Extraction Of Pleural Indentation And Lung Cavity Of Lung Nodules In CT Images

Nowadays lung cancer is the first cause of death of malignant tumors, and early diagnosis and treatment is very important to increasing the survival rate of lung cancer patients. CT image is the best diagnostic imaging tool for pulmonary nodules, and the computer-aided diagnosis of pulmonary nodules based on CT images is the current domestic and international research hotspot. The feature extraction and quantification of pulmonary nodules is the basis of determining a nodule's malignant or benign. Pleural indentation and lung cavity both have great clinical diagnostic value. Pleural indentation is one of the common signs of early lung cancer, and malignant lung cavity often has thick walls, be eccentric and so on. So, it is of great value to extracts pleural indentation sign and lung cavity sign.In this study, we present a new method of pleural indentation extraction according to the features of pleural indentation. First, we should find a convex which contains all pulmonary parenchyma points within a certain range around the nodules, and the convex perhaps include the two point where the pleural depressed; then, for every two adjacent convex point, we should determine whether the area which is surrounded by its connected line and both sides of the pleural is connected with the nodules and whether there is a V-shaped depression here.The study of extracting the feature of lung cavity has two parts: determining whether there is a cavity in the nodule, extracting kinds of features of the cavity such as its volume, density, wall thickness, and eccentricity and so on. How to determining whether there is a cavity in the nodule and computing the wall thickness correctly is the core algorithm. How to determining whether there is a cavity in the nodule contains the following steps: First, we separated the pulmonary parenchyma by threshold segmentation and get some candidate region, then, we eliminates noise, removes bronchi and determine whether it is a cavity through the size of the candidate region. Computing the wall thickness accurately was often difficult because the cavity's irregular shape. We builds a mathematical model to make the sum of the distances between all inner wall points and the corresponding outer wall points minimum, and compute by greedy algorithm.The experiments results show that the sensibility of the algorithm of pleural indentation extraction up to 100% and the true positive rate was 80%, the algorithm of extracting the feature of lung cavity can determine the existence of a cavity in the nodule and compute the features accurately.