| 摘要 |
System and method of delineating object within volumetric medical image, comprising: loading auto-generated contour, step 315, data forming object delineation structure; selecting, step 320, or using pre-selected (step 820, Fig. 8) subset of automatically generated contours for manual editing; presenting selected subset to user, step 325; receiving user feedback (eg. approval, modification or replacement with user generated contour) for at least one contour, step 335; deriving full contour set forming revised object delineation structure based on interpolation of edited outline, step 350; storing revised contour data, step 360. User may indicate object or image slice for editing, step 330. Auto-generated contours may be decimated to create subset. Subset may be chosen based on: user configured or pre-defined image slice spacing; out-of-plane curvature assessment; differences between neighbouring contours; prior definition in model or atlas; learning from previous editing; estimated contour accuracy assessment (eg. by comparison to shape model; comparing region statistics calculated at contour boundary with expected range for an organ; from variation of multiple contouring results; from variation in multi-atlas results before fusion; calculating confidence measures during contouring procedure). Automatic edge detection process may be atlas-based, segmentation-based or model-based. |
