Non-Rigid Registration Techniques for Automatic Follow-up of Lung Nodules

Our long term research goal is to develop an automatic approach for early detection of lung nodules that may lead to lung cancer. This paper focuses on the monitoring of the progress (growth or shrinking) of lung nodules in successive chest low dose CT (LDCT) scans of a person using non-rigid registration. The overall nodule detection approach consists of four main steps: 1) Extraction of lung region by modeling the gray level distribution of LDCT slices using a linear combination of Gaussians, 2) Classifying the lung region into homogenous tissues, arteries, veins and lung nodules, 3) Applying rigid registration using a combination of segmentation information and mutual information as a similarity measure, and 4) Compensating for the effects of heartbeats and respiration using non-rigid registration. Our earlier work addressed a number of steps in this protocol; this paper focuses on the registration step. We show that proper registration could lead to precise identification of the progress of the lung nodules.