Lung Nodules Classification in CT Images Using Shannon and Simpson Diversity Indices and SVM

摘要

In this work, we present the use of Shannon and Simpson Diversity Indices as texture descriptors for lung nodules in Computerized Tomography (CT) images. These indices will be proposed to characterize the nodules into two classes: benign or malignant. The investigation is done using the Support Vector Machine (SVM) for classification in a dataset consisting of 73 nodules, 47 benign and 26 malignant; the results of the methodology were: sensitivity of 85.64%, specificity of 97.89% and accuracy of 92.78%.