Rib suppression in chest radiographs for lung nodule enhancement

摘要

Lung cancer is one of the malignant tumors with the fastest increasing speed of incidence and death rate. It appears in the form of spherical nodules in a conventional radiograph. However, some lung nodules are not be able to be detected due to their overlap with normal anatomic structures such as ribs and clavicles. In this paper, a rib suppression method based on principle component analysis (PCA) is presented to improve the visibility of lung nodules. Firstly, the rib models are built by using PCA; secondly, the pixel intensity of background is calculated and added to the subtracted image of original image and rib model to recover the original brightness; finally, the border of ribs is detected and smoothed. The JSRT database and dual-energy images were employed to evaluate the rib suppression method. The experiment indicates that when the method is applied on the test images, the ribs can be suppressed substantially. The proposed rib suppression method can improve the visibility of tumor in lung field.