CT and PET lung image registration and fusion in radiotherapy treatment planning using the chamfer-matching method

Abstract

Purpose: We present a validation study of CT and PET lung image registration and fusion based on the chamfer-matching method.Methods and Materials: The contours of the lung surfaces from CT and PET transmission images were automatically segmented by the thresholding technique. The chamfer-matching technique was then used to register the extracted lung surfaces. Arithmetic means of distance between the two data sets of the pleural surfaces were used as the cost function. Matching was then achieved by iteratively minimizing the cost function through three-dimensional (3D) translation and rotation with an optimization method.Results: Both anatomic thoracic phantom images and clinical patient images were used to evaluate the performance of our registration system. Quantitative analysis from five patients indicates that the registration error in translation was 2–3 mm in the transverse plane, 3–4 mm in the longitudinal direction, and about 1.5 degree in rotation. Typical computing time for chamfer matching is about 1 min. The total time required to register a set of CT and PET lung images, including contour extraction, was generally less than 30 min.Conclusion: We have implemented and validated the chamfer-matching method for CT and PET lung image registration and fusion. Our preliminary results show that the chamfer-matching method for CT and PET images in the lung area is feasible. The described registration system has been used to facilitate target definition and treatment planning in radiotherapy.