XeMRI to CT Lung Image Registration Enhanced with Personalized 4DCT-Derived Motion Model
This paper presents a novel method for multi-modal lung image registration constrained by a motion model derived from lung 4DCT. The motion model is estimated based on the results of intra-patient image registration using Principal Component Analysis. The approach with a prior motion model is particularly important for regions where there is not enough information to reliably drive the registration process, as in the case of hyperpolarized Xenon MRI and proton density MRI to CT registration. Simultaneously, the method addresses local variations between images in the supervoxel-based motion model parameters optimization step. We compare our results in terms of the plausibility of the estimated deformations and correlation coefficient with 4DCT-based estimated ventilation maps using state-of-the-art multi-modal image registration methods. Our method achieves higher average correlation scores, showing that the application of Principal Component Analysis-based motion model in the deformable registration, helps to drive the registration for the regions of the lungs with insufficient amount of information.
KeywordsLung 4D CT XeMRI Multi-modal image registration Lung motion model Ventilation estimation
AS and BWP would like to acknowledge funding from the CRUK and EPSRC Cancer Imaging Centre in Oxford. BWP acknowledges Oxford NIHR Biomedical Research Centre (Rutherford Fund Fellowship at HDR UK). JAS was supported by EP/P023509/1 and Wellcome Trust/EPSRC Centre for Medical Engineering.
- 13.Liu, X., Oguz, I., Pizer, S.M., Mageras, G.S.: Shape-correlated deformation statistics for respiratory motion prediction in 4D lung (2010)Google Scholar
- 24.Szmul, A., Papież, B.W., Matin, T., Gleeson, F., Schnabel, J.A., Grau, V.: Regional lung ventilation estimation based on supervoxel tracking. In: Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling, vol. 10576, pp. 10576-1–10576-7 (2018)Google Scholar