Lung respiration motion modeling: a sparse motion field presentation method using biplane x-ray images
Dong Chen,Hongzhi Xie,Shuyang Zhang,Lixu Gu
DOI: https://doi.org/10.1088/1361-6560/aa8841
IF: 3.5
2017-09-21
Physics in Medicine and Biology
Abstract:Respiration-introduced tumor location uncertainty is a challenge in the precise lung biopsy for lung lesions. Current statistical modeling approaches hardly capture the complex local respiratory motion information. In this study, we formulate a statistical respiratory motion model using biplane x-ray images to improve the accuracy of motion field estimation by efficiently preserving local motion details for specific patients. Given CT data sets of 18 healthy subjects at end-expiratory and end-inspiratory breathing phases, the respiratory motion field is constructed based on deformation vector fields which are extracted from these CT data sets, and a lung contour motion repository respiratory is generated dependent on displacements of boundary control points. By varying the sparse weight coefficients of the statistical sparse motion field presentation (SMFP) method, the newly-input motion field is approximately presented by a sparse linear combination of a subset of the motion repository. The SMFP method is employed twice in the coefficient optimization process. Finally, these non-zero coefficients are fine-tuned to maximize the similarity between the projection image of reconstructed volumetric images and the current x-ray image. We performed the proposed method for estimating respiratory motion field on ten subject datasets and compared the result with the PCA method. The maximum average target registration error of the PCA-based and the SMFP-based respiratory motion field estimation are 3.1(2.0) and 2.9(1.6) mm, respectively. The maximum average symmetric surface distance of two methods are 2.5(1.6) and 2.4(1.3) mm, respectively.
engineering, biomedical,radiology, nuclear medicine & medical imaging