Fast Segmentation of Abdominal Wall: Application to Sliding Effect Removal for Non-rigid Registration
The non-rigid registration of abdominal images is still a big challenge due to the breathing motion. Indeed, the sliding between the abdominal wall and the abdominal viscera makes the local deformation field discontinuous; it means that the classical registration approach, which assumes a smooth global deformation field cannot provide accurate and clinical-required results. Other new approaches intend to add in regularization a term to allow discontinuous deformation field near sliding boundary, however, the performance of such approaches needs to be further evaluated. We propose a new approach to perform abdominal image registration including a priori knowledge of the sliding area. Our strategy is to firstly delineate the abdominal wall in source and target images and create new images containing viscera only. Then a state-of-the-art non-rigid registration algorithm is adopted for the registration of the viscera region. In this paper, we firstly show why and how a quick interactive delineation of the full abdominal wall (AW) can be performed using B-spline interpolation. Secondly, we evaluate our registration approach on arterial and venous phase CT images. The results of our approach are compared to the one obtained using the same algorithm with the same parameters on the original data (without segmentation). The registration errors (mean ± SD) with our approach are: liver (1.94 ± 2.76 mm), left kidney (0.38 ± 0.66 mm), right kidney (0.42 ± 0.82 mm), spleen (4.15 ± 3.68 mm), which is much better than the registration result without segmentation: liver (6.48 ± 10.00 mm), left kidney (3.14 ± 3.39 mm), right kidney (2.79 ± 3.12 mm), spleen (17.45 ± 12.39 mm). The results clearly demonstrate our approach is a promising method to remove the sliding motion effect on the non-rigid registration of abdominal images.
KeywordsSliding motion abdominal image registration image segmentation
Unable to display preview. Download preview PDF.
- 3.Rohlfing, T., Maurer Jr., C.R., O’Dell, W.G., Zhong, J.: Modeling Liver Motion and Deformation during the Respiratory Cycle Using Intensity-Based Free-Form Registration of Gated MR Images. In: Medical Imaging: Visualization, Displaying, and Image-Guided Procedures. SPIE, vol. 4319, pp. 337–348 (2001)Google Scholar
- 5.Pace, D.F., Enquobahrie, A., Yang, H., Aylward, S.R., Niethammer, M.: Deformable Image Registration of Sliding Organs Using Anisotropic Diffusive Regularization. In: 8th IEEE International Symposium on Biomedical Imaging, pp. 407–413. IEEE Press, New York (2011)Google Scholar
- 7.Freiman, M., Voss, S., Warfield, S.: Demons Registration with Local Affine Adaptive Regularization: Application to Registration of Abdominal Structures. In: 8th IEEE International Symposium on Biomedical Imaging, pp. 1219–1222. IEEE Press, New York (2011)Google Scholar
- 14.Collignon, A., Maes, F., Delaere, D., Vandermeulen, D., Seutens, P., Mar, G.: Automated Multimodality Image Registration Using Information Theory. In: 14th International Conference on Information Processing in Medical Imaging, IPMI 1995, pp. 263–274 (1995)Google Scholar
- 16.Insight Segmentation and Registration Toolkit (ITK), http://www.itk.org