A semantic contour based segmentation of lungs from chest x‐rays for the classification of tuberculosis using Naïve Bayes classifier

Abstract Tuberculosis (TB) is an infectious disease that primarily affects the lungs. If the TB is left undiagnosed in its early stage, it may affect other body parts leading to sudden death. A posterior anterior chest radiographs are used to diagnose pulmonary TB. This work presents a novel automatic system for the early detection of pulmonary TB. Initially, the Chan‐Vese active contour model is generated to segment the lung region from the preprocessed chest x‐rays. Different textural, statistical, and morphological features are extracted from the segmented lungs. Finally, the pertinent feature vector is considered for the classification of chest x‐rays using the Naïve Bayes classifier (NBC). The NBC is trained using 10‐fold cross validation technique. The proposed method achieves an average accuracy, the area under the curve, specificity, and sensitivity as 95.5%, 98%, 93.3%, and 94.6% respectively, when tested on different available datasets. Comparison results reveals that the Naïve Bayes classifier outperforms than the multinomial Naïve Bayes classifier and Bernoulli's Naïve Bayes classifier in classifying the chest x‐rays into healthy and TB.