Automatic Orientation Detection of Hand Structures in Digital X-Ray Images

n the last decade portable flat panel detectors have gained increased usage in medical X-ray imaging. Since portable detectors can be used in different orientation, one cannot assume the anatomical structure of interest is upright. This means clinical staff has to manually rotate images. The aim of this paper is to improve the clinical workflow by using automatic orientation detection for hand structures and apply an automatic rotation. We include knowledge of the given anatomy by searching for intersection points of the arm with any image border. The algorithm extracts all border intersections. After analyzing those based on their histogram the intersection of the arm is found. Orientation can be calculated with a perpendicular line to the intersection line and the center of gravity. The results demonstrate that our algorithm is very fast and accurate for anteroposterior and posteroanterior projection images and can be employed to ensure an upright position.