Improving Subtraction Radiography via Automated 2D Image Registration

Applications of Computer Vision in Medical Image Processing

Volume

Papers from the 1994 AAAI Spring Symposium

Contents

We have developed an initial system for performing 2D image registration for detecting bone density changes in dental radiographs based on aligning automatically extracted stable tooth features. The basic approach is to warp an x-ray image to align its landmark features with an earlier image of the same region to attempt to correct for imaging geometry variations. Although a 3D model is needed to perform truly accurate analysis, they are generally unavailable for these purposes. The goal of this project is to therefore identify the range of geometric variations and bone density changes that can be identified using 2D features and to evaluate the benefits of this approach over the manual change detection techniques currently being used.