A new region-of-interest (ROI) detection method using the chan-vese algorithm for lung nodule classification

Abstract

Suspicious regions in chest x-rays are detected automatically, and these regions are classified intothree types, including “malignant nodule”, “benign nodule”, and “no nodule” in this study. Firstly,the areas except the lung tissues are removed in each chest x-ray using the thresholding method.Then, Poisson noise was removed from the images by applying the gradient filter. Ribs mayoverlap onto nodules. Since this circumstance will make the detection of a nodule difficult, it isnecessary to distinguish and suppress the ribs. The location of the rib bones is determined by atemplate matching method, and then the corresponding bones are suppressed by applying theGabor filter. After this stage, suspicious tissues in the chest x-rays are specified using the ChanVese active contour without edges. Then, some features are extracted from these suspiciousregions. Six different features are extracted: Statistical, Histogram of Oriented Gradients (HOG)-based, Local Binary Pattern (LBP)-based, Geometrical, Gray Level Co-Occurrence Matrix(GLCM) Texture-based and Dense Scale Invariant Feature Transform (DSIFT)-based. Then, theclassification stage is achieved using these features. The best classification result is obtained usingstatistical, LBP-based, and HOG-Based features. The classification results are evaluated withsensitivity, accuracy, and specificity analyses. K-Nearest Neighbour (KNN), Decision Tree (DT),Random Forest (RF), Logistic Linear Classifier (LLC), Support Vector Machines (SVM), Fisher’sLinear Discriminant Analysis (FLDA), and Naive Bayes (NB) methods are used for theclassification purpose separately. The random forest classifier gives the best results with 57%sensitivity, 66% accuracy, 81% specificity values.