Towards Model-based 3-D Reconstruction of the Human Rib Cage from Radiographs

This work is concerned with the three-dimensional (3-d) shape and pose reconstruction of the human rib cage from few two-dimensional (2-d) radiographs. The reconstruction method is based on a statistical shape model of the rib cage that is adapted to the 2-d image data of a patient. An underlying optimization process minimizes a distance measure which quantifies the dissimilarities between 2-d projections of the 3-d shape model and the X-ray images and thereby estimates the shape model’s parameters. We propose a distance measure especially suited for the topology of the rib cage. A validation was performed on 29 sets of simulated, biplanar X-ray images.