Academic Journals Database
Disseminating quality controlled scientific knowledge

A Wavelet Based Automatic Segmentation of Brain Tumor in CT Images Using Optimal Statistical Texture Features

ADD TO MY LIST
 
Author(s): A.Padma & Dr.R.Sukanesh

Journal: International Journal of Image Processing
ISSN 1985-2304

Volume: 5;
Issue: 5;
Start page: 552;
Date: 2011;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Discrete Wavelet Transform(DWT) | Genetic Algorithm(GA) | Receiver Operating Characteristic(ROC)analysis | Spatial Gray Level Dependence Method (SGLDM) | Probabilistic Neural Network (PNN).

ABSTRACT
This paper presents an automated segmentation of brain tumors in computed tomographyimages (CT) using combination of Wavelet Statistical Texture features (WST) obtained from 2-level Discrete Wavelet Transformed (DWT) low and high frequency sub bands and Wavelet CooccurrenceTexture features (WCT) obtained from two level Discrete Wavelet Transformed(DWT) high frequency sub bands. In the proposed method, the wavelet based optimal texturefeatures that distinguish between the brain tissue, benign tumor and malignant tumor tissue isfound. Comparative studies of texture analysis is performed for the proposed combined waveletbased texture analysis method and Spatial Gray Level Dependence Method (SGLDM). Ourproposed system consists of four phases i) Discrete Wavelet Decomposition (ii) Featureextraction (iii) Feature selection (iv) Classification and evaluation. The combined WaveletStatistical Texture feature set (WST) and Wavelet Co-occurrence Texture feature (WCT) setsare derived from normal and tumor regions. Feature selection is performed by Genetic Algorithm(GA). These optimal features are given as input to the PNN classifier to segment the tumor. AnProbabilistic Neural Network (PNN) classifier is employed to evaluate the performance of thesefeatures and by comparing the classification results of the PNN classifier with the Feed ForwardNeural Network classifier (FFNN).The results of the Probabilistic Neural Network, FFNNclassifiers for the texture analysis methods are evaluated using Receiver OperatingCharacteristic (ROC) analysis. The performance of the algorithm is evaluated on a series of braintumor images. The results illustrate that the proposed method outperforms the existing methods.
Affiliate Program     

Tango Rapperswil
Tango Rapperswil