Automatic Extraction of Control Points for Chest X-Ray Image and Elastic Registration

This paper presents a novel approach for control points detection and region of interest (ROI) extraction for chest X-ray images. We develop a search strategy to achieve the control points and ROI using speciality of intensity distribution of chest X-ray images, including mean, variance and difference. Therefore, a bounding box of lung fields is acquired and can be considered as ROI for image analysis and diagnosis. Based on the corresponding anatomical point landmarks, we also consider a thin-plate spline deformation as a method to solve non-rigid registration for chest X-ray images. Finally, we demonstrate it's efficiency for real-time ROI extraction and accurate registration.