0.9 µm CMOS technology has been used.