Automatic matching of spine images to assess changes in scoliosis

This paper presents an image matching approach that can be used to measure changes in scoliotic curves. The proposed approach uses a novel fuzzy logic controller to estimate all open parameters. Using fluoroscopy images of a spine phantom, it was found that, with minimal user interaction, the matching of spine images could be achieved with high accuracy (the average errors were around 0.03 mm) and high computational efficiency (requiring less than 1 minute for matching each vertebra).