Classification of Lung Carcinoma using Texture Features of Ultrasound Images

Abstract

Lung cancer is the main cause of cancer death in women and men across the planet. The lung carcinoma is divided into two categories: Small Cell Lung Cancer (SCLC) and Non Small Cell Lung Cancer (NSCLC). A large number of techniques are being used for detection and diagnosis of Lung Cancer. The Computer Aided Diagnosis (CAD) is the most common and accurate technique for early detection of abnormal cells which can cause cancer to healthy lung tissues. CAD system works on the basis of analysis of condition of ultrasound images. CAD system follows different steps: Data collection (ultrasound image), Data Preprocessing (ROI Selection), Feature Extraction, Data Partitioning (hold- out method), Feature Classification and Result Calculation. To classify input ultrasound images into benign and malignant, different classifiers were used. The system work is based on the calculation of parameters such as individual accuracy, overall accuracy and sensitivity. These benchmarks are obtained by calculating the matrix of Support Vector Machine (SVM). The results were obtained by using various features using Statistical Methods. The best results achieved were having accuracy of 91.4% by using Gray Level Difference Statistics (GLDS). The results will be used in CAD system for detection of Lung Cancer in initial stage to enhance the capability of survival of patient.