Region based ct lung image analysis with neural network multi level classifier

Abstract

Lung cancer is a severe social health problem, as the most significant data today can indicate that its incidence continues to increase in women at 3.1%annually over the past 20 years; on the other hand in men, an annual decline of0.8% is also observed for the past 20 years. The modalities used for capturing the images are X-Ray, Computed Tomography (CT) scans and Magnetic Resonance Imaging (MRI) and among these CT is the standard for detecting pulmonary nodules. Detection of lung nodules in chest Computed Tomography (CT) images become very necessary in the present clinical world. The neuralnetwork has been the important ares in the last 20 years of research. The chief objective of this research is to provide anautomatic segmentation method for lung cancer detection. In the first work of this research, a neural network system for detection of lung cancer nodules from the Chest Computed Tomography images. An artificial neural network group is a learningmodel where several artificial neural networks are gathered to solve a problem. The techniques utilized for feature extraction included morphological, statistical features, as well as features traced out from texture analysis, Fourier-based features, and wavelet-based features. In the second work, a New Dynamic Multi-Level classifier Computer-Aided Diagnosis (CAD) system for enhanced automatic detection of CT Lung Images for abnormalities isproposed. In this work, we also proposed to enhance the CT Lung images using Selective Median Filter (SMF) for improving the quality of the image by reducing the noise. We present a new neural network Multi- Level Classifier segmentation method to reduce the suspicious region from the enhanced CT scan image. In this paper, we also design a New Neural Network based Multi-Level Classifier for classifying the CT scan images using the extracted textural features newline