Robust and Accurate Reconstruction of Patient-Specific 3D Surface Models from Sparse Point Sets: A Sequential Three-Stage Trimmed Optimization Approach

摘要: Constructing an accurate patient-specific three-dimensional (3D) bone model from sparse point sets is a challenging task. A priori information often required to handle this otherwise ill-posed problem. Previously we have proposed optimal approach for anatomical shape reconstruction [1], which uses dense surface distribution (DS-PDM) as the and formulates problem three-stage estimation process including (1) affine registration; (2) statistical extrapolation; (3) kernel-based deformation. In paper, propose important enhancement that enables realize stable reconstructions robustly reject outliers. Handling of outliers very crucial requirement especially in surgical scenario. This achieved by consistently employing Least Trimmed Squares (LTS) with roughly estimated outlier rate all three stages process. If value preferred, hypothesis testing procedure automatically determine it. Results new on dry cadaveric femurs different rates are shown.