Hatice Sezgin

Effects of dimension reduction in mammograms classification

Breast cancer is the most common type of cancer among women and causing deaths in women. In this paper, a CAD system is presented to investigate effects of dimension reduction for classifying mammograms. Proposed system consists of preprocessing, feature extraction, dimension reduction and classification steps. Multiscale top-hat transform is used to enhance mammograms and to remove noise. First order and second order textural features are extracted from enhanced mammograms. Principal component analysis (PCA) is used for dimension reduction. Two multilayer perceptron neural networks (MLP) are used to classify mammograms as normal or abnormal. All twenty features (without PCA) and selected seven features by PCA are applied each of two classifiers. First MLP classifier with all features achieved accuracy of 79,4%. Second MLP classifier with selected features by PCA achieved accuracy of 91,1%. PCA feature dimension reduction improved the classification performance, increasing accuracy value from 79,4% to 91,1%.

Approach for determining the optimum pilot placement in orthogonal frequency division multiplexing systems

One of the major concerns in wireless communications is multipath fading that causes some adverse effects and significantly limits the performance of a system. To overcome these effects in orthogonal frequency division multiplexing, system channel estimation must be performed. Pilot arrangement is very crucial in pilot aided channel estimation. There is an inevitable trade-off between the number of pilot symbols used in channel estimation, the accuracy of such estimation, and spectral efficiency. In this study, the analytical formulation of the channel estimation mean squared error based on a channels sampled power density is presented. A novel mathematical rule for the optimum pilot symbol interval and a novel adaptive pilot placement are proposed to improve channel estimation quality and bandwidth efficiency. The optimum pilot symbol intervals which are related to channels’ coherence times and coherence bandwidths can be calculated easily with the use of the proposed method. As the proposed method effectively reduces the number of searches for the optimum pilot symbol interval, it gives rise to simply constructing adaptive pilot placement in real-time systems. Compared with fixed pilot placement, the proposed adaptive placement shows obvious improvement up to 6% in bandwidth efficiency while maintaining the bit error rate performance.

Describing the optimum pilot symbol interval via curve fitting methods in OFDM system

Increasing demand on higher data rates, makes multi carrier systems popular. Orthogonal frequency division multiplexing (OFDM), a form of multicarrier modulation, has been widely applied in wireless communication systems for major advantages. In order to obtain channel state information, and overcome multipath fading channel estimation must be performed in these systems. The channel estimation quality significantly depends on the pilot pattern. In this study, the relationship between the optimum pilot symbol interval that yields the minimum channel estimation mean squared error and coherence bandwidth, coherence time at 0,9 correlation is investigated in OFDM systems for different time and frequency selective channels. Different curve fitting methods are applied to data and performances are compared in terms of normalized root mean squared errors (NRMSE). Considering both simplicity and NRMSE; the optimum pilot symbol interval is calculated by cubic polynomial in frequency direction, and by quadratic polynomial in time direction.

Determination of the optimum pilot density for an OFDM system over empirical channels

In orthogonal frequency division multiplexing (OFDM) systems, in order to overcome distortion caused by multipath fading, channel estimation is performed. Pilot symbol assisted channel estimation is effectual way to obtain channel state information. However, use of pilot symbols, leads overhead and the number of pilot symbols should be kept as minimum as possible. In this study, in order to evaluate the optimum pilot density, channel state information is obtained by using two dimensional (2-D) pilot based channel estimation. The analytical and simulated optimum pilot intervals in frequency and time direction are determined in terms of mean squared error (MSE). Then these intervals related to channels coherence time (Tc) and bandwidth (Bc) at 0.9 correlation coefficient. It is seen from the results that by using approximately 2.4 pilot symbols per Bc,0.9, and per Tc,0.9 the optimum pilot interval can be calculated easily.