Evaluation of Orthodontic Treatment Based on the Registration of X-ray Image Features

This paper proposes a method using feature-based registration, which quantitatively evaluates orthodontic treatment on X-ray lateral skull radiographs. Considering the problem of inconsistent scales of images before and after orthodontic, and to make the registration more robust to noise, this paper introduces a registration algorithm based on maximum correntropy criterion with scale constraints. By aligning the crania and mandibles separately and aligning teeth on this basis, the quantified movement of four key teeth can be acquired, which indicates the result of orthodontic treatment. Experimental results demonstrate that the proposed method makes evaluation of orthodontic treatment under the same reference and is not sensitive to the error of the mark points, which is accurate and robust.