OBJECTIVE: To capture 3D shape variation of the levator ani during straining with open access MR imaging. METHODS: Optimal scan-planning based on statistical shape modeling is developed for recovering the entire 3D structure of the levator ani from a limited number of 2D imaging planes. Statistical shape modeling ensures optimum material correspondence, and Subspace Reprojection is used to identify the optimum orientation of the imaging plans. The accuracy of the method in using limited 2D imaging planes to instantiate the dynamic structure of the levator ani is assessed with data acquired from 10 asymptomatic subjects. RESULTS: Leave-one-out analysis was performed whereby a model based on a training set consisting of all but one surface was used to instantiate the dynamic surface structure from the corresponding optimal planes. The mean surface distance error for the proposed Subspace Reprojection method is 3.989 +/- 0.790 mm, which is significantly smaller than other approaches. CONCLUSIONS: Surfaces of the levator ani may be instantiated using a limited number of imaging planes as well as a statistical shape model based on a training set of subjects. The proposed technique offers a new way forward for studying dynamic shape changes of 3D structures where complete volumetric imaging is prohibited by the inherent temporal resolution of the scanning technique.
pubs.doc.ic.ac.uk: built & maintained by Ashok Argent-Katwala.