Search

Breast Cancer has very low survival rate when detected at later stages. Hence, Computer Aided Diagnostic Models can assist the medical practitioners by diagnosing reports without human intervention. In this paper the sample size considered is 401(200 benign and 201 malignant images) and has been acquired from the Digital Database for Screening Mammography. This paper propose a novel method to detect breast malignancies using texton based analysis. The filter bank that has been used here to obtain the texton based response is the Maximum Response 8 Filter Bank. Further Haralick’s features from the Gray Level Co-occurence Matrix , histogram based features from Local Binary Pattern and statistical features namely skewness and kurtosis are extracted from each filter response. Genetic Algorithm and Linear Discriminant Analysis has been used for feature selection and feature reduction respectively. Classification is performed using three classifiers namely Naive bayes, Logistic Regression and Linear SVM. The proposed algorthm exhibit an Accuracy of 87.5% and Area under Curve of 0.95 using Logistic Regression classifier.

In this paper a virtual instrumentation based electronic tongue has been described by applying the principles of cyclic voltammetry. The set up consists of a three electrode system. A triangular pulse was applied as the input from a data acquisition card via an amplification and level shifter circuit and the voltage equivalent of the output current from the working solution was considered for data analysis. Initially, the huge amount of data has been compressed using discrete wavelet transform (DWT). Principal component analysis (PCA) and linear discriminant analysis (LDA) have been performed on the compressed data. Further, different pattern recognition models based on neural networks were used to carry out a correlation study with the tea tasters' score of 4 different grades of black tea. With unknown tea samples, encouraging results have been obtained with more than 90% classification rate.

This paper aims at quantitative analysis of histopathological features of precancerous lesion and condition using image processing technique. The algorithm involves median and low pass filtering, segmentation by adaptive region growing, optimal and local thresholding, morphological operations such as opening and closing of gray scale and binary images and some numerical methods. Differentiation on the basis of type and level of precancerous type or condition is carried out based on image marker, defined as a vector of cancer related features viz. length and curvature of radius of rete-ridges and papillae, population density of cells within epithelium, etc. Implementation of presented algorithms is done in MATLAB. The results support quantitative analysis of pathological condition in respect with progression towards malignancy. This analysis may help in developing automated analysis tool.

A knowledge-baaed decision system developed for SHV transmission line Protection. This Paper describes the two major aspects of the system, (1) Knowledge reprenentation, (2) logic process, for decission fault distance calculation. Different knowledge representation techniques are discussed with their relative performances. The developed system is shown to be reliable and stable under different fault condition, Including zero fault inception angle. Computing burden is quite reasonable and ultra high speed (UHS) performance can be achieved with commercial microprocessors.

A knowledge-based decision system is developed for EHV transmission line distance protection. This paper describes the two major aspects of the system, (1) Knowledge representation, (2) logic process, for fault distance calculation. Different knowledge representation techniques are discussed with their relative performances, The developed system is shown to be reliable and stable under different fault condition, including zero fault inception angle. Computing burden is quite reasonable and ultra high speed (UHS) performance can be easily achieved with commercial microprocessors. The need and significance of UHS relaying for EHV transmission lines are by now well known. It improves power system reliability, transient stability and reduces the damagre caused to the equipments. There are caused to the equipments. There are apparently two distinct approaches in distance protection taken to achieve high speed performance. One is based on fast estimation of power frequency components using Kalman filters symmetrical. components etc, The other approach relier on the information content in the high frequency part of the fault induced components. The later is popularly called as traveling wave (TW) based distance rela ying. Both the schemes have their own advantage and dioadvantages. The initial position of the fault induced voltage and currents are heavily dominated by travelling waves. So for the initial period of a few milliseconds, signal to noise ratio is higher for TW-based relaying and lower for power frequency based relaying. However, for near zero-fault inception angles the magnitudes of the travelling waves are very low and it is very difficult to detect these faults using TW-baoed distance relaying. The problem can be solved by combining both the methods with greater computing burden and algorithm complexity. A novel approach is presented here to solve this problem using knowledge-based system. The beauty of this system is that the whole spectrum of the information -ontalned In the fault Induced signals (limited by the sampling frequency) is considered in a very simple manner. The idea of the knowledge-based system came from the fact that post fault voltage and current waveforms are dependent on fault distance. The authors, for a time, were engaged in applying different signal processing techniques to develop TW-based distance relaying. It is observed that fault distance can be estimated from the fault-induced signals using various signal processing techniques e.g. correlations of the signals, power spectrum of the signals signal cepstrum . The decission process related to all the algorithms are more or less same : detecting some strongest component and its position. The information in the other part of the power spectra on correlation functions cannot be easily treated or related to fault distance always, thus motivating the application of knowledge-based system. The informations in fault induced signals are stored as a function of fault distance with the help of digital simulation. The information may be the raw post fault signals or some pre- processed versions of it. However, fault-generated waveforms are also dependent on various system parameters, other than fault distance. Some of the system parameter variations are taken account of in the knowledge representation part. Even then the searching procedure will be too much time consuming and some amount of higher level knowledge has to be applied to cut down the searching procedure. This technique helped the proposed relay achieving UHS performance with commercial microprocessors.

An optical method for rapid assessment of SOI structures is developed. This method is based upon angle lapping and etching to observe crystal defects in SOI structures. Traditional beveling methods introduce defects due to mechanical damage resulting from unintentional scratches caused by the material breaking loose from the sides of the sample. The proposed approach covers the sidewalls and the surface of the sample during lapping with a waxy material which provides a thick soft matrix to embed loose particles and eliminate the scratches. The method has been used to investigate both SIMOX and ISE structures. The proposed method has a greater sensitivity of defect detection than the cross-sectional transmission electron microscopy commonly used in defect studies and it is also well suited for evaluating process induced defects in bulk silicon devices. >